A virosome-mimotope approach to synthetic vaccine design and optimization: Synthesis, conformation, and immune recognition of a potential malaria-vaccine candidate

Article abstract of DOI:10.1002/anie.200250348

An optimized cyclic peptidomimetic (see picture) of the NPNA-repeat region in the circumsporozoite (CS) protein of P. falciparum was designed. When delivered to the immune system on human-compatible viruslike particles (virosomes), the mimetic effectively

Full text of DOI:10.1002/anie.200250348

Angewandte
Chemie
Communications
Making the influenza virus look more like the malaria parasite could
have important implications for synthetic-vaccine design. New syn-
thetic mimotopes displayed on virosomes elicit a strong antibody
response against the malaria parasite P. falciparum. For more
information see the following Communication by Pluscke, Robinson,
and co-workers.
Angew. Chem. Int. Ed. 2003, 42, 2367
DOI: 10.1002/anie.200250348
ꢀ 2003 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
2367

Communications
Peptidomimetic Vaccine Design
AVirosome-Mimotope Approach to Synthetic
Vaccine Design and Optimization: Synthesis,
Conformation, and Immune Recognition of a
PotentialMaal ria-Vaccine Candidate
Bernhard Pfeiffer, Elisabetta Peduzzi, Kerstin Moehle,
Rinaldo Zurbriggen, Reinhard Glück, Gerd Pluschke,*
and John A.Robinson*
Attempts to produce a vaccine against the malaria parasite
Plasmodium falciparum have so far met with limited success.
The molecules used to stimulate a protective antibody
response have in general shown only poor efficacy in
humans.[1] Synthetic peptides are attractive targets for vaccine
design, as they are readily synthesized and are able to target
specific antigens, but their use has suffered a number of
drawbacks.[2] These include poor immunogenicity and poor
mimicry of conformational epitopes, poor proteolytic stabil-
ity, and the lack of a suitable delivery system in humans. To
enhance immunogenicity, short peptides are usually coupled
to a carrier protein and then injected together with an
immunostimulatory adjuvant, which for humans is typically
an alum-absorbed antigen.
We introduce herein an alternative approach to synthetic-
vaccine design and optimization (see Figure 1), which
involves the use of peptidomimetics delivered to the
immune system on the surface of immunopotentiating
reconstituted influenza virosomes (IRIVs).[3] The IRIVs are
similar to liposomes, but contain influenza-derived, mem-
brane-bound hemagglutinin and neuraminidase, which impart
fusogenic activity and facilitate antigen delivery to immuno-
competent cells. The potential advantages of IRIVs for
vaccine delivery include their fusogenic activity, the surface
display of molecularly defined antigens, and compatibility
with both animals (e.g. mice) and humans.[4] Sequential
rounds of mimetic optimization based on structure–activity
relationships may ideally lead to vaccine candidates, which
after appropriate preclinical profiling, can be tested in human
Figure 1. Using virosome technology, sequential cycles of design, test-
ing and optimization of peptidomimetics in animal models can be
followed by human clinical trials with the same molecularly defined
adjuvant/delivery system.
clinical trials.[5] It is expected that the extensive clinical
experience obtained with two IRIV-based vaccines already
licensed[5] will facilitate the development of new IRIV-
mimotope vaccine candidates. We illustrate this approach
herein with a model system comprising the NPNA-repeat
epitope from Plasmodium falciparum.
One of the early synthetic malaria-vaccine candidates to
be tested in humans was a linear (NANP)3 peptide conjugated
to tetanus toxin in alum.[6] This peptide mimics epitopes in the
central repeat region (NPNA)ꢀ37 of the major surface protein
on sporozoites, the circumsporozoite (CS) protein.[7,8] How-
ever, the efficacy of this construct in humans was poor.
Building on these earlier studies, we describe below an
improved NPNA mimetic incorporated into IRIVs and its
potential as a new malaria-vaccine candidate.
A key step in the design of the new mimetic evolved from
earlier studies[9] of 1 (Figure 2) as well as of linear (NPNA)n
[*] Prof. J. A. Robinson, Dr. B. Pfeiffer, Dr. K. Moehle
Institute of Organic Chemistry, University of Zurich
8057 Zurich, Winterthurerstrasse 190, 8057 Zürich (Switzerland)
Fax : (+41)1-635-6833
E-mail: robinson@oci.unizh.ch
Prof. G. Pluschke, E. Peduzzi
Swiss Tropical Institute
Socinstrasse 57, 4002 Basel (Switzerland)
Fax: (41)1-61-271-8654
E-mail: Gerd.Pluschke@unibas.ch
Dr. R. Zurbriggen
Pevion-Biotech
Rehhagstrasse 79, 3018 Bern (Switzerland)
Figure 2. Mimetic 1 and a computer model of Ac-(NPNA)5-NH2 with a
backbone conformation in which each NPNA motif adopts a helical
b turn (see text). The arrows indicate the position of the proposed
amide cross-link between (2S,3R)-3-aminoproline (Apro, residue 6) and
glutamic acid (Glu, residue 16). Color coding: Asn=pink, Pro=cyan,
Asn=coral, Ala=yellow. The Ca atoms are each marked along the
ribbon with a ball.
Dr. R. Glück
Berna-Biotech
Rehhagstrasse 79, 3018 Bern (Switzerland)
Supporting information for this article is available on the WWW
under http://www.angewandte.org or from the author.
2368
ꢀ 2003 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
DOI: 10.1002/anie.200250348
Angew. Chem. Int. Ed. 2003, 42, 2368 – 2371

Angewandte
Chemie
peptides.[10] Modeling and NMR spectroscopic studies sug-
gested[9] that the central NPNA motif in 1 may adopt
conformations in which the first Asn residue is in the b region
of f/y space (e.g. À120/ + 1208) and the next three residues
Pro–Asn–Ala are each in the helical region (e.g. À60/À408).
Based on these f/y angles, a computer model was built for the
linear sequence Ac-(NPNA)5-NH2, such that this helical turn
is tandemly repeated. The resulting model (which differs from
previously reported theoretical models[11–13]) was stable in
molecular-dynamics simulations (data not given), and dis-
plays a novel repetitious secondary structure (see Figure 2).
Examination of the model suggested that this folded con-
formation could be stabilized by cross-linking an amino group
at the b position of Pro6 to a spatially adjacent side-chain
carboxy group of Glu, to be used in turn as a replacement for
Ala16. This would require cyclic peptidomimetics 3–5
(Scheme 2), containing (2S,3R)-3-aminoproline (Apro)[16] at
position 6 and glutamate at position 16 (Figure 2), linked by
an amide bond. The required orthogonally protected (2S,3R)-
3-aminoproline was prepared as shown in Scheme 1
Scheme 2. Synthesis of mimetics. a) TFA, iPr3SiH, H2O (95:2.5:2.5);
b) HATU, HOAt, DMF, iPr2EtN; c) piperidine, DMF (20% v/v); d) suc-
cinyl-rac-1-palmitoyl-3-oleoyl-glycero-2-phosphoethanolamine, HATU,
HOAt, iPr2EtN; e) Ac2O, pyridine/CH2Cl2. HATU=2-(1H-7-azabenzo-
triazol-1-yl)-1,1,3,3-tetramethyluronium hexafluorophosphate;
HOAt=1-hydroxy-7-azabenzotriazole; Mtt=4-methyltrityl; TFA=tri-
fluoroacetic acid.
NOE interactions between Pro Ca-H and Ala N-H (i + 2 ) in
each motif, provide evidence for helical turns formed by the
residues Asn5–Asn9, Asn9–Asn13, and Asn13–Asn17. The
same pattern of NOE interactions was also found for 6
(Scheme 2). Signal overlap of side-chain resonances seen in
the NMR spectrum of 4 was less severe for 6. Mimetics 4 and 6
showed similar coupling patterns of 3J (Ca-H, N-H) and
amide-proton temperature coefficients (see Supporting Infor-
mation) in their common substructures. However, the amide
temperature coefficients were mostly more negative than
À4 ppb/K and all the amide protons in 4 and 6 exchanged
within a few minutes at pH 4 in D2O, which suggests the
absence of stable intramolecular hydrogen bonds involving
the amide NHs.
Average solution structures were calculated for 4 and 6 by
simulated annealing based on NOE-derived distance
restraints, and gave similar results, except for the very flexible
N- and C-termini in 4. For example, the average structures for
6 could be clustered into four main families, as depicted in
Figure 3 (see Supporting Information). The lack of conver-
gence to a single structure reflects a low number of long range
NOE restraints, which we presume is a result of significant
backbone flexibility, compounded by problems of signal
overlap. However, in some of the calculated structures the
anticipated helical turns are populated (see Figure 4).
For immunological studies, the mimetic 5 was incorpo-
rated into IRIVs and then used to immunize BALB/c mice.
After preimmunization with the IRIV-based influenza vac-
cine Inflexal Berna[17] (Berna Products), and three doses of 5-
IRIV, the sera of all immunized mice contained mimetic-
specific antibodies, as demonstrated by ELISA (enzyme-
Scheme 1. Reagents and conditions: a) (CF3SO2)2O, CH2Cl2, pyridine,
98%; b) NaBH4, THF/DMF, 58%; c) K2S2O8, Na2HPO4, MeCN/H2O,
80%; d) (Boc)2O, CH2Cl2, DMF, Et3N, 76%; e) LiOH, THF, H2O, 99%;
f) Pd/C, MeOH, H2, 93%; g) Fmoc-OSucc, iPr2NEt, CH2Cl2, 68%.
Apro=(2S,3R)-3-aminoproline; Boc=tert-butyloxycarbonyl;
DMB=3,4-dimethoxybenzyl; DMF=N,N-dimethylformamide;
Fmoc=9-fluorenylmethoxycarbonyl; Succ=succinimide.
The mimetic 4 and an analogue 5, which was linked to a
phospholipid for incorporation into IRIVs, were synthesized
as shown in Scheme 2. The linear sequence was built through
solid-phase Fmoc chemistry on Rink-amide MBHA resin
(Novabiochem). After cleavage from the resin, the 3-amino
group of Apro6 and the carboxy group of Glu16 could be
coupled in high yield (> 90%) in DMF in the presence of
HATU.
The conformation of 4 was studied by NMR spectroscopy
1
in H2O/D2O (9:1) at pH 5. H NMR spectra indicated the
existence of a major conformer and two minor conformers
(ratio 80:14:6), of which the latter two most likely arise from
cis–trans isomerism at Asn–Pro peptide bonds, in analogy to
earlier studies.[10,14,15]
Only the major form of 4 was considered further.
2D NOESY plots showed strong dNN(i,i + 1) connectivities
between the peptide NH groups of Asn7 and Ala8, and Ala8
and Asn9 in one NPNAN motif, Asn11 and Ala12, and
Ala12and Asn13 in the next, and between Asn15 and Glu16,
and Glu16 and Asn17 in the last motif. These, together with
Angew. Chem. Int. Ed. 2003, 42, 2368 – 2371
www.angewandte.org
ꢀ 2003 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
2369

Communications
Figure 3. Average solution structures for 6 (Scheme 2). The calculated structures cluster into four main groups as shown. The sausage-shaped
average backbone was calculated from the average displacement of the Ca atoms, which is represented by the spline radius. The first helical
NPNA turn is in blue, the second in green, and the third in orange. The Apro6–Glu16 cross-link is at the bottom of the structure. The figure was
prepared with MOLMOL.[20]
humoral immune response against antigens delivered by
virosomes.[5,18] Primed mice presumably reflect the situation
in humans better than nonprimed mice, as most humans are
already exposed to the influenza virus early in life. Currently
it is not clear whether influenza antigen-specific helper T cells
or antibodies are primarily responsible for the effects of
influenza preimmunity.
The cross-reactivity of the antisera with peptidomimetic 1
was also analyzed by ELISA (data not shown). The sera from
three of four mice immunized with 5-IRIV cross-reacted with
mimetic 1. Interestingly, a helical NPNA turn (see above) is
Figure 4. Superimposition of the computer model of Ac-(NPNA)5-NH2
also possible in 1.[9] That a significant part of the antibody
shown in Figure 2 and average solution structures for 6 shown in
Figure 3.
response to 5-IRIV cross-reacts with 1 also shows that these
cross-reacting antibodies must not simply recognize the ends
of the peptide backbone in 5.
linked immunosorbent assay) (Figure 5a). The binding of
anti-5-IRIV antisera to native circumsporozoite (CS) protein
was analyzed by an indirect immunofluorescence assay using
P.falciparum sporozoites. In all immunized animals a
significant antisporozoite antibody response was detected.
The specificity of the cross-reaction was demonstrated by a
competition experiment (Figure 5b). Incubation of the anti-
serum with the sporozoites in the presence of 4 completely
suppressed immunostaining. The IRIV formulations thus
elicited a significant proportion of parasite-binding antibodies
among the total antimimetic immune response. Priming of
mice with influenza antigens has been found to enhance the
Monoclonal antibodies (mAbs) were prepared from
spleen cells of mice immunized with 5-IRIV, and two mAbs
(EP3 and EP9) that bound both the mimetic 5-IRIV and P.
falciparum sporozoites were isolated. Both EP3 and EP9 have
recently also been shown to inhibit the invasion of human-
liver hepatocytes by live sporozoites in the midnanomolar
range.[19] This inhibitory activity most likely arises as a result
of binding by the mAbs of neutralizing epitopes on the
parasite surface.
The cross-reactivity of EP3 and EP9 with peptidomimetic
1 and the linear peptide (NANP)ꢀ50 were also tested by
ELISA. The results show that both EP3 and EP9 bind to 1,
2370
ꢀ 2003 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
Angew. Chem. Int. Ed. 2003, 42, 2368 – 2371

Angewandte
Chemie
antigens and developmental stages of P.falciparum , or indeed
against other infectious agents.
Received: October 11, 2002
Revised: January 31, 2003 [Z50348]
Keywords: antibodies · conformation analysis ·
.
immunochemistry · peptides · peptidomimetics
[1] M. Tsuji, E. G. Rodrigues, R. S. Nussenzweig, Biol Chem. 2001,
382, 553.
[2] Ciba Foundation Symposium 119: Synthetic Peptides as Antigens
(Eds.: R. Porter, J. Whelan), Wiley, Chichester, UK, 1986.
[3] R. Glück, Vaccine 1999, 17, 1782.
[4] I. P. Hunziker, R. Zurbriggen, R. Glueck, O. B. Engler, J.
Reichen, W. J. Dai, W. J. Pichler, A. Cerny, Mol.Immunol.
2001, 38, 475.
[5] R. Zurbriggen, R. Glück, Vaccine 1999, 17, 1301.
[6] D. A. Herrington, D. F. Clyde, G. Losonsky, M. Cortesia, J. R.
Murphy, J. Davis, S. Baqar, A. M. Felix, E. P. Heimer, D.
Gillessen, E. Nardin, R. S. Nussenzweig, V. Nussenzweig, M. R.
Hollingdale, M. M. Levine, Nature 1987, 328, 257; for other
examples, see reference [1].
[7] F. Zavala, J. P. Tam, M. R. Hollingdale, A. H. Cochrane, I.
Quakyi, R. S. Nussenzweig, V. Nussenzweig, Science 1985, 228,
1436.
[8] J. B. Dame, J. L. Williams, T. F. McCutchan, J. L. Weber, R. A.
Wirtz, W. T. Hockmeyer, W. L. Maloy, J. D. Haynes, I. Schneider,
D. Roberts, G. S. Sanders, E. P. Reddy, C. L. Diggs, L. H. Miller,
Science 1984, 225, 593.
[9] R. Moreno, L. Jiang, K. Moehle, R. Zurbriggen, R. Glück, J. A.
Robinson, G. Pluschke, ChemBioChem 2001, 2, 838.
[10] H. J. Dyson, A. C. Satterthwait, R. A. Lerner, P. E. Wright,
Biochemistry 1990, 29, 7828.
[11] K. D. Gibson, H. A. Scheraga, Proc.Natl.Acad.Sci.USA 1986,
83, 5649.
Figure 5. a) Serum IgG titers in BALB/c mice immunized three times
with 5-IRIV. ELISA was performed in microtiter plates coated with 5
and incubated with serial dilutions of the sera of individual mice.
Bound IgG was detected using alkaline phosphatase-conjugated anti-
bodies specific for mouse gamma heavy chains. No antibody binding
was observed with uncoated plates and preimmune sera exhibited no
significant antimimotope antibody titer in ELISA (data not shown).
b) Immunofluorescence staining of P. falciparum sporozoites by
mouse anti-5-IRIV antiserum, using a FITC-labeled secondary anti-
mouse IgG antibody. No significant staining was observed with preim-
mune sera and incubation of the sporozoites in the presence of 4 sup-
pressed immunostaining (data not shown). FITC=(fluorescein isothio-
cyanate).
[12] I. K. Roterman, K. D. Gibson, H. A. Scheraga, J.Biomol.Struct.
Dyn. 1989, 7, 391.
[13] B. R. Brooks, R. W. Pastor, F. W. Carson, Proc.Natl.Acad.Sci.
USA 1987, 84, 4470.
[14] C. Bisang, C. Weber, J. Inglis, C. A. Schiffer, W. F. van Gun-
steren, I. Jelesarov, H. R. Bosshard, J. A. Robinson, J.Am.
Chem.Soc. 1995, 117, 7904.
whereas (NANP)ꢀ50 is only bound by EP9 and not by EP3 in
ELISA. These data suggest that EP3 recognizes epitopes in
both 5-IRIV and on the parasite surface that are more poorly
represented, if at all, in the linear peptide (NANP)ꢀ50
.
[15] C. Bisang, L. Jiang, E. Freund, F. Emery, C. Bauch, H. Matile, G.
Pluschke, J. A. Robinson, J.Am.Chem.Soc.
1998, 120, 7439.
The results show that the IRIV-bound peptidomimetic 5
effectively elicits sporozoite-binding antibodies, and also lead
to a new proposal for the folded conformation of the NPNA-
repeat region in the CS protein (Figure 2). A key issue now is
whether the use of this peptidomimetic with the IRIV
delivery system leads to a stronger antiparasite immune
response in humans than in mice, as has been observed with
other virosome-based vaccine candidates. In this case, influ-
enza preimmunity may play a role, as suggested by the
enhancing effects of influenza priming in mice. The suitability
of virosomal mimetic formulations for use in humans will
have to be evaluated in a phase I clinical safety and
immunogenicity trial. At a later stage, efficacy trials will
reveal whether the IRIV-based vaccine candidate offers
improvements in protective efficacy over linear (NANP)3
peptides delivered by the classical route.[6] The virotope
approach, however, could also be applied to the design and
testing of new combination vaccines targeted against multiple
[16] For the starting bicyclic lactam, see: P. Angehrn, R. L. Charnas,
K. Gubernator, E.-M. Gutknecht, C. Hubschwerlen, M. Kania,
C. Oefner, M. G. P. Page, S. Sogabe, J.-L. Specklin, F. Winkler, J.
Med.Chem. 1998, 41, 3961 – 3971.
[17] R. Glück, R. Mischler, B. Finkel, J. U. Que, B. Scarpa, S. J. Cryz,
Lancet 1994, 344, 160.
[18] F. Poltl-Frank, R. Zurbriggen, A. Helg, F. Stuart, J. A. Robinson,
R. Glück, G. Pluschke, Clin.Exp.Immunol. 1999, 117, 496 – 503.
[19] Full details of the immunological data will be published else-
where.
[20] R. Koradi, M. Billeter, K. Wüthrich, J.Mol.Graphics 1996, 14,
51 – 55.
Angew. Chem. Int. Ed. 2003, 42, 2368 – 2371
www.angewandte.org
ꢀ 2003 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
2371