Báo cáo khoa học: Apolipoprotein E-derived antimicrobial peptide analogues with altered membrane affinity and increased potency and breadth of activity pdf

Apolipoprotein E-derived antimicrobial peptide analogues
with altered membrane affinity and increased potency and
breadth of activity
Bridie A. Kelly
1
, Stuart J. Neil
2,
*, A
´
ine McKnight
2,
†, Joana M. Santos
3,
‡, Photini Sinnis
3
,
Edward R. Jack
1,
§, David A. Middleton
1,
– and Curtis B. Dobson
1
1 Faculty of Life Sciences, The Mill, The University of Manchester, UK
2 Wohl Virion Centre, Windeyer Building, University College London, UK
3 Department of Medical and Molecular Parasitology, New York University School of Medicine, NY, USA
Keywords
antimicrobial peptides; apolipoprotein E;
HIV; Plasmodium; membrane perturbation
Correspondence
C. B. Dobson, Faculty of Life Sciences,
The Mill, The University of Manchester,

PO Box 88, Sackville Street, Manchester
M60 1QD, UK
Fax: +44 (0)161 306 4433
Tel: +44 (0)161 306 8765
E-mail:
Present address
*Aaron Diamond AIDS Research Center,
New York, NY, USA
†Centre for Infectious Disease, Institute of
Cell and Molecular Science, Barts and The
London, Queen Mary’s School of Medicine
and Dentistry, London, UK
‡Department of Microbiology and Molecular
Medicine, CMU, University of Geneva,
Switzerland
§Division of Structural Biology, Department
of Biological sciences, University of
Warwick, Coventry, UK
–School of Biological Sciences, University of
Liverpool, UK
(Received 4 May 2007, revised 22 June
2007, accepted 6 July 2007)
doi:10.1111/j.1742-4658.2007.05981.x
Host-derived anti-infective proteins represent an important source of
sequences for designing antimicrobial peptides (AMPs). However such
sequences are often long and comprise diverse amino acids with uncertain
contribution to biological effects. Previously, we identified a simple highly
cationic peptide derivative of human apolipoprotein E (apoEdp) that inhib-
ited a range of microorganisms. Here, we have dissected the protein chem-
istry underlying this activity. We report that basic residues and peptide

length of around 18 residues were required for activity; however, the Leu
residues can be substituted by several other residues without loss of activity
and, when substituted with Phe or Trp, resulted in peptides with increased
potency. These apoEdp-derived AMPs (apoE-AMPs) showed no cytotoxi-
city and minimal haemolytic activity, and were active against HIV and
Plasmodium via an extracellular target. CXCR4 and CCR5 strains of HIV
were inhibited though an early stage in viral infection upstream of fusion,
and a lack of inhibition of vesicular stomatitis virus G protein pseudotyped
HIV-1 suggested the anti-HIV activity was relatively selective. Inhibition of
Plasmodium invasion of hepatocytes was observed without a direct action
on Plasmodium integrity or attachment to cells. The Trp-substituted apoE-
AMP adhered to mammalian cells irreversibly, explaining its increased
potency; NMR experiments confirmed that the aromatic peptides also
showed stronger perturbation of membrane lipids (relative to apoEdp).
Our data highlight the contribution of specific amino acids to the activity
of apoEdp (and also potentially unrelated AMPs) and suggest that apoE-
AMPs may be useful as lead agents for preventing the early stages of HIV
and Plasmodium cellular entry.
Abbreviations
AMP, antimicrobial peptide; apoE, apolipoprotein E; apoE-AMP, apoE-derived AMP; apoEdp, cationic peptide derivative of human
apolipoprotein E; DMPC-d
4
, 1,2-dimyristoylphosphatidyl-1,1,2,2-
2
H
4
-choline; DOPG, dioleoylphosphatidylglycerol; FFU, focus forming unit;
HSPG, heparan sulfate proteoglycan; HSV1, herpes simplex virus type 1; HSV2, herpes simplex virus type 2; LDLR, low density lipoprotein
receptor; LPS, lipopolysaccharide; MIC, minimum inhibitory concentration; MTT, 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl-tetrazolium bromide;
VSV-G, vesicular stomatitis virus G protein.

FEBS Journal 274 (2007) 4511–4525 ª 2007 The Authors Journal compilation ª 2007 FEBS 4511
Antimicrobial peptides (AMPs) represent a promising
source of novel molecules with potential as therapeutic
leads against bacteria, viruses and parasites. A large
number of AMPs have been reported previously with
many being derived from nonhuman naturally occur-
ring sequences, especially from invertebrates or amphi-
bians [1]. In addition, AMPs have been developed from
entirely synthetic sequences, often comprising cationic
residues (usually Arg or Lys, though sometimes His)
separated by Ala, Leu, Met, Phe, Pro, Tyr or Trp resi-
dues. Peptides derived from human or mammalian
sources, such as lactoferrin [2], defensins [3] and
LL-37 [4], are often long and complex, making rational
modification and testing of their entire sequences
difficult, and meaning that such peptides may be too
costly for clinical use.
APOE, the gene coding for human apolipopro-
tein E (apoE), influences the outcome of infection [5].
The APOE-e4 allele is associated with high preva-
lence of cold sores and with an increased risk of Alz-
heimer’s disease in elderly people possessing latent
herpes simplex virus type 1 (HSV1) in the brain [6],
and protection from liver damage in those infected
with hepatitis C virus [7]. In addition, APOE geno-
type influences risk of developing herpes simplex
encephalitis [8], malaria in children [9] and shingles
and postherpetic neuralgia in females caused by vari-
cella zoster virus infection [10]. Many infectious
agents, including viruses and intracellular parasites

such as Plasmodium, use heparan sulfate proteogly-
cans (HSPGs) as their initial attachment sites for
cells and, furthermore, many also bind to low density
lipoprotein receptor (LDLR) family receptors to enter
cells, or interact directly with lipoproteins, with the
latter sometimes being mediated through apolipopro-
teins [11]. ApoE also uses both HSPGs and LDLR
family receptors in its entry pathway. We recently
showed that the region of apoE responsible for entry
(i.e. the HSPG ⁄ LDLR receptor binding region) has
direct and broad anti-infective activity [12], when sta-
bilized by constructing a tandem repeat peptide of
apoE141-149 (an approach widely used to study the
biology of this region) [13], with this octadecamer
peptide being referred to as apoEdp. ApoEdp also
shows antibacterial action, possibly due to its highly
cationic nature, although it is not amphipathic like
many AMPs. In addition, apoEdp also inhibits the
attachment of HSV1 to cells, possibly due to block-
ade of cellular HSPG sites, reflecting its derivation
from the HSPG binding domain of apoE. The human
origin of this sequence, coupled with its broad
anti-infective activity and its lack of haemolytic
action, suggests that it might be modified to obtain
safe and potent AMPs targeted towards serious infec-
tions.
The strategy of developing peptide-based therapies
against HIV or other serious infections is strongly sup-
ported by the clinical success of the 36-residue peptide
Enfuvirtide (T20) HIV-fusion inhibitor. Moreover, this

approach has validated entry blockade as a viable
anti-HIV strategy, and a number of other agents are
now being developed as inhibitors of HIV fusion [14].
Peptides targeting even earlier events in HIV-entry (i.e.
upstream of fusion) could offer new approaches for
developing antiviral compounds [15]. HIV, like many
viruses, has been shown to attach to cellular HSPGs
prior to CD4 attachment [16], potentially offering a
further target for development of HIV therapeutics
[17]. This possibility is strengthened by the finding that
HPSGs on the surface of nonpermissive endothelial
cells may act as a reservoir for the virus and mediate
its transfer to T lymphocytes [16]; similarly, HSPGs
have been implicated in brain invasion by HIV [18].
Plasmodium spp., the parasites responsible for malaria,
also enter (liver) cells after first adhering to HSPGs
[19]. Furthermore, Plasmodium sporozoites bind to a
subset of HSPGs used by apoE-containing lipoproteins
for uptake by the liver [20]. Thus, blockade of HSPGs
may offer an attractive target to prevent cellular inva-
sion by a range of pathogenic organisms.
Here, we dissect those features of the apoEdp pep-
tide involved in its activity, and examine the activity
and mechanism of action of variants of this sequence.
We demonstrate the minimal length for activity, the
contribution of individual residues, and show that sub-
stitutions for aromatic residues increase potency and
breadth of activity, giving rise to further apoE-derived
antimicrobial peptides (apoE-AMPs). In addition, we
show for the first time that the most effective substi-

tuted peptides prevent initial attachment of both
CXCR4- and CCR5-HIV strains and of Plasmodium
berghei to cells, with this most likely involving
increased membrane perturbation associated with their
aromatic groups.
Results
We examined the influence of peptide length on activ-
ity using a series of dodecamer and pentadecamer pep-
tides derived from the octadecamer apoEdp sequence
(we previously found nonomer peptides to be inactive)
(Table 1). Figure 1 shows that the greatest antiviral
and antibacterial action was found in the full length
tandem repeat peptide apoEdp (P<0.001). The eico-
sameric tandem repeat peptide of apoE141-150 (i.e.
including the additional Arg residue found in position
Aromatic substitution of ApoE-derived AMPs B. A. Kelly et al.
4512 FEBS Journal 274 (2007) 4511–4525 ª 2007 The Authors Journal compilation ª 2007 FEBS
150 of apoE) had almost identical antiviral and anti-
bacterial activity to apoEdp. ApoEdp therefore
appears to provide the highest activity relative to pep-
tide length amongst the peptides derived from the
apoE HSPG receptor binding region we tested.
The apoEdp sequence comprises a simple pattern of
three amino acid residues. The contribution of individ-
ual residues to its activity can thus be studied by gen-
eration of a relatively small number of variant peptides
(unlike many host-derived AMPs). We initially tested
whether the cationic residues were critical for its activ-
ity. Replacing either all the Lys or all the Arg residues
with a hydrophobic residue (Trp), or replacement of

both Lys and Arg with His or the acidic residues Asp
or Glu, resulted in peptides with no measurable anti-
viral or antibacterial action (data not shown).
The anti-infective activity of apoEdp depends on the
ability of the peptide to form an a-helical structure,
although the basic residues are not distributed in an
amphipathic pattern [12] as is often the case for a-heli-
cal AMPs. We therefore examined whether the eight
bulky Leu residues separating the peptide’s basic resi-
dues mediated its activity. We prepared peptides in
which all eight Leu residues were substituted by
another amino acid (but maintaining the naturally
occurring distribution of cationic Arg and Lys residues
within the sequence). Substitutions were made with
one of each of the other 16 nonbasic amino acid resi-
dues, and also with His (which is only cationic under
acidic conditions) (Table 1). The ability to inhibit
HSV1 infection was retained by many of these
peptides, with more potent activity found after Cys-
substitution and especially after Trp-substitution
(Fig. 2A,B). Very similar data were found for herpes
simplex virus type 2 (HSV2) (data not shown). We also
examined antibacterial activity in this series of substi-
tuted peptides, and found this to be less prevalent.
Peptides in which the Leu of apoEdp had been substi-
tuted by Ile, Trp and Phe also inhibited Plasmodium
aeruginosa, but none were as potent as apoEdp
itself against this resistant bacterium. Interestingly
Table 1. Amino acid sequences of apoE-derived peptides. Se-
quences of peptides derived from the apoEdp sequence are

shown, including peptides both of altered length and in which
leucine residues had been substituted for another residue (shown
in bold).
Peptide Amino acid sequence
ApoEdp L R K L R K R LLLRKLRKRLL
Truncated peptides
ApoEdp N-3 L R K R LLLRKLRKRLL
ApoEdp N-6 R LLLRKLRKRLL
ApoEdp C-3 L R K L R K R LLLRKLRK
ApoEdp C-6 L R K L R K R LLLRK
ApoE141–149 L R K L R K R L L
Elongated peptides
ApoE141–150 dp L R K L R K R L L R L R K L R K R L L R
Substituted peptides
ApoEdpL-E E RKE RKREEERKE RKREE
ApoEdpL-A A RKA RKRAAARKA RKRAA
ApoEdpL-D D RKD RKRDDDRKD RKRDD
ApoEdpL-W W RKW RKRWWWRKW RKRWW
ApoEdpL-M M RKM RKRMMMRKM RKRMM
ApoEdpL-Y Y RKY RKRYYYRKY RKRYY
ApoEdpL-F F RKF RKRFFFRKF RKRFF
ApoEdpL-I I RKI RKRIIIRKI RKR
II
ApoEdpL-Q Q RKQ RKRQQQRKQ RKRQQ
ApoEdpL-N N RKN RKRNNNRKN RKRNN
ApoEdpL-C C RKC RKRCCCRKC RKRCC
ApoEdpL-S S RKS RKRSSSRKS RKRSS
ApoEdpL-V V RKV RKRVVVRKV RKRVV
ApoEdpL-T T RKT RKRTTTRKT RKRTT
ApoEdpL-G G RKG RKRGGGRKG RKRGG

ApoEdpL-H H RKH RKRHHHRKH RKRHH
ApoEdpL-P P RKP RKRPPPRKP RKRPP
A
B
Fig. 1. Anti-infective activity of peptides constructed from trunca-
tions or elongations of the apoEdp seqeunce. (A) Infectivity of
HSV1 as shown by plaque reduction assay in Vero cells, after treat-
ment of virus with various concentrations of apoE-AMPs, showing
apoEdp has optimum activity. Typical data are shown; bars indicate
standard errors. (B) Growth of P. aeruginosa, in the presence of
various concentrations of ApoE-AMPs, again showing that apoEdp
has optimum activity. Typical data are shown; bars indicate stan-
dard errors.
B. A. Kelly et al. Aromatic substitution of ApoE-derived AMPs
FEBS Journal 274 (2007) 4511–4525 ª 2007 The Authors Journal compilation ª 2007 FEBS 4513
apoEdpL-A, which resembles the previously reported
[21] Ala ⁄ Arg ⁄ Lys antimicrobial peptides derived from
human heparan binding sequences, was inactive as
were the other 13 substituted peptides (Fig. 2A). Even
fewer peptides inhibited the Gram-positive bacterium
Staphylococcus aureus, with the Ala substituted pep-
tides again proving inactive along with the other
12 substituted peptides (Fig. 2A). Tyr and Ile substi-
tuted peptides slightly inhibited S. aureus, whereas Trp
and Phe substituted peptides had IC
50
concentrations
lower than that of apoEdp.
The nontandem-repeat apoE141-149 sequence (previ-
ously found to be inactive, possibly due to its inability

to form a-helix) inhibited HSV1 after substitution of
its four Leu residues with Trp, although this activity
was low (IC
50
concentration ¼ 125 lgÆmL
)1
(95%
confidence interval ¼ 48–160 lgÆmL
)1
) compared with
the tandem repeat of this sequence (apoEdpL-W),
which had an IC
50
concentration of 9.3 lgÆmL
)1
(95%
confidence interval, 8.5–10.5 lgÆmL
)1
) or 3.1 lm
(Fig. 2C). Furthermore, CD measurements revealed
that neither apoEdpL-W or apoEdpL-F have a-helical
structure even in 50% 2,2,2-trifluoroethanol (data not
shown), suggesting that such structure was not
required for their activity, unlike apoEdp [12].
Although the biological effects of apoEdp on mam-
malian cell cultures have been previously reported,
these occur only under unusual physiological condi-
tions; for example, in the absence of serum [22] or the
absence of full length apoE [23]. Indeed, we previously
found that apoEdp had minimal activity in standard

cytotoxicity and haemolytic assays. Here, we examined
whether the substituted peptides showed similar appar-
ent biocompatibility. ApoEdpL-W had no effect
on 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl-tetrazolium
bromide (MTT) reduction in Vero cells, even after
2 days of incubation at 40 lm (Fig. 3A). We previ-
ously found that apoEdp had no haemolytic action,
and thus examined this for apoEdpL-W, finding only
mild haemolytic action at high concentrations (15%
red blood cell lysis at 35 lm) (Fig. 3B), with this being
much less than that for other cationic antimicrobial
peptides tested. Indeed, Fig. 3C shows that the related
Trp-substituted peptide apoE 141–150dpL-W shows
only 30% haemolysis even when tested at 160 lm.By
contrast, we found similar levels of haemolysis with
AB
C
Fig. 2. Anti-infective activity of Leu-substituted apoEdp-derived peptides, whereby all eight Leu residues were substituted for another resi-
due. (A) Relative antiviral and antibacterial activity of Leu-substituted forms of apoEdp (unsubstituted apoEdp is indicated by the letter ‘L’).
Values shown are average IC
50
concentrations (bacteria) or IC
25
concentrations (HSV1) in lM, after turbidity or plaque reduction assay.
(B) Anti-HSV1 activity of apoEdp and apoEdpL-W, measured by a plaque reduction assay. Typical data are shown; bars indicate standard
errors. (C) Anti-HSV1 activity of apoEdpL-W relative to that for the nonomer apoE141–149-L-W. Concentrations are plotted in weight per vol-
ume rather than moles per volume, thus allowing a direct comparison of the 18mer apoEdpL-W with activity of its two 9mer components
when not joined together. Activity was measured by plaque reduction assay. Typical data are shown; bars indicate standard errors.
Aromatic substitution of ApoE-derived AMPs B. A. Kelly et al.
4514 FEBS Journal 274 (2007) 4511–4525 ª 2007 The Authors Journal compilation ª 2007 FEBS

the previously described lytic antimicrobial peptides
RLLR5 and N-RLLR3 when these were used at 5 lm
and 31 lm, respectively. In conclusion, we found no
significant cytotoxic or haemolytic action of these pep-
tides at concentrations of apoEdp, apoEdpL-W, and
apoE141–150dpL-W found to be efficacious against
microorganisms.
We also compared the activities of other previously
described antimicrobial peptides that were constructed
from Trp, Arg, Leu or Lys residues alone, like the
apoE-AMPs described here (Table 2). Few of these
peptides showed the broad anti-infective activity of
apoEdp-derived peptides, although constructing
tandem repeats of some of these previously described
shorter peptides did increase activity, as had been
found for the apoE141–149 sequence, suggesting that
peptides of length 14–18 consisting of at least two
charged regions separated by hydrophobic residues
may be an optimum feature for AMPs. The strongest
antiviral activity was found in the buforin-related
peptide RLLR5 [24,25]; however, the antibacterial
effects of this peptide were limited, perhaps due to the
presence of physiological levels of salt in our assays,
known to reduce the anti-infective activity of this
peptide [25].
ApoEdp inhibits HIV replication and, for herpes
viruses, blocks the early stages of replication [12].
Accordingly, we next examined whether the more
potent peptide apoEdpL-W might inhibit the early
stages of HIV replication. In initial experiments, we

found that apoEdpL-W had strong inhibitory action
against HIV-IIIB, under conditions in which the pep-
tide was added with viral inoculum, and then washed
from the system, suggesting that an early stage in repli-
cation was targeted. Interestingly, apoEdp itself did
not show activity in this assay system, possibly reflect-
ing either the low concentrations of peptide tested or
the different strain of virus used. ApoE141–149 and
the control peptide apoE263–286 also had no activity,
as expected (Fig. 4A). To test whether the action might
involve selectively targeting either CXCR4 or CCR5
coreceptors during viral fusion, we tested activity
against HIV strains that use one or other or both these
coreceptors. Similar strong activity was found against
strains 2028 (CXCR4 ⁄ CCR5), 2044 (CXCR4) as well
as SF162 and BaL (CCR5) (Fig. 4B), suggesting that
A
C
B
Fig. 3. Lack of toxicity of apoE-AMPs against mammalian cells. (A) MTT reduction in Vero cells treated for extended periods with apoE-
AMPs. Values shown are expressed relative to untreated control wells; bars indicate standard errors. (B) Haemolytic assessment of apoE-
AMPs. ApoEdp and other apoE-AMPs were introduced to red blood cells at various concentrations, Values shown indicate A
540
relative to
that for red blood cells completely lysed in 0.45% ammonium hydroxide; bars indicate standard errors. (C) Relative haemolytic activity of
apoE-AMPs compared with that for the lytic peptides N-[RLLR]
3
and RLLR5 [24,25]. Values shown indicate A
540
relative to that for red blood

cells completely lysed in 0.45% ammonium hydroxide; bars indicate standard errors.
B. A. Kelly et al. Aromatic substitution of ApoE-derived AMPs
FEBS Journal 274 (2007) 4511–4525 ª 2007 The Authors Journal compilation ª 2007 FEBS 4515
the peptide does not operate through blockage of one
or other coreceptor.
To test whether activity may relate to blockage of a
step prior to viral fusion (e.g. initial attachment of virus
particles to the cell surface), assays were repeated at
4 °C, and 10 lm peptide was introduced either before
the virus or after initial viral preadsorption to cells. In
neither case was antiviral activity abolished or signifi-
cantly diminished relative to that found in earlier exper-
iments (Fig. 4C), suggesting that a step prior to viral
fusion was targeted involving disruption of reversible
attachment of virus to cells, or a direct lytic action on
virus occurring whether or not the virus was cell-bound.
To confirm this, we tested whether apoEdpL-W
could also inhibit the entry of vesicular stomatitis virus
G protein (VSV-G) pseudotyped HIV-1. This virus is
identical to HIV-1, but has a VSV envelope, with its
entry mediated by clathrin-mediated endocytosis lead-
ing to pH dependent fusion [26,27]. We found that this
virus was not inhibited (Fig. 4D), demonstrating that
the action of apoEdpL-W against HIV was relatively
selective, either involving blockage of attachment of
virus to the cell or by interfering with early envelope-
mediated events although without compromising mem-
brane integrity.
The Plasmodium sporozoite (i.e. the infective stage
of the malaria parasite) enters hepatocytes via attach-

ment to HSPGs, and apoE-containing lipoproteins can
inhibit sporozoite infectivity both in vitro and in vivo
[20]. Using a rodent model of the disease, we therefore
examined whether apoEdp, apoEdpL-F and ⁄ or apo-
EdpL-W could also disrupt this process. Figure 5A
shows that both apoEdp and apoEdpL-W inhibit entry
of P. berghei into Hepa 1-6 cells; the effect of apo-
EdpL-F at this level was not significant. We also con-
firmed that the active peptides had no cytotoxic action
against Hepa 1-6 cells (as this might account for the
apparent Plasmodium inhibition): MTT reduction was
not inhibited after treatment of these cells with
5–25 lm apoEdp or apoEdpL-W for 2 days (i.e. 48
times the length of incubation in the Plasmodium inhi-
bition assays) (Fig. 5B).
In further assays in which nonbound peptide was
washed away prior to introducing the sporozoites, apo-
EdpL-W (but not apoEdp) retained its ability to block
P. berghei entry (Fig. 5C). These data suggest apo-
EdpL-W is likely binding to the cell surface irreversibly,
thus preventing entry of the parasite. Consistent with
this, we found that apoEdpL-W had no action on the
attachment of sporozoites to cells, but in fact blocks
invasion of the parasite after attachment (data not
shown). Additionally apoEdpL-W had no direct action
against the parasite itself, unlike apoEdp, which caused
extensive lysis of the sporozoites (Fig. 5D). Thus,
apoEdp appears to directly inactivate the sporozoite,
whereas apoEdpL-W likely acts through interactions
with the mammalian cell membrane.

We next examined the relative ability of apoE-AMPs
to perturb model membrane systems, to indicate their
relative propensity for their biological activity to be
mediated through a direct effect on membrane lipid.
Wideline deuterium NMR spectra were collected to
Table 2. Antiviral and antibacterial activities of various non-apoE-derived antimicrobial peptides comprised predominantly of Leu or Trp and
Arg. Anti-HSV1 activity was assessed by plaque reduction assay, antibacterial assays were carried out using turbidity assessment of serial
dilutions of peptide inoculated with bacteria. In both cases, concentrations yielding 50% inhibition of growth were obtained from the plots
(averages for several experiments are shown). Peptides had either been used previously in other studies, were tandem repeats of those
peptides (indicated by the subscript ‘dp’), or were devised for the present study (MU103 and MU104).
Peptide Amino acid sequence
IC
50
concentration (lM)
HSV1
Pseudomonas
aeruginosa
Staphylococcus
aureus
Octa 1
a
RRWYRWWR >20 21 42
MU89 (Octa 1 dp) R R W Y R W W R R R W Y R W W R 8.3 30 20
Hepta 1
a
R W W R W W R > 20 28.5 14
MU92 (Hepta 1 dp) R W W R W W R R W W R W W R 3.3 > 53 > 53
Deca 1
a
YRWWRWARRW >20>53 >53

MU94 (Deca 1 dp) Y R W W R W A R R W Y R W W R W A R R W 5 > 53 > 53
N-[RLLR]
3
b
APKAMRL L RRLLRRL L R 7.3 >53 2
RLLR5
b
RL LRRL L RRL LRRL L RRLLR<3.3 28 16
MU103 R R R R R R R W W W R R R R R R R R > 20 > 53 9
MU104 R R R R R R R R R R R R R R R W W W 9.25 > 53 26
ApoEdpL-W W R K W R K R W W W R K W R K R W W 3.1 7 7
a
Sequences described in Strom et al. [30].
b
Sequences described in Park et al. [24,25].
Aromatic substitution of ApoE-derived AMPs B. A. Kelly et al.
4516 FEBS Journal 274 (2007) 4511–4525 ª 2007 The Authors Journal compilation ª 2007 FEBS
detect interactions of the apoEdp, apoEdpL-W and
apoEdpL-F derived peptides with the surface of
multilamellar vesicles composed of 1,2-dimyristoyl-
phosphatidyl-1,1,2,2-
2
H
4
-choline (DMPC-d
4
) and di-
oleoylphosphatidylglycerol (DOPG) mixed in a 2 : 1
molar ratio. The NMR spectrum arises from the repor-
ter molecule, choline methylene-deuterated DMPC, and

is sensitive to any interactions between peptides and the
membrane surface that alter the mean orientation of
the choline moiety [28]. Such orientational changes are
manifest in the spectrum as changes in the quadrupole
splittings, which is the frequency separation between
the two pairs of Pake doublets for the a- and b-deute-
rons. The anionic DOPG headgroups provide an over-
all negative surface charge and this lipid membrane
system has previously been used in NMR studies of this
type to provide a model membrane for studying the
behaviour of amphipathic peptides [29].
The peptides were titrated into the sample to a
lipid ⁄ peptide molar ratio of 50 : 1 or 20 : 1. The
dashed lines in Fig. 6A indicate the scaled changes in
splittings (an increase for the b-deuterons and a
decrease for the a-deuterons) consistent with the pep-
tide interacting with lipid head groups at increasing
concentration. ApoEdp appears to interact relatively
weakly with the mixed lipid bilayers of the multila-
mellar vesicles because it induces only minor changes
in the splitting values, yet aromatic substitution greatly
enhances the effect on the splittings. The marked
effects of apoEdpL-W and apoEdpL-F on the spec-
trum may reflect a high affinity of these peptides for
the lipid head groups or a greater destabilizing effect
on the membrane surface, perhaps as a result of the
aromatic groups inserting deeper into the lipid bilayer.
By contrast, the spectra in the presence of apoEdp are
consistent with the peptide being situated away from
AB

CD
Fig. 4. Anti-HIV activity of the apoE-AMP apoEdpL-W. (A) Potent anti-HIV action of apoEdpL-W relative to apoEdp, apoE141–149 and
apoE263–286. Experiments were performed by treating NP2 cells with peptide prior to introducing HIV IIIB. Values show level of HIV infec-
tion (in FFU) relative to untreated control; bars indicate standard errors. (B) Anti-HIV-1 activity of apoEdpL-W against CXCR4 using HIV-1
strain 2044, CXCR4 ⁄ CCR5 using HIV-1 strain 2028, and CCR5 using HIV strains SF162 and BaL. Values show the level of HIV infection (in
FFU) relative to untreated control; bars indicate standard errors. (C) Anti-HIV-1 activity of apoEdpL-W after incubation of virus with NP2 cells
at 4 °C, either before or after addition of peptide. Values show the level of HIV infection (in FFU) relative to untreated control; bars indicate
standard errors. (D) Lack of effect of apoEdpL-W on VSV-G infection of NP cells. Virus was added to cells at 4 °C, either in the presence or
absence of apoEdpL-W at 10 l
M (a concentration that strongly inhibits HIV). Infectivity was assessed by enumerating green fluorescence
protein-positive cells using fluorescence activated cell sorting (Becton Dickinson) 48 h after transduction.
B. A. Kelly et al. Aromatic substitution of ApoE-derived AMPs
FEBS Journal 274 (2007) 4511–4525 ª 2007 The Authors Journal compilation ª 2007 FEBS 4517
the membrane surface and participating in a weak elec-
trostatic interaction. In all three cases, the degree of
interaction scales with increasing peptide concentration
confirming that the observed effects are peptide-medi-
ated (Fig. 6B) but suggesting that the binding sites for
the peptides are not fully saturated at lipid ⁄ peptide
ratios of greater than 20 : 1.
Discussion
We previously reported that a highly cationic sequence
within human apoE has broad anti-infective properties
when presented as a tandem repeat peptide [12]. Here,
we have shown that a large family of similarly active
peptides can be obtained by systematic modification of
this highly cationic human sequence. The activity of
apoEdp-AMPs was greatest for tandem repeat octa-
decamer sequences and, as expected, was abolished by
replacing cationic residues. We found that substitution

of some or all of the Leu residues with the aromatic
residue Trp increased the potency of activity for most
species, although unsubstituted apoEdp itself was most
active against the Gram negative bacterium P. aerugi-
nosa. Additionally, Phe substitutions increased antibac-
terial activity against S. aureus, and Cys substitutions
maintained anti-HSV1 activity. Other substitutions for
Leu reduce or abolished antiviral or antibacterial activ-
ity. In general, anti-infective activity was associated
with peptides where bulkier residues separated the
cationic amino acids, although exceptions occurred
(e.g. apoEdpL-Y had little antibacterial activity).
Furthermore, our data show that the increased
inhibitory action of apoEdpL-W and apoEdpL-F
against certain bacteria and viruses, including HIV, and
the stronger association of apoEdpL-W peptide with
mammalian cell membranes in the Plasmodium assays,
may be caused by stronger membrane interactions of
these aromatic substituted peptides. This may directly
AB
CD
Fig. 5. Inhibition of Plasmodium invasion by apoE-AMPs. (A) Plasmodium berghei infection of Hepa 1-6 cells after incubation of sporozoites
and cells with various apoE-AMPs at 50 lgÆmL
)1
. Values shown are expressed relative to untreated control wells; bars indicate standard
errors. (B) Confirmation that growth of Hepa 1-6 cells was not inhibited by apoE-AMPs. Hepa 1-6 cells were grown in the presence of up to
25 l
M (> 60 lgÆmL
)1
) apoEdp or apoEdpL-W for 2 days, before assessment of cell viability by MTT reduction. Values shown are expressed

relative to untreated control wells; bars indicate standard errors. (C) Effect of pretreatment of Hepa 1-6 cells with apoEdp and apoEdpL-W at
50 lgÆmL
)1
on entry of P. berghei, after washing peptide-treated cells prior to introduction of Plasmodium. Values shown are expressed rela-
tive to untreated control wells. (D) Effect of apoEdp and apoEdpL-W treatment (50 lgÆmL
)1
)onP. berghei membrane integrity, assessed by
propidium iodide exclusion relative to organisms treated with heat (positive control) or untreated (negative control). Values show proportion
of organisms taking up stain; bars indicate standard errors.
Aromatic substitution of ApoE-derived AMPs B. A. Kelly et al.
4518 FEBS Journal 274 (2007) 4511–4525 ª 2007 The Authors Journal compilation ª 2007 FEBS
damage bacterial membranes, or anchor peptides to
mammalian membranes, allowing later influence on
HIV or Plasmodium invasion of cells.
The activity of apoE-AMPs appears to be broader
and more potent than that of comparable short (pent-
amer to undecamer) Trp ⁄ Arg ⁄ Tyr ⁄ Ala peptides previ-
ously described [30], which showed only modest
antibacterial activity in our assays, and no antiviral
activity (Table 2). We found that activities were lower
than those for apoE-AMPs; however, antiviral activity
was found in tandem repeat peptides constructed from
the latter (paralleling the increase in activity obtained
by presenting apoE141–149 as a tandem repeat). These
data suggest that only cationic peptides of a certain
length have potent antiviral activity. Nonetheless,
unlike apoE141–149, constructing tandem repeats of
the short Trp ⁄ Arg ⁄ Tyr ⁄ Ala peptides did not increase
antibacterial activity; indeed, in the case of Hepta1dp,
activity was abolished.

The antibacterial effects of the most potent apoE-
AMPs compares favourably with published activities
for commercially available peptidic antimicrobials. For
example, the minimum inhibitory concentration (MIC)
of nisin against Pseudomonas aeruginosa is reported to
be 32 lgÆmL
)1
[31], whereas the same value for
apoEdp is 7.3 lgÆmL
)1
or 3 lm (Fig. 1B). Similarly
magainin II and cecropin P1, both commercially avail-
able antimicrobial peptides, have reported MICs
A
B
Fig. 6. Wideline NMR spectra (at 303 °K) and corresponding quadrupole splitting values for multilamellar vesicles composed of 2 : 1 DMPC-
d
4
⁄ DOPG, before and after introduction of apoEdp, apoEdpL-F and apoEdpL-W. (A) Spectra are shown with dashed lines indicating the
changes in a- and b-splittings observed for untreated lipid (bottom) and after introduction of each of the three peptides to a lipid ⁄ peptide
molar ratio of 50 : 1 (middle) and 20 : 1 (top). The distance between outer pair of dotted lines represents the splittings for the choline a -deu-
terons and the distance between the inner pair represents the splitting for the choline b-deuterons. (B) Peptide concentration-dependent
changes in the values of the measured quadrupole splittings for the choline a- and b-deuterons after addition of apoEdpL-W, apoEdpL-F and
apoEdp.
B. A. Kelly et al. Aromatic substitution of ApoE-derived AMPs
FEBS Journal 274 (2007) 4511–4525 ª 2007 The Authors Journal compilation ª 2007 FEBS 4519
against (nonresistant) S. aureus strains in the range
16–128 lgÆmL
)1
[32], whereas we found apoEdpL-F to

have an MIC against S. aureus of 9.5 lm or
25.5 lgÆmL
)1
(data not shown).
Cationic AMPs may act against bacteria by disrupt-
ing the negatively charged inner membrane, to which
such peptides are electrostatically attracted [33]. At
least one function of the positive charge associated
with such peptides is to promote selective adherence of
the peptide to the bacterial membrane. The ‘carpet’
model is one of several proposed to explain cell death,
in which peptides accumulate on the membrane sur-
face and cover (carpet) the bilayer, ultimately destroy-
ing the membrane by a detergent-like action [34].
Alternately monomers may form either ‘barrel-stave’
[35] or ‘toroidal’ [36] pores in the membrane, with loss
of cellular contents or membrane potentially killing
the bacterial cell. A further model proposes the forma-
tion of pores formed by peptide aggregates, allowing
entry of further peptides into the cell and interference
with either protein synthesis or DNA replication [37].
It is unlikely that the octadecamer apoEdp-AMPs
would form barrel stave pores because the membrane
spanning barrel stave mechanism is linked to peptides
greater than 22 amino acids in length [36], and
because they are not amphipathic [12]. However, both
the ‘carpet’ and ‘aggregate pore’ mechanisms do
appear possible.
One structural motif previously considered to medi-
ate binding to bacterial lipopolysaccharide (LPS) is

two positive amino acid residues separated from a
third by several hydrophobic residues. Such motifs are
found in a number of antimicrobial peptides, including
bovine lactoferrin (RRWQWR) [38], polyphemusin 1
(RRWCFR) and tachylpepsin (KWCFR) [39] and the
synthetic hexapeptide RRWWCR [40]. ApoEdp and
apoEdpL-W contain similar motifs (‘KRLLLR’ and
‘KRWWWR’, respectively), which might therefore be
suitable to mediate such interactions with LPS. How-
ever our finding that the pentadecamer and dodecamer
apoEdp-related peptides (which also contain the
‘KRLLLR’ motif) showed relatively little antibacterial
activity suggests this is not the case. Nonetheless, it
would be interesting to examine whether apoE-AMPs
might inactivate immunostimulatory LPS released
from bacterial cells.
The anti-HSV1 activity of apoEdp involves inhibition
of virus particle attachment to cells, with this likely
being related to the derivation of this peptide from the
apoE HSPG ⁄ LDLR binding region [12]. Inhibition of
viral (or Plasmodium) attachment to cells may relate to
the blockade of either HSPGs on the eukaryotic cell sur-
face or blockade of LDLR family receptors. An alterna-
tive would be direct lysis of the virus or parasite, which
appears to occur with apoEdp itself and P. berghei,
thereby indirectly preventing their attachment to
cells. Our finding that VSV-pseudotyped HIV was not
inhibited by apoEdpL-W suggests that the effect of the
latter does at least in part involve disruption specific to
the HIV viral membrane. This is consistent with our

finding that apoEdpL-W had a far greater propensity to
perturb model membrane systems than apoEdp. Such
membrane interactions did not appear to directly medi-
ate the activity of apoEdpL-W against Plasmodium,
which appeared to remain viable after exposure to
apoEdpL-W. In conclusion, the mechanism of action
of apoEdpL-W against HIV may involve a selective
biophysical detergent-like action or attachment inhibi-
tion, whereas its activity against Plasmodium sporozoites
involves inhibition of parasite invasion of cells, medi-
ated through attachment of the peptide to mammalian
cells. These findings are consistent with those of a recent
study which demonstrated that the membrane effects of
Leu- and Trp-containing cationic peptides varies consi-
derably, depending on the distribution of cationic
residues (and resulting degree of amphipathicity), the
hydrophobic residues and the nature of the membrane
interacting with the peptide [41].
Only a subset of AMPs have antiviral activity. For
example, indolicidin, which superficially resembles
apoEdpL-W, is relatively inactive against HIV, with a
reported IC
50
concentration of 35–50 lm [42]. A
recent study surveyed potential anti-HIV activity in a
range of amphibian-derived AMPs, and found only
three peptides with suitable activities [15]. ApoE-
AMPs are active against HIV in the single digit lM
concentration range, unlike the low concentration
(nm) activity range of many small molecule anti-HIV

lead compounds. However, unlike many such com-
pounds, apoE-AMPs also appear nontoxic in the lM
range. ApoE-AMPs offer a means to interfere with a
very early stage in the attachment of HIV (and herpes-
viruses) to cells, and with peptides based on a human
cationic sequence. The HIV strains inhibited included
those using both CCR5 and CXCR4 coreceptors
(2028), those using CXCR4 alone (2044 and IIIB) or
those using CCR5 alone (SF162 and BaL). With the
exception of IIIB, all can infect both macrophages as
well as CD4
+
T cells [43]. These data support a tar-
get upstream of viral fusion events (or upstream of
those fusion events involving HIV coreceptors). In
addition, the clear inhibition of HIV when peptide
was introduced to cells both before and after the
virus adsorbs at 4 °C suggests that any action
involves blockade of the reversible attachment of the
virus to cells.
Aromatic substitution of ApoE-derived AMPs B. A. Kelly et al.
4520 FEBS Journal 274 (2007) 4511–4525 ª 2007 The Authors Journal compilation ª 2007 FEBS
ApoEdpL-W has much greater anti-HIV potency
than apoEdp, suggesting apoE-AMPs may be developed
as a new approach for HIV-therapy. The early anti-HIV
target of apoEdpL-W suggests that it may also be used
as a microbicide, especially because it also inhibits
HSV2, an important cofactor for the transmission and
acquisition of HIV infection [44]. Additionally, com-
pounds preventing Plasmodium sporozoite entry into

the liver would represent an attractive new means to pre-
vent infection by this parasite. ApoE-AMPs have
unusually broad anti-infective activity, relative to well-
known AMPs such as cecropins, clavanins and LL-37,
which show little antiviral activity [45], and are highly
active, with IC
50
concentrations against many organisms
(including HIV and P. aeruginosa) in the 1–4 lm region.
Although there is precedent for safe clinical use of pep-
tides based on nonhuman sequences (e.g. the 36 amino
acid peptide T20), such peptides run the risk of an
immunogenic response. The human origin and relatively
simple nature of the apoEdp sequence should expedite
the rational design of apoE-AMPs for potential clinical
use.
Experimental procedures
Cell cultures
Vero cells were maintained in Eagle’s minimum essential
medium supplemented with 10% (v ⁄ v) fetal bovine serum,
2mml-glutamine, penicillin (100 IU ÆmL
)1
) and strepto-
mycin (100 lgÆmL
)1
), hereafter referred to as growth med-
ium. Hepa 1-6 cells (ATCC CRL-1830, American Type
Culture Collection, Manassas, VA, USA) were maintained
in Dulbecco’s modified Eagle’s medium containing 10%
(v ⁄ v) fetal bovine serum, hereafter referred to as growth

medium (DMEM). NP2 ⁄ CD4 ⁄ CXCR4 or NP2 ⁄ CD4 ⁄ CCR5
cells were maintained in growth medium (DMEM) [46].
Growth medium containing only 2% or 0.5% fetal bovine
serum is referred to as ‘2% medium’ or ‘0.5% medium’.
Microorganisms
HSV1 stocks (strain SC16 provided by Professor Roy Jen-
nings, Sheffield University, UK) and HSV2 stocks (clinical
isolate provided by Professor Anthony Hart, Liverpool
University, UK) were prepared in Vero and HEp2a cells,
respectively [12]. HIV-1 stocks were prepared in human
peripheral blood mononuclear cells stimulated with phyto-
hemagglutinin and interleukin-2 except strain IIIB which
was prepared in H9 T cells. Titres determined on NP2 ⁄ CD4
cells bearing the appropriate coreceptor by counting p24+
foci by immunostaining [47]. The VSV-G pseudotype of
HIV-1 based vectors were made by transient transfection of
293T cells with p8.91 (HIV-1 Gag-Pol), pCSGW (Lentivec-
tor genome encoding enhanced green fluorescent protein
under spleen focus-forming virus promoter control) and
pMD-G (CMV-VSV-G) [48]. Infectivity of stocks was mea-
sured by enumerating green fluorescence protein-positive
cells using fluorescence activated cell sorting (Becton Dick-
inson), 48 h after transduction.
Bacterial stocks were grown by inoculating Luria–Bertani
broth with either P. aeruginosa (ATCC strain 9027) or
S. aureus (ATCC strain 6538P) obtained in Cultiloop for-
mat (Oxoid, Basingstoke, UK) [12]. Plasmodium berghei
stocks were prepared by feeding 3–5 day-old Anopheles
stephensi mosquitoes on anesthetized P. berghei (NK65)-
infected Swiss Webster mice, which had been checked by

blood smear for the abundance of gametocyte-stage para-
sites. Salivary gland sporozoites were harvested on days
18–21 postinfective blood meal. The mosquitoes were rinsed
in 70% ethanol and washed in DMEM before salivary
gland dissection. The glands were gently ground, centri-
fuged (80 g for 3 min using an Eppendorf microfuge, model
5417C, fixed angle rotor S45-30-11) to remove mosquito
debris, and sporozoites counted in a hemocytometer. Ethics
approval for the use of experimental animals was provided
by NYU Animal Ethics Committee (IACUC ref, Animal
Study Protocol 050310-03).
Peptides
Peptides were obtained commercially (Alta Bioscience, Bir-
mingham, UK) having been synthesized using 9-fluorenyl-
methyl carbamate chemistry and purified by HPLC as
described previously [12]. For peptides apoEdp and apo-
EdpL-W, peptide weight was confirmed by amino acid
analysis, and far-UV CD spectra obtained for peptides sol-
ubilized in either distilled water or 50% trifluoroethanol.
Measurements were carried out at 20 °C, using a Jasco
J810 spectropolarimeter (Jasco Inc., Easton, MD, USA).
Peptide stocks were solubilized in NaCl ⁄ Pi or growth med-
ium at 400 lm, aliquoted and stored at ) 80 °C.
Herpes virus plaque reduction assays
Confluent Vero or HEp2a cells in 24-well plates were inocu-
lated with 90 plaque forming units per well HSV1 or HSV2
in 0.5% medium, containing various concentrations of pep-
tide. This was removed after 1 h, and 1% medium containing
0.2% high viscosity carboxymethylcellulose was added. After
2 days of incubation, cells were fixed in formal saline, stained

with carbol fuchsin and plaques were enumerated. Concen-
trations of peptide that inhibited infection by 50% were cal-
culated (IC
50
concentrations). Because peptide was only
present during inoculation, the assay measured inhibition of
only the very early steps of infection; we have previously
confirmed this using acyclovir and heparin controls [12].
B. A. Kelly et al. Aromatic substitution of ApoE-derived AMPs
FEBS Journal 274 (2007) 4511–4525 ª 2007 The Authors Journal compilation ª 2007 FEBS 4521
HIV inhibition assays
Dilutions of peptide were added to NP2 ⁄ CD4 ⁄ CXCR4
or NP2 ⁄ CD4 ⁄ CCR5 cells which had been plated at
1 · 10
4
cells per well the previous day. Peptides were
allowed to associate with the cells for 30 min before addi-
tion of 100 focus forming units (FFU) of the test virus
(with additional peptide to maintain peptide concentration).
Cells were washed 2 h post infection and cultured for 72 h.
After fixation in cold acetone methanol (1 : 1), cells were
immunostained for HIV p24 with secondary b-galactosidase
conjugate (stained with 5-bromo-4-chloro-3-indolyl-b-d-
galactopyranoside), and the recovered titre enumerated.
For experiments in which the activity of peptides was exam-
ined both before and after viral attachment to cells,
100 FFU of virus was bound to NP2 ⁄ CD4 cells expressing
CXCR4 or CCR5 at 4 °C for 1 h in the presence or
absence of peptide. Cells were washed and the medium
replaced (with or without peptide). Cells were incubated at

37 °C for a further 2 h and then processed as above.
Plasmodium assays
Sporozoite development assay
Hepa 1-6 cells, a mouse hepatoma cell line permissive
for P. berghei sporozoite development, were seeded
(8 · 10
4
cells ⁄ well) in Laboratory-Tek permanox chamber
slides (Nalgene Nunc Corp., Naperville, IL, USA) and
grown until confluent. On the day of the experiment, spor-
ozoites were dissected from mosquitoes and preincubated in
DMEM with 1% bovine serum albumin alone or with the
indicated peptide for 45 min at 28 °C and plated on cells in
the continued presence of the peptide in growth medium.
After 1 h at 37 °C, the medium with unattached sporozoites
and peptide was removed and replaced with complete med-
ium. Twenty four hours later, the cells were fixed with cold
methanol and developing exoerythrocytic stages were
stained with mAb 2E6 [49] followed by anti-mouse immu-
noglobulin conjugated to fluorescein isothiocyanate. The
number of exoerythrocytic stages in each well was counted
with a · 40 objective on a Nikon fluorescent microscope
(Nikon Corp., Tokyo, Japan). In other experiments,
Hepa 1-6 cells were pretreated with the peptides in com-
plete medium for 1 h, washed and then sporozoites were
added in medium without peptide and the assay was con-
tinued as outlined above.
Sporozoite toxicity assay
To check for toxicity of the inhibitors on sporozoites, we
used propidium iodide. Plasmodium berghei sporozoites

were incubated with the peptides for 45 min at 28 °C, pro-
pidium iodide was then added to a final concentration of
1 lgÆmL
)1
for 10 min at 28 °C, sporozoites were washed
and viewed under a fluorescent microscope. Controls
included sporozoites that were heat-killed (and so the
majority should have taken up the propidium iodide) and
sporozoites incubated in medium without the peptide.
Antibacterial assays
A microdilution method was used [12]: paired dilutions of
compounds in Luria–Bertani broth, arranged in 96-well
plates were inoculated with around 1 · 10
5
colony forming
units of P. aeruginosa (ATCC 9027) or S. aureus (ATCC
6538P). After overnight incubation at 37 °C, absorbance at
620 nm (A
620
) was assessed, and the IC
50
concentrations
were determined.
Haemolytic assays
Fresh washed human red blood cells were added to pep-
tides diluted in NaCl ⁄ Pi in 96-well plates (20 · 10
6
red
blood cells were added per well). After 2 h of incubation at
37 °C, plates were centrifuged (3000 g for 5 min using a

Sorvall RT6000B centrifuge and H1000B rotor) and 80 lL
of supernatant transferred to further 96-well plates contain-
ing 0.75% ammonium hydroxide in distilled water. After
assessment of absorbance (A
540
), the concentrations of pep-
tide resulting in 5% or 50% haemolysis (referred to as EC
5
and EC
50
concentrations) were calculated (100% haemoly-
sis was considered to be the average A
540
for red blood cells
treated directly with 0.45% ammonium hydroxide).
Data analysis
For anti-infective assays, activity was expressed as percent-
age reduction relative to control. Standard error was calcu-
lated using a special case of Fieller’s theorem, and
significance assessed using analysis of variance.
Cytotoxicity assessment
Vero cells growing in 96-well plates were treated with vari-
ous peptide concentrations. After 48 h of incubation,
25 lL MTT in 0.5% medium was added (1 mgÆmL
)1
final
concentration), and cells were incubated for 2 h before
removal of growth medium and solubilization of formazan
crystals in 100 lL of dimethyl sulfoxide, prior to reading
absorbance (A

570
).
Solid-state NMR
The lipids DMPC-d
4
and DOPG in chloroform were mixed
in a 2 : 1 molar ratio, respectively, with a total lipid content
of 10 mg (Avanti Polar Lipids, Inc., Alabaster, AL, USA).
The sample was then dried under argon and then under
Aromatic substitution of ApoE-derived AMPs B. A. Kelly et al.
4522 FEBS Journal 274 (2007) 4511–4525 ª 2007 The Authors Journal compilation ª 2007 FEBS
reduced pressure overnight. The thin film of lipids was then
resuspended in 50 lL of deuterium depleted phosphate
buffer (20 mm, pH 7.4) with 1 mm EDTA and transferred
to a 4 mm diameter zirconia magic angle spinning NMR
rotor. In some cases, the total lipid was reduced to attain
the maximum lipid : peptide molar ratio of 20 : 1 (main-
taining the lipid concentration by decreasing the volume of
resuspension buffer). The lipid mixture was then vigorously
mixed to promote the formation of vesicles, followed by
five cycles of freeze thawing on dry ice to integrate the pep-
tide into the multilamellar vesicles that form. A Bruker
Avance 400 NMR spectrometer (Bruker, Ettlingen, Ger-
many) equipped with a magic angle spinning triple reso-
nance probe and cooling unit was used to make deuterium
NMR measurements (at 61 MHz) on nonspinning samples.
The NMR experiments were carried out at 303 °K using a
single deuterium p ⁄ 2 pulse of 4.5 ls and a 1 s recycle time.
Typically, 6 °K transients were accumulated for every
10 mg lipid sample, increasing to 80 °K for the smallest

lipid sample (1 mg) to maintain a satisfactory signal.
Acknowledgements
We thank Ruth Itzhaki, Matthew Wozniak, and Keith
Crutcher for review of the manuscript, and Andrew
Doig for discussion of peptide CD and structural data.
C.B.D. and B.A.K. were supported by grants from
UMIP Ltd, and the Manchester Technology Fund.
A
´
.McK. and S.J.N. were funded by a grant awarded
to Professors Robin Weiss and A
´
ine McKnight by the
Medical Research Council (UK). P.S. was supported
by NIH R01 AI056840.
References
1 Jenssen H, Hamill P & Hancock RE (2006) Peptide
antimicrobial agents. Clin Microbiol Rev 19, 491–511.
2 Farnaud S, Spiller C, Moriarty LC, Patel A, Gant V,
Odell EW & Evans RW (2004) Interactions of lacto-
ferricin-derived peptides with LPS and antimicrobial
activity. FEMS Microbiol Lett 233, 193–199.
3 Wu Z, Li X, de Leeuw E, Ericksen B & Lu W (2005)
Why is the Arg5-Glu13 salt bridge conserved in mam-
malian alpha-defensins? J Biol Chem 280, 43039–43047.
4 Durr UH, Sudheendra U.S. & Ramamoorthy A (2006)
LL-37, the only human member of the cathelicidin family
of antimicrobial peptides. Biochim Biophys Acta 1758,
1408–1425.
5 Burgos JS, Ramirez C, Sastre I & Valdivieso F (2006)

Effect of apolipoprotein E on the cerebral load of latent
herpes simplex virus type 1 DNA. J Virol 80, 5383–5387.
6 Dobson CB, Wozniak MA & Itzhaki RF (2003) Do
infectious agents play a role in dementia? Trends
Microbiol 11, 312–317.
7 Wozniak MA, Itzhaki RF, Faragher EB, James MW,
Ryder SD & Irving WL (2002) Apolipoprotein E-epsi-
lon 4 protects against severe liver disease caused by hep-
atitis C virus. Hepatology 36, 456–463.
8 Lin WR, Wozniak MA, Esiri MM, Klenerman P &
Itzhaki RF (2001) Herpes simplex encephalitis: involve-
ment of apolipoprotein E genotype. J Neurol Neurosurg
Psychiatry 70, 117–119.
9 Wozniak MA, Faragher EB, Todd JA, Koram KA,
Riley EM & Itzhaki RF (2003) Does apolipoprotein E
polymorphism influence susceptibility to malaria? J Med
Genet 40, 348–351.
10 Wozniak MA, Shipley SJ, Dobson CB, Parker SP, Scott
FT, Leedham-Green M, Breuer J & Itzhaki RF (2007)
Does apolipoprotein E determine outcome of infection
by varicella zoster virus and by Epstein Barr virus? Eur
J Hum Genet 15, 672–678.
11 Dobson CB & Itzhaki RF (1999) Herpes simplex virus
type 1 and Alzheimer’s disease. Neurobiol Aging 20,
457–465.
12 Dobson CB, Sales SD, Hoggard P, Wozniak MA &
Crutcher KA (2006) The receptor-binding region of
human apolipoprotein E has direct anti-infective activ-
ity. J Infect Dis 193, 442–450.
13 Clay MA, Anantharamaiah GM, Mistry MJ,

Balasubramaniam A & Harmony JA (1995) Locali-
zation of a domain in apolipoprotein E with both
cytostatic and cytotoxic activity. Biochemistry 34,
11142–11151.
14 Lin PF, Blair W, Wang T, Spicer T, Guo Q, Zhou N,
Gong YF, Wang HG, Rose R, Yamanaka G, Robinson
B, Li CB, Fridell R, Deminie C, Demers G, Yang Z,
Zadjura L, Meanwell N & Colonno R (2003) A small
molecule HIV-1 inhibitor that targets the HIV-1 enve-
lope and inhibits CD4 receptor binding. Proc Natl Acad
Sci USA 100 , 11013–11018.
15 VanCompernolle SE, Taylor RJ, Oswald-Richter K,
Jiang J, Youree BE, Bowie JH, Tyler MJ, Conlon JM,
Wade D, Aiken C, Dermody TS, KewaI, Ramani VN,
Rollins-Smith LA & Unutmaz D (2005) Antimicrobial
peptides from amphibian skin potently inhibit human
immunodeficiency virus infection and transfer of virus
from dendritic cells to T cells. J Virol 79 , 11598–11606.
16 Bobardt MD, Saphire AC & Hung HC, Yu X, Van der
Schueren B, Zhang Z, David G & Gallay PA (2003)
Syndecan captures, protects, and transmits HIV to
T lymphocytes. Immunity 18, 27–39.
17 Saphire AC, Bobardt MD, Zhang Z, David G & Gallay
PA (2001) Syndecans serve as attachment receptors for
human immunodeficiency virus type 1 on macrophages.
J Virol 75, 9187–9200.
18 Bobardt MD, Salmon P, Wang L, Esko JD, Gabuzda D,
Fiala M, Trono D, Van der Schueren B, David G &
Gallay PA (2004) Contribution of proteoglycans to
B. A. Kelly et al. Aromatic substitution of ApoE-derived AMPs

FEBS Journal 274 (2007) 4511–4525 ª 2007 The Authors Journal compilation ª 2007 FEBS 4523
human immunodeficiency virus type 1 brain invasion.
J Virol 78, 6567–6584.
19 Pinzon-Ortiz C, Friedman J, Esko J & Sinnis P (2001)
The binding of the circumsporozoite protein to cell sur-
face heparan sulfate proteoglycans is required for plas-
modium sporozoite attachment to target cells. J Biol
Chem 276, 26784–26791.
20 Sinnis P, Willnow TE, Briones MR, Herz J &
Nussenzweig V (1996) Remnant lipoproteins inhibit
malaria sporozoite invasion of hepatocytes. J Exp Med
184, 945–954.
21 Andersson E, Rydengard V, Sonesson A, Morgelin M,
Bjorck L & Schmidtchen A (2004) Antimicrobial activi-
ties of heparin-binding peptides. Eur J Biochem 271,
1219–1226.
22 Tolar M, Keller JN, Chan S, Mattson MP, Marques
MA & Crutcher KA (1999) Truncated apolipoprotein
E (ApoE) causes increased intracellular calcium and
may mediate ApoE neurotoxicity. J Neurosci 19,
7100–7110.
23 Lilley HN, Narayanaswami V & Crutcher KA (2006)
Both Full-Length ApoE3 and ApoE4 Protect Against
ApoE Peptide Neurotoxicity. Program no. 466.9. Society
for Neuroscience, Atlanta, GA.
24 Park CB, Yi KS, Matsuzaki K, Kim MS & Kim SC
(2000) Structure-activity analysis of buforin II, a histone
H2A-derived antimicrobial peptide: the proline hinge is
responsible for the cell-penetrating ability of buforin II.
Proc Natl Acad Sci USA 97, 8245–8250.

25 Park IY, Cho JH, Kim KS, Kim YB, Kim MS &
Kim SC (2004) Helix stability confers salt resistance
upon helical antimicrobial peptides. J Biol Chem 279,
13896–13901.
26 Sun X, Yau VK, Briggs BJ & Whittaker GR (2005) Role
of clathrin-mediated endocytosis during vesicular stoma-
titis virus entry into host cells. Virology 338, 53–60.
27 Marsh M & Helenius A (1989) Virus entry into animal
cells. Adv Virus Res 36, 107–151.
28 Scherer PG & Seelig J (1989) Electric charge effects on
phospholipid headgroups. Phosphatidylcholine in mix-
tures with cationic and anionic amphiphiles. Biochemis-
try 28, 7720–7728.
29 Madine J, Doig AJ & Middleton DA (2006) A study of
the regional effects of alpha-synuclein on the organiza-
tion and stability of phospholipid bilayers. Biochemistry
45, 5783–5792.
30 Strom MB, Rekdal O & Svendsen JS (2002) Antimicro-
bial activity of short arginine- and tryptophan-rich pep-
tides. J Pept Sci 8, 431–437.
31 Giacometti A, Cirioni O, Barchiesi F, Fortuna M &
Scalise G (1999) In-vitro activity of cationic peptides
alone and in combination with clinically used antimicro-
bial agents against Pseudomonas aeruginosa. J Antimic-
rob Chemother 44 , 641–645.
32 Giacometti A, Cirioni O, Greganti G, Quarta M &
Scalise G (1998) In vitro activities of membrane-active
peptides against gram-positive and gram-negative aero-
bic bacteria. Antimicrob Agents Chemother 42, 3320–
3324.

33 van ‘T, Hof W, Veerman EC, Helmerhorst EJ & Amer-
ongen AV (2001) Antimicrobial peptides: properties and
applicability. Biol Chem 382, 597–619.
34 Shai Y (1999) Mechanism of the binding, insertion and
destabilization of phospholipid bilayer membranes by
alpha-helical antimicrobial and cell non-selective mem-
brane-lytic peptides. Biochim Biophys Acta 1462, 55–70.
35 Boheim G (1974) Statistical analysis of alamethicin
channels in black lipid membranes. J Membr Biol 19,
277–303.
36 Sato H & Feix JB (2006) Peptide–membrane interac-
tions and mechanisms of membrane destruction by
amphipathic alpha-helical antimicrobial peptides.
Biochim Biophys Acta 1758, 1245–1256.
37 Hancock RE & Chapple DS (1999) Peptide antibiotics.
Antimicrob Agents Chemother 43, 1317–1323.
38 Kang JH, Lee MK, Kim KL & Hahm KS (1996) Struc-
ture-biological activity relationships of 11-residue highly
basic peptide segment of bovine lactoferrin. Int J Pept
Protein Res 48, 357–363.
39 Masuda M, Nakashima H, Ueda T, Naba H, Ikoma R,
Otaka A, Terakawa Y, Tamamura H, Ibuka T, Mura-
kami T et al. (1992) A novel anti-HIV synthetic peptide,
T-22 ([Tyr5,12,Lys7]-polyphemusin II). Biochem Biophys
Res Commun 189, 845–850.
40 Blondelle SE & Houghten RA (1996) Novel antimicro-
bial compounds identified using synthetic combinatorial
library technology. Trends Biotechnol 14, 60–65.
41 Lamaziere A, Burlina F, Wolf C, Chassaing G, Trugnan
G & Ayala-Sanmartin J (2007) Non-metabolic mem-

brane tubulation and permeability induced by bioactive
peptides. PLoS ONE 2, E201.
42 Robinson W, McDougall B, Tran D & Selsted M
(1998) Anti-HIV-1 activity of indolicidin, an antimi-
crobial peptide from neutrophils. J Leukoc Biol 63,
94–100.
43 Simmons G, Reeves JD, McKnight A, Dejucq N,
Hibbitts S, Power CA, Aarons E, Schols D, De Clercq
E, Proudfoot AE & Clapham PR (1998) CXCR4 as a
functional coreceptor for human immunodeficiency
virus type 1 infection of primary macrophages. J Virol
72, 8453–8457.
44 del Mar Pujades Rodriguez M, Obasi A, Mosha F,
Todd J, Brown D, Changalucha J, Mabey D, Ross D,
Grosskurth H & Hayes R (2002) Herpes simplex virus
type 2 infection increases HIV incidence: a prospective
study in rural Tanzania. AIDS 16, 451–462.
45 Jenssen H, Andersen JH, Mantzilas D & Gutteberg TJ
(2004) A wide range of medium-sized, highly cationic,
Aromatic substitution of ApoE-derived AMPs B. A. Kelly et al.
4524 FEBS Journal 274 (2007) 4511–4525 ª 2007 The Authors Journal compilation ª 2007 FEBS
alpha-helical peptides show antiviral activity against
herpes simplex virus. Antiviral Res 64, 119–126.
46 Soda Y, Shimizu N, Jinno A, Liu HY, Kanbe K,
Kitamura T & Hoshino H (1999) Establishment of a
new system for determination of coreceptor usages of
HIV based on the human glioma NP-2 cell line.
Biochem Biophys Res Commun 258, 313–321.
47 Neil SJ, Aasa-Chapman MM, Clapham PR, Nibbs RJ,
McKnight A & Weiss RA (2005) The promiscuous CC

chemokine receptor D6 is a functional coreceptor for
primary isolates of human immunodeficiency virus type
1 (HIV-1) and HIV-2 on astrocytes. J Virol 79, 9618–
9624.
48 Neil S, Martin F, Ikeda Y & Collins M (2001) Postentry
restriction to human immunodeficiency virus-based
vector transduction in human monocytes. J Virol 75,
5448–5456.
49 Tsuji M, Mattei D, Nussenzweig RS, Eichinger D &
Zavala F (1994) Demonstration of heat-shock protein
70 in the sporozoite stage of malaria parasites. Parasitol
Res 80, 16–21.
B. A. Kelly et al. Aromatic substitution of ApoE-derived AMPs
FEBS Journal 274 (2007) 4511–4525 ª 2007 The Authors Journal compilation ª 2007 FEBS 4525