Differences in the binding capacity of human apolipoprotein
E3 and E4 to size-fractionated lipid emulsions
Matthew A. Perugini
1
, Peter Schuck
2
and Geoffrey J. Howlett
1
1
Department of Biochemistry and Molecular Biology, The University of Melbourne, Parkville, VIC, Australia;
2
Division of
Bioengineering and Physical Science, ORS, OD, National Institutes of Health, Bethesda, MD, USA
We describe sensitive new approaches for detecting and
quantitating protein–lipid interactions using analytical ultra-
centrifugation and continuous size-distribution analysis
[Schuck (2000) Biophys. J. 78, 1606–1619]. The new methods
were developed to investigate the binding of human apo-
lipoprotein E (apoE) isoforms to size-fractionated lipid
emulsions, and demonstrate that apoE3 binds preferentially
to small lipid emulsions, whereas apoE4 exhibits a preference
for large lipid particles. Although the apparent binding
affinity for large emulsions is similar (K
d
% 0.5 l
M
), the
maximum binding capacity for apoE4 is significantly higher
than for apoE3 (3.0 and 1.8 amino acids per phospholipid,
respectively). This indicates that apoE4 has a smaller binding
footprint at saturation. We propose that apoE isoforms

(R112/R158), differ by single amino acid substitutions at
positions 112 and 158 [13–15]. ApoE2 is linked to type III
hyperlipoproteinemia and has low affinity for the low
density lipoprotein (LDL) receptor [13,16]. ApoE3 is the
most common isoform and is associated with normolipide-
mia [13,17], while the apoE4 isoform is independently
associated with an increased risk for atherosclerosis [17,18]
and late-onset Alzheimer’s disease [19,20]. Recent studies
compare the structure–function relationships of the apoE
isoforms, including their stability [21], self-association in the
presence and absence of phospholipid [4], and their ability to
bind preferentially to different lipoprotein classes [7–10].
ApoE is composed of two independently folded domains.
The 10 kDa COOH-terminal domain (residues 225–299)
possesses high lipid affinity, while the 22 kDa NH
2
-terminal
domain (residues 1–191) binds weakly to lipid and mediates
receptor interactions [13]. In the absence of lipid, the NH
2
-
terminal domain of apoE3 forms an elongated four helical-
bundle, stabilized by hydrophobic contacts and intra- and
inter-helical salt bridges [22]. The substitution of cysteine for
arginine at position 112 in apoE4, results in an additional
salt-bridge between Glu109 and Arg112 and the displace-
ment of Arg61 from the surface of the four-helix bundle [8].
The displaced Arg61 side chain forms an interdomain salt-
bridge with Glu255, an interaction that is critical for
directing the preference of apoE4 to bind large VLDL

receptor-active conformation in the presence of phospho-
lipid [13,27]. Evidence in support of this model is provided
by recent lipid binding studies employing synthetic lipid
emulsions and intact apoE4, its 22 kDa NH
2
-terminal
fragment, and the 10 kDa COOH-terminal fragment [28].
At a low surface concentration of protein, the binding
enthalpy of intact apoE4 was consistent with the sum of the
enthalpies for the 22 kDa and 10 kDa derivatives, indica-
ting that both NH
2
- and COOH-terminal domains bind
to the emulsion surface [28]. At saturation, however, the
enthalpy for intact apoE4 was similar to that of the 10 kDa
fragment, suggesting that only the COOH-terminal domain
of intact apoE4 interacts with the emulsion surface [28]. It is
not known whether the apoE3 isoform displays a similar
phenomenon on the surface of lipid particles.
In the present study we develop methods for the
characterization of lipid emulsions of well-defined compo-
sition but different particle size. We compare the binding of
apoE3 and apoE4 to small and large lipid particles, and
show that apoE4 has a greater capacity than apoE3 to bind
large lipid emulsions. These results suggest that at satura-
tion, apoE4 binds primarily in a more compact conforma-
tion, whereas apoE3 adopts an expanded conformation on
the lipoprotein surface.
EXPERIMENTAL PROCEDURES
Materials

)1
) and the sample layered beneath 0.1
M
sodium
phosphate buffer, pH 7.4. The sample was centrifuged at
110 000 g for 35 min in a Beckman 70.1 Ti rotor and model
L8-70 ultracentrifuge. After centrifugation, the top 1.5 mL
was collected (unfractionated) and the density adjusted to
1.018 gÆmL
)1
with solid sucrose (5 gÆdL
)1
), before adding a
0–5% (w/v) linear sucrose gradient prepared in 0.1
M
sodium phosphate, pH 7.4. The sample was then centri-
fuged at 4500 g for2hinaBeckmanSW-40rotorand
model L8-70 ultracentrifuge. Fractions (1.0 mL) were
collected from the bottom of the tube using a peristaltic
pump and dialyzed exhaustively against 50 m
M
sodium
phosphate, pH 7.4. Phospholipid and triacylglycerol con-
centrations were determined using enzymatic spectropho-
tometric phospholipid and glycerol assay kits (Roche).
Peptide synthesis and purification
ApoE(263–286) peptide (SWFEPLVEDMQRQWAGLV
EKVQAA, M
r
¼ 2818) [30–32], was synthesized by auto-

microsampler (Protein Solutions, Charlottesville, VA,
USA). Samples, suspended in 50 m
M
sodium phosphate,
pH 7.4, were centrifuged for 5 min in a microcentrifuge to
remove dust particles, and 20 lL sample was inserted in the
cuvette with the temperature control set to 20 °C. The light
scattering signal was collected at 90°, and diffusion coeffi-
cients and Stokes-radii of the emulsion fractions were
calculated with the instrument software.
Flotation velocity
Flotation velocity experiments were performed using a
Beckman model XL-A analytical ultracentrifuge. Prior to
centrifugation, apoE and emulsion samples were exhaust-
ively dialyzed (> 20 h) against 50 m
M
sodium phosphate,
pH 7.4. Samples (300–400 lL) and reference (320–420 lL)
solutions were loaded into a conventional double-sector
quartz cell and mounted in a Beckman An-60 Ti rotor.
Experiments were conducted at 20 °C and at a rotor speed
of 5000 r.p.m. Data was collected in continuous mode, at a
single wavelength (230 nm or 250 nm), time interval of 360 s
and a step-size of 0.003 cm without averaging. Multiple
scans at different time points were fitted to a single species or
5940 M. A. Perugini et al. (Eur. J. Biochem. 269) Ó FEBS 2002
to a continuous size distribution (see below) using the
program
SEDFIT
(which is available at www.analyticalultra

; D; r; tÞds
f
ð1Þ
where a(r,t) denotes the observed optical density at
radius r and time t, c(s
f
) denotes the differential flotation
coefficient distribution, L(s
f
,D,r,t) denotes the solution
to the Lamm equation [34], calculated with an adapta-
tion of the moving frame of reference method [35] to
flotation velocity, taking into consideration the rotor
acceleration phase. One feature of boundary modeling
with Eqn (1) is that it allows interconversion of the
flotation coefficient distribution to a molar mass distri-
bution via the Stokes-Einstein and Svedberg equation
[36], upon consideration of the spherical shape and the
size-dependent particle density of the polydisperse sol-
utes. For the functional dependence between the density
and molar mass of the fractionated emulsion particles,
we assumed a spherical monolayer of Myr
2
Gro-PCho
surrounding a core of TO, supported by transmission
electron microscopy (data not shown). Consequently,
the relationship between density of the particles as a
function of particle mass was determined using values
for M and


Gro-
PCho/TO and EggPtdCho/TO emulsion fractions were
calculated using Eqns (2) and (3), respectively.
For the mixtures of emulsion particles and protein,
because of the unknown contribution of the protein to the

vv,
a flotation coefficient distribution was calculated by
approximating the diffusion with an average diffusion
coefficient measured by dynamic light scattering. Because of
the size of the emulsion particles, diffusional broadening of
the flotation profiles is not very large, and variation of the
diffusion coefficient throughout the distribution can be
considered a second order effect. However, this method
does not allow the transformation of the flotation coefficient
distribution in a molar mass distribution.
To prevent an ill-conditioned analysis when performing
continuous size-distribution analysis with many species, a
regularization technique was employed that selects the most
parsimonious distribution of species that fits the data within
a predetermined confidence limit. Consistent with observa-
tions in previous studies [4,36], this resulted in smooth
distributions. However, in contrast to earlier studies with
sedimentation coefficient distributions of proteins, for which
maximum entropy regularization seemed advantageous
because of its potential to produce sharp peaks for discrete
mixtures [4,36,39], we found the Tikhonov-Phillips regular-
ization with second derivative functional more useful,
because it avoids possible oscillatory artifacts known to be
encountered with the maximum entropy method for broad

range of 0.1 cm in the plateau region using data from the
final radial scan. The average optical density due to free
protein was converted into concentration units via a five-
point standard curve. The concentration of bound protein
was calculated based on the measured free and the known
total amount of protein. The apparent dissociation
constant (K
d
) and maximum binding capacity (B
max
)were
estimated on the basis of a Langmuir isotherm, by
plotting the amount of free protein (P
f
), against total
phospholipid concentration (PC) multiplied by the ratio of
free to bound protein (P
f
/P
b
) [2,6,41]:
P
f
¼ PCðP
f
=P
b
ÞB
max
À K

scattering data, the fractionation procedure yielded individ-
ual fractions of particles with radii in the range of 38–52 nm.
Flotation velocity analysis of fractionated
lipid emulsions
The solution properties of the major emulsion fractions
were further characterized by analytical ultracentrifugation.
Figure 2 shows flotation velocity data of fractions 2, 4 and 6
at 360 s intervals. A significant time-dependent broadening
of the flotation boundary is observed in each, suggesting the
emulsion fractions are moderately heterogeneous. This
assertion is supported by the poor fits (rmsd ¼ 0.0574)
and nonrandom distribution of residuals, when for example,
the data for fraction 4 is fitted assuming a single species with
the average diffusion coefficient as determined independ-
ently by dynamic light scattering (data not shown). Similar
observations were made with fractions 2, 3, 5, and 6. We
therefore sought a method to characterize the residual
polydispersity.
In a previous study [29], dc/dt analysis [42] was employed
to determine the apparent flotation distribution function
g(s*) of fractionated lipid emulsions. In particular for large
particles and when using absorbance optical ultracentrifuge
data, this method is intrinsically limited due to artificial
broadening that is introduced by the finite time-difference
between the scans considered for dc/dt analysis [43]. There-
fore, in the present study we took advantage of the con-
tinuous size distribution method c(s)andc(M) for direct
boundary modeling [36]. Although the direct boundary
Fig. 1. Dynamic light scattering of size-fractionated lipid emulsions. The
TO/Myr

best fits for fractions 2, 4 and 6 are shown in Fig. 2 (solid
lines), which result in a random distribution of residuals
(Fig. 2, insets) and low rmsd values (< 0.005) when
compared to fits assuming a single species. The resulting
c(s
f
) size-distributions for the major emulsion fractions,
including fractions 2, 4 and 6, are shown in Fig. 3. The data
indicate that the size-fractionated emulsions have well-
separated size-distributions, albeit with some degree of
overlap. The continuous size distributions are broader with
increasing fraction size, suggesting that the larger fractions,
F5 and F6, are more polydisperse than the smaller fractions,
F2 and F3. This may be an artifact of the fractionation
process, as the emulsions are harvested smallest to largest
following sucrose gradient ultracentrifugation. The c(s
f
)
distribution for unfractionated emulsions is also shown,
demonstrating a high degree of heterogeneity, as expected,
and a double maxima in c(s
f
) at approximately 200 S and
550 S (Fig. 3). Similar c(s
f
) distributions were obtained
when the emulsions were prepared and fractionated at a
10-fold higher total lipid concentration of 10 mgÆmL
)1
(data

Interaction of ApoE(263–286) with lipid emulsions
We initially examined the binding of a synthetic peptide
comprising residues 263–286 of human apoE to the
fractionated lipid emulsions. ApoE(263–286) is amphipathic
in nature, and has previously been reported to bind to
Myr
2
Gro-PCho bilayers [31] and SDS micelles [32], a
common lipid-mimetic. Figure 4 shows the continuous
flotation, c(s
f
), distribution of emulsion fraction 4 in the
absence and presence of apoE(263–286), calculated with a
fixed diffusion coefficient of 0.46 · 10
)7
cm
2
Æs
)1
.Relativeto
the control, the flotation rate of fraction 4 after the addition
of 1.0 l
M
peptide is significantly reduced, accompanied also
by an increase in the area under the distribution curve
(Fig. 4A). These changes are attributed to peptide binding
to the emulsion particles. Monte-Carlo analysis demon-
strates that the observed increase in area under the curve
and shifts to lower values of s
f

vv (as above).
Fraction #
TO:Myr
2
Gro-PCho
molar ratio
s
f
(S)
M
(· 10
8
Da)

vv
(mLÆg
)1
)
R
S
(nm)
2 2.05 166 0.94 1.061 34
3 2.62 333 2.3 1.067 46
4 3.18 506 4.00 1.072 55
5 3.86 664 5.60 1.076 62
6 4.29 795 7.00 1.079 67
Ó FEBS 2002 Binding of apolipoprotein E3 and E4 to emulsions (Eur. J. Biochem. 269) 5943
statistically significant, and cannot be attributed to noise
affecting the data analysis (Fig. 4B). At a 10-fold higher
peptide concentration of 10.0 l

, against the phospholipid concentration
multiplied by the ratio of free to bound peptide, a linear plot
results (Fig. 5, inset), yielding an apparent dissociation
constant, K
d
,of75l
M
and binding capacity, B
max
,of
approximately four amino acids per phospholipid (Table 2).
Interaction of ApoE3 and ApoE4 with Myr
2
Gro-PCho/TO
lipid emulsions
To compare the interactions of apoE3 and apoE4 isoforms
with size-fractionated lipid emulsions, flotation velocity
experiments were conducted in the analytical ultracentri-
fuge. Fraction 2 was employed as a synthetic model for
small lipoprotein particles and fraction 6 for large lipopro-
tein particles.
The c(s
f
) distributions of fraction 2 in the absence and
presence of 1.0 l
M
apoE3 or apoE4 are compared in
Fig. 6A. Relative to the control, the c(s
f
) distribution of

M
apoE(263–286) (solid line, solid sym-
bols). (B) Results of Monte-Carlo statistical analysis distributions,
calculated from 1000 synthetic data sets to a confidence level of P ¼
0.95. The lower (0.025) and upper (0.975) quantiles are depicted as
dashed lines, enclosing the mean distribution (solid lines) for fraction 4
alone (labelled 1), and fraction 4 + 1.0 l
M
apoE(263–286) peptide
(labelled 2).
Fig. 5. Binding of apoE(263–286) peptide to Myr
2
Gro-PCho/TO
emulsion fraction 6. The amount of bound apoE(263–286) (symbols +
solid line) is plotted as a function of free protein (lg/mL). The con-
centrations of Myr
2
Gro-PCho and TO in lipid emulsion fraction
6are110l
M
and 390 l
M
, respectively. Binding data was obtained by
analytical ultracentrifugation using the direct binding assay as des-
cribed in Experimental procedures. Inset: Linearized plot of the
binding data for apoE(263–286) (symbols) shown in panel A. The solid
line represents the linear least-squares fits to the data according to Eqn
(4), where the y-intercept and slope equate to the apparent K
d
and

shown).
Interaction of ApoE3 and ApoE4 with EggPtdCho/TO
lipid emulsions
The binding of apoE3 and apoE4 to lipid particles was
also assessed using size-fractionated emulsions comprised
of egg yolk phosphatidylcholine (EggPtdCho) and TO.
EggPtdCho is comprised of saturated and unsaturated
phospholipids, which also differ in fatty acyl chain length,
and are therefore more biologically relevant than mono-
layers comprised of Myr
2
Gro-PCho alone [46]. The
EggPtdCho/TO emulsions were synthesized by pressure
extrusion and fractionated by sucrose-gradient ultracentri-
fugation, employing identical procedures to those used for
the synthesis of Myr
2
Gro-PCho/TO emulsions. Following
fractionation, flotation velocity experiments were performed
to characterize the solution properties of these emulsions,
which are summarized in Table 3. The data presented in
Table 3 demonstrates that the EggPtdCho/TO emulsions
share similar physical properties to the fractionated
Myr
2
Gro-PCho/TO emulsions (Table 1). Accordingly, flo-
tation velocity experiments were employed to compare the
binding of apoE3 and apoE4 to small (fraction 2) and large
(fraction 6) EggPtdCho/TO emulsions.
As for Myr

maximum and area under the curve is obvious in the
presence of the apoE4 isoform (Fig. 7C, open symbols),
indicating that a greater proportion of apoE4 binds the
larger particles. Together with the results obtained in the
previous section employing Myr
2
Gro-PCho/TO emulsions,
these data support the conclusion that apoE3 and apoE4
bind preferentially to small and large lipid particles,
respectively.
Direct binding analysis of ApoE3 and ApoE4 to large
emulsion particles
To verify and quantify the size-dependent interaction of
apoE3 and apoE4 to fractionated lipid emulsions, we also
employed a direct binding using Myr
2
Gro-PCho/TO
fraction 6 and physiological concentrations of apoE. The
binding isotherms for the apoE3 and apoE4 isoforms are
presented in Fig. 8A. Both curves show evidence of
saturation, although the proportion of bound apoE4 is
significantly greater at all protein concentrations
employed, particularly in excess of 30 lgÆmL
)1
or 1.0 l
M
(Fig. 8A). The apparent K
d
and B
max

d
B
max
(lgÆmL
)1
)(l
M
) ApoE/particle PL/ApoE Amino acids/PL
ApoE(263–286) 210 75 3.14 · 10
4
5.9 4.1
ApoE3 15 0.44 1010 163 1.83
ApoE4 17 0.51 1630 101 2.96
Ó FEBS 2002 Binding of apolipoprotein E3 and E4 to emulsions (Eur. J. Biochem. 269) 5945
Fig. 6. Flotation velocity analysis of Myr
2
Gro-PCho/TO and
EggPtdCho/TO emulsion fraction 2 in the absence and presence of
apoE3 and apoE4 isoforms. The c(s
f
) distribution calculated using an
invariant D ¼ 0.63 · 10
)7
cm
2
Æs
)1
is plotted as a function of flotation
coefficient. (A) Myr
2

apoE3 (labelled 2), and Myr
2
Gro-PCho/TO fraction 2 + 1.0 l
M
apoE4 (labelled 3). (C) EggPtdCho/TO fraction 2 alone (solid line, no
symbols); EggPtdCho/TO fraction 2 + 1.0 l
M
apoE3 (solid symbols
+ line) and EggPtdCho/TO fraction 2 + 1.0 l
M
apoE4 (open sym-
bols + line). The total lipid concentration in EggPtdCho/TO fraction
6 ¼ 280 l
M
, i.e. [EggPtdCho] ¼ 135 l
M
+[TO]¼ 145 l
M
.
Fig. 7. c(s
f
) distribution analysis of Myr
2
Gro-PCho/TO and
EggPtdCho/TO fraction 6 in the presence and absence of apoE3 and
apoE4. The c(s
f
) distributions calculated using an invariant D ¼
0.38 · 10
)7

Gro-PCho] ¼ 23 l
M
+[TO]¼
102 l
M
.(B)Myr
2
Gro-PCho/TO fraction 6 alone (solid line, no sym-
bols); Myr
2
Gro-PCho/TO fraction 6 + 0.5 l
M
apoE4 (dashed line),
Myr
2
Gro-PCho/TO fraction 6 + 1.0 l
M
apoE4 (dashed-dotted line)
and Myr
2
Gro-PCho/TO fraction 6 + 2.0 l
M
apoE4 (solid line +
open symbols). (C) EggPtdCho/TO fraction 6 alone (solid line, no
symbols); EggPtdCho/TO fraction 6 + 1.0 l
M
apoE3 (solid symbols
+ line) and EggPtdCho/TO fraction 6 + 1.0 l
M
apoE4 (open sym-

notypes [18] may be due to the inability of apoE4 to bind
and initiate the clearance of small lipoproteins in plasma. In
contrast, the observation that chylomicron remnants are
cleared faster in subjects with apoE4, compared to those
with apoE3 [53], may be explained by the superior ability of
apoE4 to bind large lipid particles. Similarly, a size-
dependent binding phenomenon may account for the
observation that apoE3 distributes preferentially with small
(density > 1.125 gÆmL
)1
) and apoE4 with large (density
<1.00 gÆmL
)1
) lipoproteins in the cerebrospinal fluid of the
brain [54].
Insight into the structural basis for apoE3 and apoE4
lipid binding preferences is provided by earlier studies,
where it is demonstrated that apoE3 and apoE4 differ in
their NH
2
- and COOH-terminal domain interactions [7,8].
In addition, it is known that truncation of apoE4 at residue
244 abolishes VLDL binding [7], although the same
truncated variant of apoE3 retains the ability to bind
HDL [23]. This suggests the important determinants for
binding large lipid particles are downstream of residue 244,
in the COOH-terminal region of the protein. More recently,
studies employing fluorescence resonance energy transfer
[24,26], intradomain disulfide bonding [3], nuclear magnetic
resonance [25], and microcalorimetry [28] experiments

vv
(mLÆg
)1
)
R
S
(nm)
2 1.07 133 0.76 1.038 31
3 1.66 313 2.31 1.049 46
4 2.18 487 4.05 1.056 55
5 2.84 649 5.80 1.061 62
6 3.54 806 7.60 1.067 68
Fig. 8. Binding of apoE3 and apoE4 to Myr
2
Gro-PCho/TO emulsion
fraction 6. (A) The amount of bound apoE3 (solid symbols + solid
line) and apoE4 (open symbols + solid line) is plotted as a function of
free protein (lgÆmL
)1
). The concentration of Myr
2
Gro-PCho and TO
in lipid emulsion fraction 6 ¼ 150 l
M
and 260 l
M
, respectively.
Binding data was obtained by analytical ultracentrifugation using
the direct binding assay as described in Experimental procedures.
(B) Linearized plot of the binding data for apoE3 (solid symbols)

phospholipid. This indicates that apoE4 has a smaller
binding footprint when bound to large lipid particles. One
explanation for this phenomenon could be that apoE4 self-
associates more readily on the surface of lipid emulsions.
However, we have previously demonstrated that both
apoE3 and apoE4 dissociate to folded monomers when
complexed with phospholipid micelles [4], consistent also
with earlier studies [55,56]. Accordingly, we propose that
apoE3 and apoE4 differ in their capacity to adopt
expanded and compact conformations when bound to
lipid particles. We suggest that the apoE4 isoform binds
more extensively to large, less curved lipid particles due its
greater propensity to adopt a more surface-compact,
closed conformation (Fig. 9A). This difference may be
mediated by interactions between Arg61 and Glu255 [7]. In
contrast, the preferential binding of apoE3 to small lipid
particles may be explained by its enhanced ability to adopt
a more expanded, open conformation (Fig. 9B), which is
likely to provide greater flexibility for interacting with
highly curved lipid surfaces. Differences in the propensity
of apoE3 and apoE4 to adopt open and closed confor-
mations (Fig. 9) may provide insight into the role of
apoE4 in disease.
ACKNOWLEDGEMENTS
We would like to thank Dr Karl Weisgraber (Gladstone Institute of
Cardiovascular Disease, San Francisco, CA, USA) for kindly supplying
the apoE3 and apoE4 pGEX-3X plasmids. We also thank Con
Dogovski, Cait MacPhee and Ben Atcliffe for advice and assistance
during the course of this work. This work was funded by the National
Health and Medical Research Council, Australia.

tional characterization of apolipoprotein E isoforms over-
expressed in Escherichia coli. Prot. Expr. Purif. 16, 224–230.
10. Weisgraber, K.H. (1990) Apolipoprotein E distribution among
human plasma lipoproteins: role of the cysteine-arginine inter-
change at residue 112. J. Lipid Res. 31, 1503–1511.
11. Eisenberg, S. (1984) High density lipoprotein metabolism. J. Lipid
Res. 25, 1017–1058.
12. Mahley, R.W. (1988) Apolipoprotein E: cholesterol transport
protein with expanding role in cell biology. Science 240, 622–630.
13. Weisgraber, K.H. (1994) Apolipoprotein E: structure-function
relationships. Adv. Prot. Chem. 45, 249–302.
14. Rall, S.C. Jr, Weisgraber, K.H. & Mahley, R.W. (1982) Human
apolipoprotein E. The complete amino acid sequence. J. Biol.
Chem. 257, 4171–4178.
15. Weisgraber, K.H., Rall, S.C. Jr & Mahley, R.W. (1981) Human E
apoprotein heterogeneity. Cysteine-arginine interchanges in the
amino acid sequence of the apo-E isoforms. J. Biol. Chem. 256,
9077–9083.
16. Utermann, G., Jaeschke, M. & Menzel, J. (1975) Familial
hyperlipoproteinemia type III: deficiency of a specific apolipo-
protein (apo E-III) in the very-low-density lipoproteins. FEBS
Lett. 56, 352–355.
17. Mahley,R.W.&Rall,S.C.Jr(2000)ApolipoproteinE:farmore
than a lipid transport protein. Ann. Rev. Genomics Hum. Genet. 1,
507–537.
18. Davignon, J., Gregg, R.E. & Sing, C.F. (1988) Apolipoprotein E
polymorphism and atherosclerosis. Arteriosclerosis 8, 1–21.
19. Corder, E.H., Saunders, A.M., Strittmatter, W.J., Schmechel,
D.E., Gaskell, P.C., Small, G.W., Roses, A.D., Haines, J.L. &
Pericak-Vance, M.A. (1993) Gene dose of apolipoprotein E type 4

binding domain of human apolipoprotein E. J. Biol. Chem. 275,
33601–33606.
25. Lund-Katz, S., Zaiou, M., Wehrli, S., Dhanasekaran, P., Baldwin,
F., Weisgraber, K.H. & Phillips, M.C. (2000) Effects of lipid
interaction on the lysine microenvironments in apolipoprotein E.
J. Biol. Chem. 275, 34459–34464.
26. Narayanaswami, V., Szeto, S.S.W. & Ryan, R.O. (2001) Lipid
association-induced N- and C-terminal domain reorganization in
human apolipoprotein E3. J. Biol. Chem. 276, 37853–37860.
27. Narayanaswami, V. & Ryan, R.O. (2000) Molecular basis of
exchangeable apolipoprotein function. Biochim. Biophys. Acta
1483, 15–36.
28. Saito, H., Dhanasekaran, P., Baldwin, F., Weisgraber, K.H.,
Lund-Katz, S. & Phillips, M.C. (2001) Lipid binding-induced
conformational change in human apolipoprotein E. Evidence for
two lipid-bound states on spherical particles. J. Biol. Chem. 276,
40949–40954.
29. MacPhee, C.E., Chan, R.Y.S., Sawyer, W.H., Stafford, W.F. &
Howlett, G.J. (1997) Interaction of lipoprotein lipase with
homogeneous lipid emulsions. J. Lipid Res. 38, 1649–1659.
30. MacPhee, C.E., Perugini, M.A., Sawyer, W.H. & Howlett, G.J.
(1997) Trifluoroethanol induces the self-association of specific
amphipathic peptides. FEBS Lett. 416, 265–268.
31. Sparrow, J.T., Sparrow, D.A., Fernando, G., Culwell, A.R.,
Kovar, M. & Gotto, A.M. Jr (1992) Apolipoprotein E: phos-
pholipid binding studies with synthetic peptides from the carboxyl
terminus. Biochemistry 31, 1065–1068.
32. Wang, G., Pierens, G.K., Treleaven, W.D., Sparrow, J.T. &
Cushley, R.J. (1996) Conformations of human apolipoprotein E
(263–286) and E (267–289) in aqueous solutions of sodium

´
zdy, F.J. &
Kaiser, E.T. (1980) The mechanism of activation of lecithin:
cholesterol acyltransferase by apolipoprotein A-I and an amphi-
philic peptide. J. Biol. Chem. 255, 7333–7339.
42. Stafford, W.F. (1992) Boundary analysis in sedimentation trans-
port experiments: a procedure for obtaining sedimentation coef-
ficient distributions using the time derivative of the concentration
profile. Anal. Biochem. 203, 295–301.
43. Schuck, P. & Rossmanith, P. (2000) Determination of the sedi-
mentation coefficient distribution by least–squares boundary
modeling. Biopolymers 54, 328–341.
44. van Holde, K.E. & Weischet, W.O. (1978) Boundary analysis of
sedimentation velocity experiments with monodisperse and pau-
cidisperse solutes. Biopolymers 17, 1387–1403.
45. Schuck, P., Perugini, M.A., Gonzales, N.R., Howlett, G.J. &
Schubert, D. (2002) Size-distribution analysis of proteins by ana-
lytical ultracentrifugation: strategies and application to model
systems. Biophys. J. 82, 1096–1111.
46. New, R.R.C. (1990) Introduction. In Liposomes a Practical
Approach (New, R.R.C., ed.), pp. 1–32. Oxford University Press,
New York.
47. Deckelbaum, R.J., Shipley, G. & Small, D.M. (1977) Structure
and interactions of lipids in human plasma low density lipopro-
teins. J. Biol. Chem. 252, 744–754.
48. Gotto, A.M. Jr, Pownall, H.J. & Havel, R.J. (1986) Introduction
to the plasma lipoproteins. Meth Enzymol. 128, 3–41.
49. Kroon, P.A. (1994) Fluorescence study of the motional states of
core and surface lipids in native and reconstituted low density
lipoproteins. Biochemistry 33, 4879–4884.