Synthesis, conformational properties, and immunogenicity of a cyclic template-bound peptide mimetic containing an NPNA motif from the circumsporozoite protein of plasmodium falciparum

Article abstract of DOI:10.1021/ja980444j

The immunodominant central portion of the circumsporozoite (CS) surface protein of the malaria parasite Plasmodium falciparum contains a tetrapeptide motif, Asn-Pro-Asn-Ala (NPNA), tandemly repeated almost 40 times. The three-dimensional structure of the CS protein, including the central repeat region, is presently unknown. We have investigated an approach to stabilize β-turns in a single NPNA motif, by its incorporation into a template-bound cyclic peptide comprising the sequence ANPNAA. The template was designed to stabilize β-turns in the peptide loop and to allow its conjugation to T-cell epitopes in a multiple-antigen-peptide. NMR studies and MD simulations with time-averaged NOE-derived upper distance restraints support the formation of a stable β-I turn conformation in the NPNA motif of this template-bound antigen. Balb/c mice immunized with a multiple-antigen-peptide containing four copies of the template-bound loop conjugated to a single universal T-cell epitope produced antibodies that bound P. falciparum sporozoites in immunofluorescence assays. These results provide further support for the immunological relevance of a type-I β-turn conformation based on the NPNA cadence in the repeat region of the CS protein and illustrate the use of a novel template for the evaluation of conformationally constrained peptide immunogens.

Full text of DOI:10.1021/ja980444j

J. Am. Chem. Soc. 1998, 120, 7439-7449
7439
Synthesis, Conformational Properties, and Immunogenicity of a
Cyclic Template-Bound Peptide Mimetic Containing an NPNA Motif
from the Circumsporozoite Protein of Plasmodium falciparum
Christian Bisang,‡ Luyong Jiang,‡ Ernst Freund,‡ Fabienne Emery,‡ Christian Bauch,§
Hugues Matile,§ Gerd Pluschke,§ and John A. Robinson*,‡
Contribution from the Institute of Organic Chemistry, UniVersity of Zu¨rich, Winterthurerstrasse 190,
8057 Zu¨rich, and Swiss Tropical Institute, Socinstrasse 57, 4002 Basel, Switzerland
ReceiVed February 9, 1998
Abstract: The immunodominant central portion of the circumsporozoite (CS) surface protein of the malaria
parasite Plasmodium falciparum contains a tetrapeptide motif, Asn-Pro-Asn-Ala (NPNA), tandemly repeated
almost 40 times. The three-dimensional structure of the CS protein, including the central repeat region, is
presently unknown. We have investigated an approach to stabilize â-turns in a single NPNA motif, by its
incorporation into a template-bound cyclic peptide comprising the sequence ANPNAA. The template was
designed to stabilize â-turns in the peptide loop and to allow its conjugation to T-cell epitopes in a multiple-
antigen-peptide. NMR studies and MD simulations with time-averaged NOE-derived upper distance restraints
support the formation of a stable â-I turn conformation in the NPNA motif of this template-bound antigen.
Balb/c mice immunized with a multiple-antigen-peptide containing four copies of the template-bound loop
conjugated to a single universal T-cell epitope produced antibodies that bound P. falciparum sporozoites in
immunofluorescence assays. These results provide further support for the immunological relevance of a type-I
â-turn conformation based on the NPNA cadence in the repeat region of the CS protein and illustrate the use
of a novel template for the evaluation of conformationally constrained peptide immunogens.
Introduction
immunization with DNA.7 Nevertheless, the protective effect
in humans of vaccination with peptides and proteins containing
NPNA repeats was initially disappointing.3 Recently, however,
with the use of a more potent adjuvant, almost complete
resistance to malaria infection was achieved in a Phase I trial
using a recombinant protein vaccine comprising residues 207-
395 of the CS protein.8
Plasmodium falciparum is the most virulent of the Plasmodial
parasites that cause malaria. Infection of human hosts is initiated
by the sporozoite stage, which is injected into the circulation
following attack by infested mosquitoes. The sporozoites
migrate to the liver where a specific ligand-receptor interaction,
mediated by one of the major surface membrane proteins on
the parasite, the circumsporozoite (CS) protein, may lead to
invasion of liver cells.1 One approach to develop a malaria
vaccine is based on the observation that antibodies against the
CS protein are able to block infection by sporozoites.2,3
Although several B-cell epitopes have been described on the
ca. 45 kDa CS protein,4 the immunodominant response is
directed against a central repeat region (approximately residues
150-300) comprising a motif Asn-Pro-Asn-Ala (NPNA), which
is tandemly repeated almost 40 times. The role of the CS
protein in protective immunity has been demonstrated in rodent
models, using synthetic peptides,5 recombinant proteins,6 and
An important question concerns the conformation of the
NPNA-repeat region in the intact CS protein. Earlier NMR
studies of linear peptides containing tandemly repeated NANP
and NPNA tetrapeptide motifs indicated the presence of
conformers containing helical and/or reverse turns based on the
NPNA cadence, in rapid dynamic equilibrium with unfolded
forms.9 Based on sequence preferences for â-turns in protein
crystal structures,10 a type-I â-turn seems a likely conformation
for the NPNA motif in aqueous solution. A significant
stabilization of â-I conformations was achieved by substituting
proline by (S)-R-methylproline (PMe) in linear peptides contain-
ing single and tandemly repeated (NPMeNA) motifs.11 More-
over, anti-(NPMeNA)3 antibodies raised in rabbits were able to
bind to P. falciparum sporozoites in a solid-phase immuno-
* Corresponding author. Tel: (++)-41-1-635-4242. Fax: (++)-41-1-
635-6812. E-mail: robinson@oci.unizh.ch.
‡ University of Zurich.
§ Swiss Tropical Institute.
(7) Sedegah, M.; Hedstrom, R.; Hobart, P.; Hoffman, S. L. Proc. Natl.
Acad. Sci. U.S.A. 1994, 91, 9866-9870.
(8) Stoute, J. A.; Slaoui, M.; Heppner, D. G.; Momin, P.; Kester, K. E.;
Desmons, P.; Wellde, B. T.; Garcon, N.; Krzych, U.; Marchand, M.; Ballou,
W. R.; Cohen, J. D. New Engl. J. Med. 1997, 336, 86-91.
(9) Dyson, H. J.; Satterthwait, A. C.; Lerner, R. A.; Wright, P. E.
Biochemistry 1990, 29, 7828-7837.
(10) Hutchinson, E. G.; Thornton, J. M. Protein Sci. 1994, 3, 2207-
2216.
(11) Bisang, C.; Weber, C.; Inglis, J.; Schiffer, C. A.; van Gunsteren,
W. F.; Jelesarov, I.; Bosshard, H. R.; Robinson, J. A. J. Am. Chem. Soc.
1995, 117, 7904-7915.
(1) Cerami, C.; Frevert, U.; Sinnis, P.; Takacs, B.; Clavijo, P.; Santos,
M. J.; Nussenzweig, V. Cell 1992, 70, 1021.
(2) Nardin, E. H.; Nussenzweig, R. S. Annu. ReV. Immunol. 1993, 11,
687.
(3) Nussenzweig, V.; Nussenzweig, R. S. AdV. Immunol. 1989, 45, 283.
(4) Calvo-Calle, J. M.; Nardin, E. H.; Clavijo, P.; Boudin, C.; Stuber,
D.; Takacs, B.; Nussenzweig, R. S.; Cochrane, A. H. J. Immunol. 1992,
149, 2695-2701.
(5) Nardin, E. H.; Oliveira, G. A.; Calvo-Calle, J. M.; Nussenzweig, R.
S. AdV. Immunol. 1995, 60, 105-149.
(6) Romero, P. Curr. Opin. Immunol. 1992, 4, 432-441.
S0002-7863(98)00444-2 CCC: $15.00 © 1998 American Chemical Society
Published on Web 07/17/1998

7440 J. Am. Chem. Soc., Vol. 120, No. 30, 1998
Bisang et al.
fluorescence assay, thus providing evidence for the biological
relevance of the â-I conformation in the context of the folded
CS protein. More recently, molecular dynamics simulations
Scheme 1a
3
employing time-averaged NOE distance and J-coupling con-
stant restraints confirmed these structural conclusions and
suggested how a compact folded structure might arise in
(NPMeNA)3 peptides, by stacking of the proline rings in the outer
two motifs to form a hydrophobic core.12 Similarly folded forms
would also be possible in longer tandemly repeated NPNA
peptides, thereby forming a stemlike structure, which conceiv-
ably might also occur in the native CS protein.
In earlier work, Satterthwait et al.13 synthesized conforma-
tionally constrained variants of (NPNA)3 in which the side
chains of two Asn residues, flanking either Pro or Ala, were
joined by ethylene bridges and showed that the former could
elicit anti-sporozoite antibodies in rabbits. This was the first
demonstration that nonproteinogenic conformational constraints
could be used to prepare effective immunogens, although the
conformations stabilized by these modifications were not
described.
In this work, we have investigated an alternative approach
to stabilize a â-I turn in the NPNA motif within a template-
bound cyclic peptide comprising the sequence ANPNAA. The
template was designed to stabilize â-turns in the peptide loop
and to allow its presentation to the immune system as a
conformationally constrained peptide immunogen conjugated
to T-cell epitopes in a multiple-antigen-peptide format.14 An
optimized route to the template 1 and peptide antigen 2 is
a Reagents: a) SOCl2, MeOH, 93%; b) BnOCOCl, Na2CO3 aqueous
1 M/dioxane, 84%; c) TMSCl, NEt3, THF; d) LDA, BrCH2CO2 Bu,
THF; e) citric acid 10%, MeOH, 72% over three steps; f) Ph3P, PhthNH,
DEAD, THF, 74%; g) Pd/C, H2, 47%; h) Fmoc-Asp(OAllyl)-OH, DCC,
HOAt, CH2Cl2; i) DBU 5%, MeCN, 64% over two steps; j) MeNHNH2,
DMF; k) Fmoc-Succ, imidazole, DMAP, DCM, 58% over two steps;
l) DCM, TFA 1:2, 97%.
t
conditions where a linear peptide containing the same epitope
fails to elicit a detectable anti-sporozoite antibody response.
Results
Design and Synthesis. The template 1 can be readily
synthesized on a gram scale using the route shown in Scheme
1, which has been optimized from that described previously in
a communication.15 The alkylation of Me3Si-protected N-Z-4-
hydroxyproline methyl ester can be performed conveniently on
a large scale and after removal of the temporary Si-protecting
group gives the desired diastereomer in about 2:1 (3:3′) excess.
After a Mitsunobu reaction16 and removal of the Z protecting
group, the desired stereoisomer 4 could be obtained easily by a
simple recrystallization, without the need for large scale
chromatography. The relative configuration of 4 was confirmed
by steady-state NOE-difference experiments. The coupling of
4 with Fmoc-Asp(allyl)-OH proceeded in good yield using DCC
and HOAt for activation. The resulting dipeptide could then
be cyclized directly to give 5, and exchange of protecting groups
yielded the template 1 in a form suitable for solid-phase peptide
synthesis.
To assemble the cyclic antigen 2 on a solid-phase (Scheme
2), the template 1 was coupled to Tentagel S-AC resin through
the free carboxylic acid, and the linear peptide chain was then
assembled using standard Fmoc-chemistry.17 The allyl ester of
the template was cleaved using Pd0, and cyclization was
accomplished with BOP and DIEA in NMP-DMSO. The
products of the cyclization were analyzed by reverse-phase
HPLC following cleavage from the resin and deprotection with
TFA. Two major products were detected in approximately 2:1
ratio. Electrospray mass spectrometry indicated they were the
desired product (30% yield) (MS: 806.5 [(M + H)+]), and a
dimeric species (MS: 1612 [(M + H)+]) that was not
investigated further. Alternatively, cleavage from the resin with
presented,15 that allows the convenient preparation of 1 on a
gram scale. In addition, conformational studies on 2 are
described, based on NMR and MD simulations with time-
averaged distance restraints. Finally, we report that the multiple-
antigen-peptide 9, incorporating 2, is able to elicit antibodies
in mice that cross-react with P. falciparum sporozoites, under
(12) Nanzer, D.; Torda, A. E.; Bisang, C.; Weber, C.; Robinson, J. A.;
van Gunsteren, W. F. J. Mol. Biol. 1997, 267, 1011-1024.
(13) Satterthwait, A. C.; Arrhenius, T.; A., H. R.; Zavala, F.; Nussen-
zweig, V.; Lerner, R. A. Philos. Trans. R. Soc. B 1989, 323, 565-572.
(14) Tam, J. P. J. Immunol. Methods 1996, 196, 17-32.
(15) Emery, F.; Bisang, C.; Favre, M.; Jiang, L.; Robinson, J. A. J. Chem.
Soc., Chem. Commun. 1996, 2155-2156.
(16) Mitsunobu, O.; Wada, M.; Sano, T. J. Am. Chem. Soc. 1972, 94,
679-680.
(17) Atherton, E.; Sheppard, R. C. Solid-phase peptide synthesis - a
practical approach; IRL Press: Oxford, 1989.

Cyclic NPNA Antigen
J. Am. Chem. Soc., Vol. 120, No. 30, 1998 7441
1% TFA in CH2Cl2 gave 7 with the side chain protecting groups
intact. The peptide 8 was synthesized in a similar way, and
the disulfide bridge was introduced in the last step by aerial
oxidation of the corresponding dithiol.
Scheme 2a
a Reagents: a) 1 (2 equiv), Tentagel-SAC, CIP (6 equiv), Pyr/
CH2Cl2; b) piperidine, DMF; c) peptide assembly using HBTU (3
equiv), HOBt (3 equiv) in DMF, DIEA, and Fmoc-protected amino
acid (3 equiv) in each coupling step, piperidine in DMF for Fmoc-
removal; then d) Pd(PPh3)4 (3 equiv), NMM, AcOH, DMF; e)
piperidine, DMF; f) BOP (1.5 equiv), DIEA (3.0 equiv), NMP, DMSO;
g) 1% TFA/CH2Cl2 (≈30% yield over last four steps); h) 95% TFA.
For immunizations, the multiple antigen peptides 9 and 10
were prepared by straightforward solid-phase methods using
Fmoc-chemistry. The final steps in the assembly of 9 involved
coupling 2 to resin bound H4-Lys2-Lys-Ala.Ala.Gln(Mtt).
Tyr(tBu).Ile.Lys(Boc).Ala.Asn(Mtt).Ser(tBu).Lys(Boc).Phe.
Ile.Gly.Ile.Thr(tBu).Glu(OtBu).Leu.Ala-O-Resin, which was
performed with a 4-fold excess of 2 and HBTU/HOBt for
activation. Both 9 and 10 were purified by HPLC and gave
electrospray mass spectra consistent with their predicted masses.
NMR and Conformational Studies. The aim of these
studies was to determine which conformations are preferred in
the cyclic peptide antigen 2. The peptide 8 served here as a
useful reference. All studies were performed in aqueous
could be seen in ROESY spectra at 300 K as cross-peaks with
the same positive sign as the diagonal. The rates of intercon-
version of these conformers were calculated from the ratio of
the intensity of the diagonal (IAA) and exchange cross-peaks
(IAB) for the NH signals, which are related to the rate of
exchange (k) according to eq 119
1
solution. The chemical shifts and H NMR line widths of 2
IAA/IAB ) [(1/k) × (1/tm)] - 1
(1)
and 8 remain unchanged in H2O/D2O (90:10, pH 5) in the
concentration range 10.0-0.3 mM, indicating an absence of
aggregation. The chemical shift assignments were made by
standard methods18 and are given in Table 1 and 2.
where tm is the mixing time. From the gradient of a plot of
IAA/IAB versus 1/tm (ROESY mixing times were 75, 150, 225,
and 300 ms) the exchange rates were estimated to be k(major->minor)
) 0.011 s-1 and k (minor->major) ) 0.12 s-1. The exchange cross-
peaks are not apparent in ROESY spectra measured at 285 K,
indicating that the exchange rate is slower than the mixing period
at this temperature. Earlier 1H NMR spectra of linear peptides
Chemical Shifts and Chemical Exchange. The amide
1
region of the H NMR spectrum of 2 revealed the presence of
a minor conformer, representing ca. 10% of the major con-
former. Chemical exchange between the major and minor forms

7442 J. Am. Chem. Soc., Vol. 120, No. 30, 1998
Bisang et al.
Table 1. 1H NMR (600 MHz) Chemical Shifts for 2 at 300 K in
10% H2O/H2O, pH ) 5.0
suggests significant shielding from solvent. The relative
exchange rates of amide NH protons were measured by
monitoring the disappearance of resonances upon dissolution
in D2O at 293 K. For 2, the peptide NH protons for Asn4 and
Ala5 NHs exhibited half-lives of ca. 12 and 25 min, respectively,
at pH 3.4, whereas the others exchanged completely within a
few minutes. A similar pattern was noted earlier in a closely
related system.21 For 8, a different result was observed, with
Pro(NH2)7 NH-C(γ) now exchanging very slowly (t1/2 ) >24
h) at pH 4.0, followed by Asn2 NH (t1/2 ) 90 min) and Ala5
NH (t1/2 ) 66 min). The other NH protons again exchanged
completely within a few minutes, and their exchange rates
presumably approximate to random coil values under these
conditions.
2
chemical shifta (ppm)
b
residue
Ala1
NH H-CR
H-Câ
1.31
others
8.34 4.29
8.07c
Asn2
8.13 4.93
2.64, 3.15
NH(δ)E 7.46, NH(δ)Z 7.06
7.69c
Pro3
4.42
1.96, 2.32
CH2(γ) 1.97, 2.02; CH2(δ) 3.78
NH(δ)E 7.65, NH(δ)Z 6.98
Asn4
8.41 waterd 2.72, 2.88
8.53c
7.60 4.38
8.52c
Ala5
Ala6
1.28
1.33
8.21 4.30
7.97c
Pro(NH2)7 e 8.46
7.74c
1.96, 2.60b CH2(δ) 3.82, 3.87
CH2(ꢀ) 2.75, 2.66
3
Coupling Constants. The J (R,â) coupling constants for
the Asn2 and Asn4 residues in 2 (Table 4) have combinations
of high and low values, indicative of preferred conformations
for each side chain. Rotational averaging, if dominant, would
result in both values being close to 6.5 Hz, which is not
observed. In 8, however, a different preferred conformation is
Asp8
8.30 4.45
2.80, 3.26b
8.11c
a Chemical shifts are measured relative to internal TSP [sodium(tri-
methylsilyl)-d4-propionate]. Pro(NH2)7 and Asp8 are the 4-aminoproline
and aspartate moieties of the template. b Stereospecific assignments are
in italics, in the order pro-R, pro-S. c The peptide NH assignments for
the minor conformer (see text). d Resonance lies under the water signal.
e The Pro(NH2)7 NH refers to NH-C(γ). The Pro(NH2)7 H-C(γ) lies
under the water signal.
3
populated by the Asn2 side chain, since both J (R,â) values
are now small. Peptide 8 is expected to be more rigid than 2,
and the other coupling constants support this. For example,
several 3J (R,NH) values are > 9.0 Hz, and all other 3J coupling
constants have nonaveraged values. This includes 3J (R,â), 3J
(â,γ), and 3J (γ,δ) for the aminoproline and aspartate moieties
of the template. In contrast, the aminoproline moiety of the
Table 2. 1H NMR (600 MHz) Chemical Shifts for 8 at 300 K in
2
10% H2O/H2O, pH ) 5.0
chemical shifta
(ppm)
3
3
template in 2 shows J (â,γ) and J (γ,δ) values in the range
5-8 Hz, which for a proline residue indicate rapid fluctuations
in ring pucker.22 In both 2 and 8, the combination of two small
3J (R,â) values for the aspartate moiety of the template indicate
a preferred torsion angle ø1 around +60°.
b
residue
NH H-CR
H-Câ
others
Cys1
8.78 5.32
8.65 5.00
4.26
3.01, 2.94
Asn2
Pro3
3.18, 2.88 NH(δ)E 7.70, NH(δ)Z 6.95
2.42, 1.75 CH2(γ) 2.13, 1.96; CH2(δ)
3.86, 3.67
NOEs and Structure Calculations. A summary of the
NOEs observed in 2 are shown in Figure 1. Noteworthy are
the medium range (i, i+2) and (i, i+3) NOE connectivities
within the four residues of the NPNA motif as well as a
relatively strong Asn4-Ala5 dNN NOE, which together are
characteristic of a highly populated â-turn conformation with
proline at the (i+1) position. The low-temperature coefficient
of the Ala5 NH proton (Table 3) and its relatively slow H/D
exchange rate (see above) are also consistent with a â-turn in
the NPNA motif stabilized by H-bonding involving the Ala5
amide proton. Although a â-I turn appears to be favored (vide
infra), the observation of a medium strength Pro3-Asn4 dRN (i,
i+1) NOE indicates a shorter Pro3 H-C(R) to Asn4 NH distance
than seen in an ideal â-I turn, most likely due to torsional motion
involving this amide plane or occasional flipping to a type-II
â-turn.
Average solution structures were calculated for 2 based on
upper-distance and one dihedral angle restraint (Asp8 ø1 ) 60
( 30°) using a simulated annealing (SA) method.21 From 30
calculations, 27 structures were within 10 kcal/mol of the
minimum energy structure and showed no distance restraint
violations > 0.3 Å; for nine of these structures, the violations
were < 0.25 Å. The final 27 SA structures revealed structural
convergence to a family of closely related backbone conforma-
tions with a pairwise rms deviation for superimposition of the
backbone N, C(R), and C′ atoms in all residues of 0.43 ( 0.32
Å. The minimum energy structure is shown in Figure 2A. Only
one distance restraint, that derived from the Pro3-Asn4 dRN (i,
i+1) NOE, was consistently violated in the range 0.2-0.3 Å.
Asn4
Ala5
Cys6
8.52 4.75c 2.86, 2.79 NH(δ)E 7.58, NH(δ)Z 6.95
7.86 4.48
8.67 5.20
1.36
3.07, 2.89
Pro(NH2)7 d 8.84
2.35, 2.54b CH2(δ) 3.49, 4.14;b CH(γ)
4.52; CH2(ꢀ) 2.83, 2.71
2.97, 3.35b
Asp8
8.29 4.53
a Chemical shifts are measured relative to internal TSP [sodium(tri-
methylsilyl)-d4-propionate]. Pro(NH2)7 and Asp8 are the 4-aminoproline
and aspartate moieties of the template. b Stereospecific assignments are
in italics, in the order pro-R, pro-S. c Resonance lies under the water
signal. d The Pro(NH2)7 NH refers to NH-C(γ).
containing tandemly repeated NPNA motifs also revealed major
and minor conformers in about the same ratio,9 due to cis-
trans isomerism at one or more Asn-Pro peptide bonds, whereas
(NPMeNA)3 peptides showed a single average structure by
NMR.11 The 1H NMR spectrum of 8 also revealed only a single
species on the NMR chemical shift time scale at 300 K.
Since correlations between chemical shift and peptide sec-
ondary structure are currently of general interest, we note here
that the Ala5 NH in 2 and 8 resonates considerably upfield from
the random coil value at this temperature,20 the former by +0.88
ppm and the latter by +0.62 ppm. Moreover, in 8, the H-C(R)
resonances of Cys1, Cys6, and Asn2 are appreciably downfield
shifted compared to the random coil values, in the case of Cys1
by -0.74 ppm.
The temperature coefficients of the amide NH resonances for
2 and 8 are collected in Table 3. The low value for Ala5 NH
(18) Wu¨thrich, K. NMR of proteins and nucleic acids; Wiley-Inter-
science: New York, 1986.
(19) Ernst, R. R.; Bodenhausen, G.; Wokaun, A. Principles of NMR
spectroscopy in one and two dimensions; Clarendon Press: Oxford, 1987.
(20) Merutka, G.; Dyson, H. J.; Wright, P. E. J. Biomol. NMR 1995, 5,
14-24.
(21) Bisang, C.; Weber, C.; Robinson, J. A. HelV. Chim. Acta 1996, 79,
1825-1842.
(22) Ma´di, Z. L.; Griesinger, C.; Ernst, R. R. J. Am. Chem. Soc. 1990,
112, 2908-2914.

Cyclic NPNA Antigen
J. Am. Chem. Soc., Vol. 120, No. 30, 1998 7443
Table 3. Amide Temperature Coefficients for 2 and 8
Temperature Coefficientsa for Peptide 2
Residue
-∆δ/K [ppb/K]
Ala1
7.6
Asn2
6.8
NHE
3.8
NHZ
6.3
Asn4
4.5
NHE
6.5
NHZ
6.3
Ala5
1.2
Ala6
6.5
Pro(NH2)7
7.4
Asp8
8.4
Temperature Coefficientsa for Peptide 8
Residue
-∆δ/K [ppb/K]
Cys1
7.9
Asn2
2.1
NHE
5.3
NHZ
4.4
Asn4
4.6
NHE
6.4
NHZ
5.5
Ala5
2.8
Cys6
7.4
Pro(NH2)7
3.0
Asp8
9.0
a The coefficients for the peptide amides and the side chain NHs of Asn are given. Measurements were made in aqueous solution (10% D2O/
H2O, pH 5) in the range 5-41°.
Table 4. Coupling Constantsa (Hz) for 2 and 8
residue
3J (R,NH) 3J (R,â)
others
2
Ala1
5.8
7.3
7.0
Asn2
4.3, 9.9b
4.7, 8.8b
J (â,â′) 15.0
Pro3
n.d.
Asn4
8.8
6.7
6.2
8.5c
10.5, 4.5b J (â,â′) 15.1
Ala5
7.0
7.1
Ala6
Pro(NH2)7
J (âR,γ) 8.0, J (âS,γ) 7.5,
J (â,â′) 12.5, J (γ,δ) 5.7,b
J (γ,δ′) 8.9,b J (δ,δ′) 13.2,
J (ꢀ,ꢀ′) 15.1
Asp8
<3.0
4.7, 2.7d
J (â,â′) 16.9
8
Cys1
9.5
8.7
4.3, 10.4 J (â,â′) 15.2
Asn2
3.9, 5.4b
J (â,â′) 15.8
Pro3
7.0, 10.8b n.d.
10.1, 5.4b J (â,â′) 14.7
6.8
Asn4
9.1
6.2
9.8
9.8c
Ala5
Cys6
4.6, 10.6 J (â,â′) 15.1
Pro(NH2)7
J (âR,γ) < 2.0, J (âS,γ) 7.3,
J (â,â′) 15.0, J (γ,δR) 4.2,
J (γ,δS) < 2.0, J (δ,δ′) 12.8,
J (ꢀ,ꢀ′) 15.6
Asp8
<2.0
3.9, 3.1d
J (â,â′) 18.0
a Measured from 1D and/or E.COSY spectra. b Stereospecific as-
Figure 1. Summary of key NOE connectivities found for (A) peptide
2 and (B) peptide 8. The Asn4 H-C(R) lies under the water signal.
NOE connectivities with Pro3 are to the H-C(δ)’s in place of NH.
Pro(NH2)7 NH refers to the NH-C(γ) in the aminoproline moiety of
the template.
3
signments not available. c Refers to the J(NH-C(γ), H-C(γ)) cou-
pling. d Given in the order pro-R, pro-S.
However, all the SA structures have the NPNA motif in a type-I
â-turn, whereas as noted above, torsional motion of the amide
plane or flipping to a type-II turn would lead to a shorter Pro3
H-C(R) to Asn4 NH distance. In addition, all the SA structures
have the same ring pucker for the aminoproline moiety of the
template, with ø1 ≈ -35°, yet the 3J coupling constants within
this ring provide evidence for flipping of ring pucker.
with both the low-temperature coefficients (Table 2) and slow
H/D exchange behavior (vide supra) of these (Asn2, Ala5, and
Pro(NH2)7) NH protons. 3J (R,NH) values ≈ 9.0 Hz are also
consistent with a â-hairpin conformation. The nonaveraged 3J
(R,â) values indicate a relatively well defined conformation for
each of the Asn and Cys side chains and are consistent with
those found in the SA-structures (Figure 2B). The 3J (â,γ) and
3J (γ,δ) values for the aminoproline moiety of the template are
also consistent with the observed ø1 values of ≈+10-12°, which
places the N-C(γ) substituent in a pseudoaxial position. Taken
together, the NMR data suggest that to a reasonable approxima-
tion the peptide 8 may be viewed as a relatively rigid molecule,
undergoing mostly small amplitude motion about backbone and
side chain dihedral angles.
Although the NMR data indicate stable secondary structure
in the backbone of 2, this molecule is not rigid but rather retains
a significant degree of flexibility which may include flipping
of pyrrolidine rings and amide planes (vide supra). An attempt
to account for flexibility in the structure calculations for 2 was
made by incorporating the atom-atom upper distance restraints
as values that are to be fulfilled, not instantaneously, but over
the course of an MD simulation with solvent water present, i.e.,
as time-averaged distance restraints (TA-DR).
The pattern of NOE connectivities observed in peptide 8 (see
Figure 1), in particular the strong Cys1-Cys6 dRR (i, i+5) and
Asn2-Ala5 dNN (i, i+3) NOEs, provides evidence for a â-hairpin
conformation from Cys1-Cys6, and this is borne out by structure
calculations. Of 30 structures generated by SA using NOE-
derived distance restraints, 25 were within 10 kcal/mol of the
global minimum and showed no NOE violations >0.2 Å.
Moreover, these structures converged to a family of closely
related conformers having a â-hairpin backbone conformation
(with a pairwise rms deviation for superimposition of the
backbone N, C(R), and C′ atoms in all residues of 0.38 ( 0.20
Å) as seen in the energy minimized SA structure deduced for
8 shown in Figure 2B. The final 25 SA structures all possess
a type-I â-turn in the NPNA motif, and the side chain Asn2
O(δ) is frequently found H-bonded to the Ala5 backbone NH,
as was observed also for 2 (Figure 2A). The residues Cys1-
Pro3 and Asn4-Cys6 form two â-strands, connected at Pro3-Asn4
by the â-turn, with the hairpin stabilized by the template, and
the Cys1-Cys6 disulfide bridge. Hydrogen bonds may form
between the backbone amides in opposing â-strands as indicated
by the dotted lines in 8A (see also Figure 2B) which is consistent
The energy minimum SA structure found for 2, embedded
in a bath of water molecules, was taken as a starting point for
2 ns MD simulations at 300 K, performed with the 43A1 force

7444 J. Am. Chem. Soc., Vol. 120, No. 30, 1998
Bisang et al.
Figure 2. (A) Energy minimum structure deduced by dynamic simulated annealing for peptide 2. (B) Energy minimum structure deduced by
dynamic simulated annealing for peptide 8. Yellow ) S atoms, red ) O atoms, blue ) N atoms, gray ) C atoms. The structures were drawn using
Molscript.37
Table 5. Results of the MD Simulation with TA-DR for Peptide
field in GROMOS96,23 both with and without TA-DR. The
results for both runs were similar (Table 5), indicating that the
2a
Part A
restraints are well satisfied by structures at and near to the energy
energies
minimum in this force field. A simulation with TA-DR was
(kJ/mol)/temp (K)
av over the run
rms fluctuations
also performed with 2 as a Na+-carboxylate salt, but again the
results were not significantly different. Three parameters
illustrate well the types of motion undergone by 2 during the
simulation: (1) the Asn2 C(R) to Ala5 C(R) distance, which in
a â-turn is less than 7 Å;24 (2) the ø1 value of the aminoproline
moiety of the template, which is ≈ -35° in the starting SA
structure (Figure 2A) but flips to ≈ +10-12° in the â-hairpin
conformation seen in 8 (Figure 2B); and (3) the Pro3 and Asn4
φ/ψ angles, since these change from ≈ -60°/-30° and -90°/0
°C, respectively, in a â-I turn to ≈-60°/+120° and +80°/0 °C
in a â-II turn.10,24
The Asn2 C(R) to Ala5 C(R) distance in 2 remains ≈6 Å
during almost the entire simulation, both with (Figure 3) and
without TA-DR (data not shown). Moreover, the Pro3 φ/ψ and
Asn4 φ angles remain for most of the simulation within (30°
of the values expected for a â-I turn (Figure 4); only very
infrequently are conformers populated close to a â-II turn. The
Pro3 H-C(R) to Asn4 NH upper distance restraint shows the
largest single violation (Table 5). The variation of ø1 in the
aminoproline moiety of the template is shown in Figure 5. For
most of the simulation the ring pucker remains in the range ø1
≈ -30° to -20°, close to that in the starting structure (ø1 ≈
-35°), with the N-C(γ) in a pseudoequatorial position.
However, ring flips to a ø1 > 0° are observed, which brings
the N-C(γ) into a pseudoaxial position and the template and
the peptide loop into a â-hairpin-like conformation (Figure 6),
similar to that deduced for 8 (see Figure 2). A similar ring
flipping was also seen in the unrestrained MD simulation of 2
(data not shown).
potential energy
bond angle energy
improper dihedral energy
dihedral energy
electrostatic energy
peptide-peptide
peptide-solvent
solvent-solvent
LJ interaction energy
peptide-peptide
peptide-solvent
solvent-solvent
distance restraint
energy
-47 817
137
165
17
7.9
11.7
34
81
-430
-502
-54 990
27
64
323
-108
-202
8060
11
21
226
1.9
302.5
0.7
3.8
temperature
Part B
sum of violations of distance restraints (Å)
largest violation (Å)
0.79
0.53
Part C
hydrogen bonds (donor:acceptor)
%
Asn2 HδE:Apro7 O
Asn2 HδZ:Apro7 O
Asn4 HN:Asn2 Oδ
23
19
26
50
11
3
Ala5 HN:Asn2 Oδ
Apro7 HN-C(γ):Asn2 Oδ
Apro7 HN-C(γ):Ala5 O
Asn2 HN:Asp8 Oδ
2
a Part A, gives the average contributions to the total energy separated
on the basis of terms in the force field and the rms fluctuations as well
as the temperature. Part B gives the distance restraint violations. Part
C shows the percentage population of hydrogen bonds to a cut-off of
2%. Apro7 refers to the aminoproline moiety of the template. All values
are taken over the entire 2 ns simulation.
Immunological Studies. To investigate whether the cyclic
mimetic 2 can elicit antibodies that bind to P. falciparum
sporozoites, peptide 9 was prepared which contains also a
promiscuous T-helper cell epitope.25 As a control, the same
ANPNAA epitope as a linear peptide was incorporated into the
immunogen 10. Groups of BALB/c mice were immunized five
(23) van Gunsteren, W. F.; Billeter, S. R.; Eising, A. A.; Hu¨nenberger,
P. H.; Kru¨ger, P.; Mark, A. E.; Scott, W. R. P.; Tironi, I. G. Biomolecular
simulation: The GROMOS96 manual and user guide; Hochschulverlag AG
an der ETH Zurich: Zurich, 1996.
(25) Kumar, A.; Arora, R.; Kaur, P.; Chauhan, V. S.; Sharma, P. J.
Immunol. 1992, 148, 1499-1505.
(24) Wilmot, C. M.; Thornton, J. M. J. Mol. Biol. 1988, 203, 221-232.

Cyclic NPNA Antigen
J. Am. Chem. Soc., Vol. 120, No. 30, 1998 7445
times subcutaneously with 9 or 10 together with the adjuvant
QS-21. A control group received only the adjuvant. The
development of IgG titers was monitored by ELISA (Figure
7). Mice immunized with 9 developed relatively low anti-9
IgG titers, which increased slowly with each immunization.
Serum antibodies of these mice did not react with 10 (Figure
7), indicating that 2 is the immunodominant determinant of 9.
No cross-reactivity with 9 developed in mice immunized with
10 or with adjuvant alone.
Figure 3. Interatomic distance Asn2 C(R) to Ala5 C(R) (Å) as a function
of time during the MD simulation of 2 with TA-DR.
Cross-reactivity with P. falciparum sporozoites was tested
with an immunofluorescence assay. None of the mice im-
munized with 10 or with adjuvant alone stained sporozoites. In
contrast, staining was obtained with the serum of some of the
15 mice immunized with 9 (Figure 8). After three immuniza-
tions, one serum was positive, after four immunizations three
sera were positive, and after an additional immunization 12 of
15 mice were positive. The specificity of the cross-reaction
was analyzed by competition experiments. Incubation in the
presence of the immunogen 9, or the mimetic 2, completely
abolished the antibody reactivity of positive sera with sporo-
zoites (Figure 8). Neither the immunogen 10 containing the
linear epitope nor cyclic peptides with different sequences
mounted on the template inhibited the binding of the cross-
reactive anti-9 antibodies to the sporozoites.
Discussion
The design of the template 1 represents an attempt to discover
novel semirigid structural units that, in the context of a cyclic
peptide, exert an influence on the peptide conformation and
allow conjugation to other building blocks, including T-cell
epitopes in a multiple-antigen-peptide suitable for immuniza-
tions.14 The template itself is readily accessible and might prove
to be generally useful in the evaluation of conformationally
restrained B-cell epitopes.
The inherent flexibility of linear synthetic peptides is a well-
known drawback in their use as synthetic vaccines, in particular
for mimics of continuous conformational protein epitopes.26
There is, however, increasing interest in the rational design of
constrained synthetic peptide vaccines. Ideally, the backbone
of the peptide should adopt conformations present in the epitope,
on the intact protein, while bound to a neutralizing antibody.
We chose to examine here a synthetic epitope comprising just
four amino acids, NPNA, which occurs naturally in a tandemly
repeated form in the CS protein of P. falciparum. The key issue
in this case was to create a constrained and conformationally
defined peptide that can elicit antibodies able to bind the
parasite. Both linear and cyclic peptides containing multiple
NPNA-motifs have been shown already to act in this way,3,5,27,28
albeit with varying levels of effectiveness, but their preferred
conformations are less well defined. An improved understand-
ing of the secondary structure of the repeat region, and its
sequence dependence, might be useful for the design of more
effective synthetic malaria vaccines.
Figure 4. Ramachandran plot for the Pro3 and Asn4 residues taken
every 0.2 ps during the 2 ns MD simulation of 2 with TA-DR.
Figure 5. Plot of the ø1 angle (N, C(R), C(â), C(γ) torsion) in the
aminoproline moiety of the template as a function of time at 0.2 ps
time steps during the MD simulation of 2 with TA-DR.
in the peptide backbone should be possible, whereas this would
not be the case with only four residues attached to the template21
(see Figure 2). Before any attempts to determine average
solution structures were made, evidence for stable secondary
structure was sought, based on analyses of chemical shifts,
coupling constants, amide temperature coefficients, relative
amide H/D exchange rates, and NOEs for both 2 and the more
highly restrained peptide 8. According to each of these
parameters, a stable â-turn appeared to be significantly populated
in both molecules (see Results section).
The predominance of a type-I â-turn in the NPNA motif in
both 2 and 8 was indicated by SA structures calculated using
NOE-derived upper distance restraints (Figure 2). The SA
structure deduced for 8 is consistent with the NMR data, which
supports our view of this molecule as being relatively rigid.
However, some of the coupling constants observed for 2 are
Since earlier studies had indicated that synthetic peptides
containing tandemly repeated NANP and NPNA motifs populate
â-I turns based on the NPNA cadence, in rapid dynamic
equilibrium with unfolded forms,9,11,12 we set out to stabilize
this conformation in a single NPNA motif in the context of a
cyclic peptide with template 1. The cyclic peptide antigen 2,
with six amino acids attached to the template, was chosen for
study, since modeling indicated that a â-I hairpin conformation
(26) Brown, F. Philos. Trans. R. Soc. London B 1994, 344, 213-219.
(27) Etlinger, H. M.; Renia, L.; Matile, H.; Manneberg, M.; Mazier, D.;
Trzeciak, A.; Gillessen, D. Eur. J. Immunol. 1991, 21, 1505-1511.
(28) Etlinger, H. M.; Trzeciak, A. Philos. Trans. R. Soc. London B 1993,
340, 69-72.

7446 J. Am. Chem. Soc., Vol. 120, No. 30, 1998
Bisang et al.
Figure 6. Snapshots taken from the MD simulation with TA-DR of 2 having different ø1 angle in the aminoproline moiety of the template (Figure
5); left, at 1.7 ns (ø1 ) -37°); middle, at 1.06 ns (ø1 ) +0.5°); right at 1.04 ns (ø1 ) +20.5°). H atoms except amide NHs are omitted for clarity.
The structures were drawn using Molscript.37
inconsistent with the average structure determined by SA (Figure
2A). In an attempt to explore which other conformations might
be populated in 2, the upper distance restraints were imple-
mented in a time-averaged mode in a MD simulation with
solvent water present. The use of time-averaged restraints in
MD simulations has been proposed29 and tested,30,31 as one
means to account for the averaging that is inherent in NMR
spectroscopic data.
During the MD simulation of 2 with TA-DR the NPNA motif
populates predominantly a type-I â-turn. However, it is
noteworthy that an Asn2 NH to Ala5 CO H-bond is seldom
formed. Rather, the most frequently formed H-bond is between
the side chain Asn2 CO(δ) and Ala5 NH (see Table 5), as seen
in Figure 2 and 6. A conformational transition in the template
between conformations with the C(γ)-N in a pseudoequatorial
position (ø1 ≈ -30°), and a pseudoaxial position (ø1 ≈ 0 to
+30°), is observed in the simulation (Figures 5 and 6), which
is consistent with the averaged 3J coupling constants observed
in the aminoproline moiety of the template in 2 (Table 4). The
peptide 8 serves as a useful reference, since here the template
is locked into the ø1 ≈ +10-20° conformation.
Thus the NMR and MD simulation give a self-consistent view
Figure 7. Serum IgG titers of 15 BALB/c mice immunized four times
of the conformation of the peptide antigen 2, which includes a
subcutaneously with 9 together with the adjuvant QS-21. ELISA was
stable â-I turn in the NPNA motif but also some flexibility in
performed with ELISA microtiter plates coated either with 9 (A) or
the template. The â-turn conformation is similar to those found
with 10 (B) and incubated with serial dilutions of the individual mouse
sera. Bound IgG was quantitated using alkaline phosphatase conjugated
antibodies specific for mouse gamma heavy chains (see Experimental
in linear peptides of the type (NPNA)3 and (NPMeNA)3 by NMR
and MD methods.11,12 For example, the backbone N, C(R), and
C′ atoms of the residues in the ANPNA loop in a typical
structure of 2 (for example, the snapshot at 1.7 ns from the
MD simulation) can be superimposed on the central ANPMeNA
motif of a typical structure taken from an MD simulation12 of
(NPMeNA)3 with an rmsd of ≈0.6 Å (Figure 9).
Section).
falciparum sporozoites (Figure 8); the remaining mice gave only
a response against 9 by ELISA. The binding of anti-9 serum
to sporozoites was inhibited both by the immunogen 9 and by
the mimetic 2 (Figure 8). This latter observation is consistent
with the specific recognition of sporozoites by anti-2 antibodies
in the sera. In control experiments, cyclic peptides with different
sequences mounted on the same template were unable to inhibit
antibody binding to sporozoites in the IFA. On the other hand,
the sera from none of the mice immunized with 10 (or with
adjuvant alone) cross-reacted with sporozoites. Possible reasons
for the inactivity of immunogen 10 may include the fact that
only a single NPNA motif is present, that its conformational
flexibility is high, and/or that its susceptibility to proteolytic
degradation in serum is high, in comparison to 9.
For antibody production, the two multiple-antigen-peptides
9 and 10 incorporating the NPNA motif and a single universal
T helper cell epitope derived from tetanus toxin25 were each
administered to BALB/c mice together with the adjuvant QS-
21. After four booster immunizations, the sera from 12/15 mice
immunized with 9 showed specific IgG responses against the
immunogen 9. Some of the elicited IgG from these 12 mice
cross-reacted in an immunofluorescence assay (IFA) with P.
(29) Torda, A. E.; Scheek, R. M.; van Gunsteren, W. F. Chem. Phys.
Lett. 1989, 157, 289-294.
(30) Pearlman, D. A. J. Biomol. NMR 1994, 4, 1-16.
(31) Nanzer, A. P.; van Gunsteren, W. F.; Torda, A. E. J. Biomol. NMR
1995, 6, 313-320.
A wider question concerns whether the unnatural template
itself is a major epitope in 9, which could result in the

Cyclic NPNA Antigen
J. Am. Chem. Soc., Vol. 120, No. 30, 1998 7447
Figure 9. Superimposition (over the N, C(R), and C′ atoms) of the
ANPNA sequence in a snapshot from the MD simulation of 2 (at 1.7
ns), with the central ANPMeNA motif taken from an earlier MD
simulation12 of the linear peptide (NPMeNA)3.
repeat peptides. Earlier MD studies12 of (NPMeNA)3 peptides
indicated how longer tandem repeat peptides might begin to
fold and hence what structures the major epitope(s) might adopt.
In particular, the NPNA motif was seen to define not only a
sequence repeat but also a conformational repeat, which could
be used for the assembly of larger epitopes on the native CS
protein. The results reported here certainly strengthen the
conclusion that a â-I turn is the major conformational epitope
within the NPNA motif.
Figure 8. (A) Immunofluorescence staining of P. falciparum sporo-
zoites by serum IgG from a mouse immunized with 9 (see Experimental
Section for method); (B) Incubation in the presence of the mimetic 2
(50 µg/mL) abolished staining of the sporozoites.
immunological response being directed toward this unnatural
antigen. This becomes an issue whenever unnatural templates
or amino acids are intended for use in synthetic immunogens
but has not been pursued in the present work, where the focus
has been on a characterization of the antigen conformation and
the anti-NPNA response elicited by 9.
Experimental Section
Methyl (2S,4R)-4-Hydroxyprolinate Hydrochloride. To a suspen-
sion of (2S,4R)-4-hydroxyproline (60.0 g, 458 mmol) in MeOH (550
mL) was added thionyl chloride (68 mL, 937 mmol) at 0 °C. The
resulting suspension was stirred 30 min at room temperature and then
refluxed for 4 h. Evaporation of solvent gave product (82.8 g, 99%)
which was used without further purification. For analytical purposes
the product was recrystallized from Et2O/MeOH (93%). Mp 162-
The finding that a single NPNA motif in the form of a cyclic
peptide, constrained into a â-I turn conformation, can elicit anti-
sporozoite antibodies provides additional support for the im-
munological relevance of the NPNA cadence in the repeat region
of the CS protein and provides a direct correlation between a
â-I turn conformation in the free antigen and immune recogni-
tion of the parasite. However, the free antigen 2 is not rigid
(vide supra) and as yet the antibody-bound conformation(s)
is(are) not known. By introducing the conformational constraint,
the specificity of the immune response to the NPNA repeat is
undoubtedly more stringent than would be the case with a linear
peptide immunogen. For this reason, the ability of 9 to elicit
anti-sporozoite antibodies is significant. With the multiple-
antigen-peptide construct used here, the immunogenicity of 2
was low, and anti-2 serum IgG titers increased only gradually
with each booster immunization. However, this will be
influenced also by the type and copy number of the T-helper
cell epitope in the immunogen. Improved immunogenicity can
be expected by conjugation to a carrier containing multiple
T-cell epitopes. It was noted earlier,27,32 that the fine specificity
of antibodies to the repeat sequence of the CS protein could be
manipulated by using sequence variants of (NANP)n. In
particular, the major epitopes in the repeat region on the parasite
seem from serological studies32-34 to encompass at least two
or three NANP repeats. This implies the adoption of more
complex, immunologically relevant structures in longer tandem
164°. [R]D ) -24.3 (c 1.05, MeOH). 1H NMR (300 MHz,
21
(D6)DMSO): 2.09 (ddd, J ) 13.3, 10.9, 4.5, 1 H); 2.20 (dddd, J )
13.3, 7.5, 1.6, 1.6, 1 H); 9.96 (br., 2 H); 3.07 (ddd, J ) 12.3, 1.6, 1.6,
1 H); 3.37 (dd, J ) 12.1, 4.5, 1 H); 3.76 (s, 3 H); 4.42-4.40 (m, 1 H);
4.45 (dd, J ) 10.9, 7.4, 1 H); 5.64 (br., 1H). CI-MS (NH3): 146.2
(100, [M + H]).
(2S,4R)-N-[(Benzyl)oxycarbonyl]-4-hydroxy-L-proline Methyl Es-
ter. To the foregoing product (41.0 g, 225.7 mmol) in dioxane (250
mL) and 1 M aqueous Na2CO3 (500 mL) at 0 °C was added
benzyloxycarbonyl chloride (50.3 mL, 339 mmol). The mixture was
stirred 1 h at 0 °C and overnight at room temperature. After acidifying
to pH 2, extraction with EtOAc, washing with brine and water, and
drying (MgSO4), the solvent was evaporated. The residue was purified
by column chromatography (silica, EtOAc/hexane 1:1, then EtOAc/
21
hexane 2:1) to yield product (51.7 g, 84%) as a colorless oil. [R]D
)
-57.1 (c 1.0, CH2Cl2). 1H NMR (300 MHz, CDCl3): 2.05 (m, 1H);
2.28 (m, 1H); 2.69 (s, 1H); 3.63 (m, 2H); 3.53 (s, 9/5H), 3.73 (s, 6/5H);
4.48 (m, 2H); 4.98 (d, J ) 12.5, 3/5H); 5.09 (d, J ) 12.5, 2/5H); 5.14
(d, J ) 12.5, 2/5H); 5.16 (d, J ) 12.5, 3/5H). CI-MS (NH3): 297 (11,
[M + NH4]+), 280 (100, [M + H]+).
(2S,4R)-N-[(Benzyl)oxycarbonyl]-4-[(trimethylsilyl)oxy]-L-pro-
line Methyl Ester. To the foregoing product (47.4 g, 170 mmol) in
THF (420 mL) were added NEt3 (118.4 mL, 849.0 mmol) and
trimethylsilyl chloride (53.7 mL, 424.5 mmol). A white precipitate
was formed. After 21 h the suspension was filtered and the filtrate
was washed with Et2O. The etheral solution was washed with 1 M
aqueous Na2CO3 (500 mL), dried (MgSO4), and evaporated to afford
the silyl ether as a clear orange oil (59.7 g, 100%). 1H NMR (300
MHz, CDCl3): 0.13 (s, 9H); 2.20-2.10 (m, 1H); 2.16-2.26 (m, 1H);
(32) DelGiudice, G.; Tougne, C.; Renia, L.; Ponnudurai, T.; Corradin,
G.; Pessi, A.; Verdini, A. S.; Louis, J. A.; Mazier, D.; Lambert, P.-H. Mol.
Immunol. 1991, 9, 1003-1009.
(33) Zavala, F.; Tam, J. P.; Hollingdale, M. R.; Cochrane, A. H.; Quakyi,
I.; Nussenzweig, R. S.; Nussenzweig, V. Science 1985, 228, 1436-1440.
(34) Togna, A. R.; DelGiudice, G.; Verdini, A. S.; Bonelli, F.; Pessi,
A.; Engers, H. D.; Corradin, G. J. Immunol. 1986, 137, 2956-2960.

7448 J. Am. Chem. Soc., Vol. 120, No. 30, 1998
Bisang et al.
3.42 (dd, J ) 11.0, J ) 3.0, 0.5H); 3.50 (dd, J ) 11.0, J ) 1.6, 0.5H);
3.70 (dd, J ) 11.0, J ) 5.0, 1H); 3.67, 3.77 (2s, 3H); 4.31-4.52 (m,
2H); 5.04 (d, J ) 12.5, 0.5H); 5.12 (d, J ) 12.5, 0.5H); 5.20 (d, J )
12.5, 0.5H); 5.22 (d, J ) 12.5, 0.5H); 7.21-7.40 (m, 5H).
DCC (2.14 g, 10.3 mmol) in CH2Cl2 (40 mL) at room temperature
was added Fmoc-L-Asp(OAllyl)-OH (4.07 g, 10.3 mmol) in CH2Cl2
(28 mL). After stirring 24 h, the solution was filtered and evaporated.
The resulting yellow foam was dissolved in a solution of DBU in MeCN
(5% v/v, 24 mL), stirred for 8 h, then diluted with CH2Cl2 (70 mL),
and washed first with 10% aqueous citric acid and then brine, and the
combined organic fractions were dried (MgSO4) and evaporated. The
residue was redissolved in CH2Cl2, filtered to remove dicyclohexyl urea,
evaporated, and then purified by flash chromatography (EtOAc/hexane
(2S,4R) and (2S,4S)-N-[(Benzyl)oxycarbonyl]-2-{[(tert-butyl)oxy-
carbonyl]methyl}-4-hydroxyproline Methyl Ester (3) and (3′). To
a solution of diisopropylamine (35.5 mL, 250.2 mmol) in anhydrous
THF (370 mL) at -5 °C was added n-BuLi (133.7 mL, 213.9 mmol,
1.6 M in hexane). After stirring for 15 min at -5 °C the solution was
cooled to -70 °C. A solution of the foregoing silyl ether (62.9 g,
178.7 mmol) in anhydrous THF (150 mL) was added dropwise and
stirred for 30 min at -70 °C, and then tert-butyl-2-bromoacetate (45.5
mL, 304 mmol) was added dropwise. After 15 min at -70 °C, 90
min at 0 °C, and overnight at room temperature, the reaction mixture
was poured into saturated aqueous NH4Cl (400 mL) and extracted with
EtOAc (3 × 500 mL). The combined organic fraction was dried
(MgSO4) and evaporated to yield a dark brown oil (97.7 g). This was
dissolved in a methanolic citric acid solution (500 mL, 10%) and stirred
3 h at room temperature. Then 250 mL of EtOAc was added, and the
MeOH was partly evaporated. After addition of EtOAc and washing
with 1 M aqueous Na2CO3, the organic fraction was dried (MgSO4)
and evaporated, and the residue was chromatographed (silica, EtOAc/
hexane 1:2) to yield a diastereomeric mixture of 3 and 3′ as a brown
oil (52.6 g, 72%). CI-MS (NH3): 411 (25, [M + NH4]+), 294 (100,
[M + H]+).
21
1:1) to afford 5 as a white solid (1.69 g, 64%). Mp 191-192°. [R]D
) -32.3 (c 0.99, CHCl3). IR (KBr): 3450w (br.), 3360m, 2980m,
2930w, 1770m, 1725s, 1710s, 1675s, 1660s, 1615w, 1440m, 1380s.
1H NMR (300 MHz, CDCl3): 1.45 (s, 9H); 2.58 (dd, J ) 13.7, J )
9.0, 1H); 2.69 (d, J ) 16.0, 1H); 2.77 (dd, J ) 13.7, J ) 7.0, 1H);
2.91 (dd, J ) 17.7, J ) 10.8, 1H); 3.03 (d, J ) 16.0, 1H); 3.47 (dd, J
) 17.7, J ) 3.1, 1H); 3.72 (dd, J ) 13.2, J ) 9.0, 1H); 4.34 (dd, J )
13.2, J ) 3.7, 1H); 4.68 (d, J ) 5.8, 1H); 4.73 (dd, J ) 10.8, 3.1, 1H);
5.08 (dddd, J ) 9.1, 8.9, 7.0, 3.7, 1H); 5.37 (d, J ) 17.2, 1H); 5.95
(ddt, J ) 17.2, 10.3, 5.8, 1H); 6.76 (s, 1H); 7.70 and 7.82 (2m, 2 ×
2H). 13C NMR (75.4 MHz, CDCl3): 27.9; 36.8; 40.4; 42.2; 45.4; 48.9;
52.3; 65.8; 82.1; 118.8; 123.5; 131.5; 134.2; 164.0; 167 0.4; 168.3;
169.2; 171.4. ES-MS (NaI): 534.5 (100, [M + Na]+). Anal. Calc.
for C26H29N3O8 (511.54): C 61.0, H 5.7, N 8.2. Found: C 60.6, H
5.7, N 8.2.
Allyl ((2S,6S,8aS)-8a-[(tert-butyl)oxy)carbonyl]methyl}perhydro-
5,8-dioxo-2-[(fluorenyl)methyloxycarbonyl]pyrrolo[1,2-a]pyrazine-
6-acetate (1). To a solution of 5 (2.00 g, 3.91 mmol) in DMF (20
mL) was added methylhydrazine (800 µL, 15.2 mmol). After stirring
4.5 h at 60 °C, the solvent was removed in vacuo, the residue was
redissolved in CH2Cl2 and filtered, and the solvent evaporated. The
residue in CH2Cl2 (20 mL) was treated with Fmoc-succinimide (2.64
g, 7.83 mmol), imidazole (0.47 g, 7.83 mmol), and DMAP (48 mg,
3.93 mmol) and stirred for 3 h at room temperature. After washing
with aqueous citric acid (0.1%) and brine and drying (MgSO4), the
solvent was evaporated, and the residue was purified by chromatography
(silica, EtOAc/hexane 3:2) to give the corresponding amine (1.37 g,
58%) as a white powder. IR (film): 3386w, 3008m, 3015w, 1728s,
1682s, 1518m, 1450s, 1155 s. 1H NMR (300 MHz, CDCl3): 1.45 (s,
9H); 2.58 (dd, J ) 13.7, J ) 9.0, 1H); 2.69 (d, J ) 16.0, 1H); 2.77
(dd, J ) 13.7, J ) 7.0, 1H); 2.91 (dd, J ) 17.7, J ) 10.8, 1H); 3.03
(d, J ) 16.0, 1H); 3.47 (dd, J ) 17.7, J ) 3.1, 1H); 3.72 (dd, J )
13.2, J ) 9.0, 1H); 4.34 (dd, J ) 13.2, J ) 3.7, 1H); 4.68 (d, J ) 5.8,
1H); 4.73 (dd, J ) 10.8, 3.1, 1H); 5.08 (dddd, J ) 9.1, 8.9, 7.0, 3.7,
1H); 5.37 (d, J ) 17.2, 1H); 5.95 (ddt, J ) 17.2, 10.3, 5.8, 1H). 13C
NMR (75 MHz, CDCl3): 27.89; 36.24; 42.32; 43.32; 47.08; 47.83;
51.41; 52.07; 65.16; 65.76; 66.77; 82.01; 118.83; 119.89; 124.91;
126.99; 127.64; 131.42; 141.19; 143.71; 155.83; 164.91; 168.42; 169.96;
170.44. ES-MS (NaI): 626.4 (100, [M + Na]+). The foregoing amine
(1.00 g, 1.66 mmol) in CH2Cl2 (2 mL) and TFA (4 mL) was stirred at
room temperature and then evaporated, and dioxane was added and
evaporated to remove traces of TFA. The residue was purified by
chromatography (silica, EtOAc, then EtOH) to yield product (880 mg,
97%) as a white hygroscopic powder. IR (film): 3378s (br.), 3007m,
2966m, 2859m, 1719s, 1745s, 1452s, 1177s, 1121 s. 1H NMR (300
MHz, CDCl3): 1.44 (s, 9H); 2.25 (dd, J ) 14.0, 6.0, 1H); 2.59 (dd, J
) 14.0, 7.5, 1H); 2.59 (dd J ) 16.0, 7.3, 1H); 2.98 (d, J ) 16.0, 1H);
3.04 (m, 2H); 3.62 (dd, J ) 13.0, 7.0, 1H); 4.05 (br. d, J ) 13.0, 1H);
4.20 (t, J ) 7.0, 1H); 4.39 (d, J ) 7.0, 2H); 4.35-4.50 (m, 2H); 4.52
(m, 2H); 5.26 (d, J ) 10.5, 1H); 5.30 (d, J ) 17.5, 1H); 5.94 (d, J )
8.5, 1H); 5.89 (dddd, J ) 17.5, 10.5, 5.5, 4.5, 1H); 6.27 (s, 1H); 7.31
(m, 2H); 7.40 (m, 2H); 7.58 (d, J ) 7.3, 2H); 7.76 (d, J ) 7.5). 13C
NMR (75 MHz, CDCl3): 35.95; 40.50; 42.56; 46.89; 47.57; 51.04;
52.13; 60.89; 64.87; 65.87; 66.91; 118.77; 119.88; 124.86; 126.99;
127.67; 131.23; 141.13; 143.55; 156.33; 160.88; 165.48; 170.70; 172.64.
ES-MS (NaI): 570.5 (100, [M + Na]+).
(2S,4R) and (2S,4S)-N-[(Benzyl)oxycarbonyl]-2-{[(tert-butyl)oxy-
carbonyl]methyl}-4-phthalimidoproline Methyl Ester. To the fore-
going product (13.0 g, 33.0 mmol), phthalimide (14.6 g, 99.0 mmol),
and PPh3 (21.6 g, 82.5 mmol) in THF (400 mL) at -25 °C was added
diethyl azodicarboxylate (11.3 mL, 72.6 mmol). After 3 h at -25°,
and at room temperature overnight, the solvent was removed in vacuo,
and the residue was purified by column chromatography (silica, EtOAc/
hexane, 1:2). The remaining phthalimide was precipitated from AcOEt/
hexane 1:3, and evaporation of the mother liquor yielded the two
diastereoisomers as a viscous brown oil (12.1 g, 70%). CI-MS (NH3):
523 (100, [M + H]+).
(2S,4R) and (2S,4S)-2-{[(tert-Butyl)oxycarbonyl]methyl}-4-phthal-
imidoproline Methyl Ester (4 and 4′). The foregoing phthalimides
(11.31 g, 13.4 mmol) in DMF (180 mL) with Pd/C (2.32 g, 10% Pd)
were stirred 4 days at room temperature and 1 atm under H2. The
solvent was removed in vacuo, the residue was dissolved in CH2Cl2,
and filtered through a Celite pad, and the solvent evaporated to give 4
and 4′ as a slightly yellow solid. This residue was dissolved in the
minimum volume of CH2Cl2 (61 mL) and 1.5 times this volume hexane
(92 mL) was added. This mixture was seeded with pure diastereoisomer
4 and left at 4 °C overnight. The resulting white precipitate was
collected and washed to give 4 (3.46 g, 41.2%). The mother liquor
was evaporated, and crystallization of the residue from hot hexane gave
diastereoisomer 4′ as a white solid. Repeating this procedure on mother
liquors gave additional 4 (yield 47%) and 4′ (yield 23%). 4: Mp 181-
183°. [R]D21 ) -14.0 (c 1.03, CH2Cl2). IR (KBr): 3460w (br.), 3355m,
3010w, 2970m, 2900w, 1770m, 1730s, 1710s, 1615w. 1H NMR (300
MHz, CDCl3): 1.37 (s, 9H); 2.17 (dd, J ) 13.7, J ) 10.0, 1H); 2.58
(d, J ) 15.7, 1H); 2.63 (dd, J ) 13.7, J ) 7.5, 1H); 2.93 (d, J ) 15.7,
1H); 3.22 (dd, J ) 9.4, J ) 8.0, 1H); 3.51 (dd, J ) 9.4, J ) 9.0, 1H);
3.80 (s, 3H); 4.85 (ddd, J ) 10.0, 9.0, 8.0, 1H); 7.64 and 7.74 (2m, 2
× 2H). 13C NMR (75.4 MHz, CDCl3): 27.9; 38.1; 45.6; 48.4; 48.8;
52.6; 66.1; 81.2; 123.2; 131.7; 134.0; 167.8; 169.5; 175.1. CI-MS
(NH3): 389 (100, [M + H]+). Anal. Calc. for C20H24N2O6 (388.42):
C 61.9, H 6.2, N 7.2. Found: C 62.0, H 6.2, N 7.3. 4′: Mp 110-
111°. [R]D21 ) -20.4 (c 1.02, CH2Cl2). IR (KBr): 3450w (br.), 3335w,
2965w, 1775w, 1735s, 1710s, 1695s. 1H NMR (300 MHz, CDCl3):
1.43 (s, 9H); 2.42 (dd, J ) 13.2, J ) 9.1, 1H); 2.54 (dd, J ) 13.2, J
) 8.9, 1H); 3.00 (s, 2H); 3.37 (dd, J ) 11.0, J ) 7.5, 1H); 3.43 (dd,
J ) 11.0, J ) 8.5, 1H); 4.82 (ddd, J ) 8.8, 8.5, 7.2, 1H); 7.72 and
7.83 (2m, 2 × 2H). 13C NMR (75.4 MHz, CDCl3): 27.9; 38.5; 44.3;
48.8; 49.5; 52.6; 67.1; 81.2; 123.2; 131.7; 134.0; 168.0; 169.8; 174.9.
CI-MS (NH3): 389 (100, [M + H]+).
Peptide Synthesis. Loop 2. Tentagel-S AC resin (Rapp Polymere,
Tu¨bingen) (1.0 g, 0.23 mmol/g) in CH2Cl2 and pyridine (3 mL, 1:1)
was treated with 1 (0.38 g, 3 equiv) and 2-chloro-1,3-dimethylimid-
azolidinium hexafluorophosphate (CIP) (386 mg, 6 equiv) with swirling
for 1.5 h. The resin was then filtered and washed successively with
CH2Cl2, methanol, and CH2Cl2. The substitution level of template on
Allyl ((2S,6S,8aS)-8a{[(tert-Butyl)oxycarbonyl]methyl}perhydro-
5,8-dioxo-2-phthalimidopyrrolo[1,2-a]pyrazine-6-acetate (5). To a
suspension of 4 (2.00 g, 5.15 mmol), HOAt (1.40 g, 10.3 mmol), and

Cyclic NPNA Antigen
J. Am. Chem. Soc., Vol. 120, No. 30, 1998 7449
the resin was 65% (0.15 mmol/g). The Fmoc group was removed with
piperidine in DMF (20%), and the amino acids Fmoc-Ala-OH, Fmoc-
Ala-OH, Fmoc-Asn(Mtt)-OH, Fmoc-Pro-OH, Fmoc-Asn(Mtt)-OH, and
Fmoc-Ala-OH were added sequentially using standard solid-phase
methods.17 The allyl ester was removed by treating resin-bound peptide
with Pd(PPh3)4 (345 mg, 2 equiv), N-methylmorpholine (0.3 mL), and
AcOH (0.6 mL) in DMF (12 mL) under oxygen-free conditions. After
2 h, the resin was filtered and washed successively with CH2Cl2,
diisopropylethylamine in DMF (0.5% v/v), sodium diethyldithiocar-
bamate in DMF (0.5%), methanol, and finally CH2Cl2. Deprotection
was monitored by reverse-phase HPLC (C18 column, eluting with a
gradient from 0 to 100% MeCN in H2O + 0.1% TFA). The N-terminal
Fmoc group was then removed with piperidine in DMF (20%), and
cyclization was performed with benzotriazol-1-yloxy-tris(dimethyl-
amino)phosphonium hexafluorophosphate (BOP) (61 mg, 1.5 equiv)
in N-methylpyrrolidone-DMSO (9 mL, 4:1) and diisopropylethylamine
(65 µL, 3 equiv) for 1.5 h. Upon completion, the peptide was cleaved
from the resin with TFA-triisopropylsilane-H2O (95:2.5:2.5) and
purified by reverse-phase HPLC (see above for conditions) to give 2
(20.0 mg) (EI-MS: 806.5 (100, [M + H]+)) and a dimer (EI-MS: 1612
(100, [M + H]+)).
For the MD simulation of 2 with and without TA-DR the GRO-
MOS96 suite of programs was used, with the 43A1 force field for
simulations with solvent water.23 The 27 upper distance restraints used
were the exact values obtained from ROE build-up curves, with a
memory decay time τdr ) 100 ps and a force constant Kdr ) 1000 kJ
mol-1 nm-2. The starting structure was the lowest energy SA structure
in a truncated octahedral box (box length 41.42 Å) with 1142 SPC
water molecules. The temperature was held constant by weak coupling
(τT ) 0.1 ps) to an external bath at 300 K. The SHAKE algorithm
was used to maintain bond lengths with a relative precision of 10-4
and the integrator time step was 0.002 ps. Nonbonded interactions
evaluated at every step were within a short-range cutoff of 8 Å. For
long-range interactions the cutoff was 14 Å. Structures were saved
for analysis every 100 steps (0.2 ps). After short simulations to relax
the solute and solvent, the simulations with and without TA-DR were
each run for 2 ns.
Immunization of Mice. Two groups of 15 female BALB/c mice,
six to eight weeks old, were immunized subcutaneously with 50 µg of
9 or 10 in phosphate-buffered saline (100 µL) (pH 6.7) containing 20
µg of the adjuvant QS-21.35 As a control, a third group of five mice
received 20 µg of QS-21 in PBS (100 µL). Mice were immunized
four times at two week intervals and a fifth time 2 months after the
fourth immunization. Mice were bled by retroorbital puncture before
the first immunization and 11-14 days after the third, fourth, and fifth
immunizations. Blood was collected and incubated at 37 °C for 1 h
and at 4 °C for 3 h, and the cells were then removed by centrifugation.
Loop 8. This loop was made following the same procedure as used
for 2, except for the use of Fmoc-Cys(Trt)-OH in peptide assembly.
After cleaving the cyclic peptide from the resin with TFA-thioanisole-
1,2-ethanedithiol-H2O-triisopropylsilane (87.5:5:2.5:2.5:2.5), air oxi-
dation of the free dithiol in ammonium acetate buffer (pH 8), monitored
by reverse phase HPLC, gave 8 (5.6 mg, 8%) (EI-MS: 868.4 (36, [M
+ H]+)).
Enzyme-Linked Immunosorbent Assay (ELISA). ELISA micro-
titer plates (Immunolon 4B, Dynatech, Switzerland) were coated at 4
°C overnight with a solution of 9 or 10 in PBS (50 µL of 5 µg/mL, pH
7.2). Wells were then blocked with 5% milk powder in PBS for 1 h
at 37 °C and washed three times with PBS containing 0.05% Tween-
20 (Merck, Germany). Plates were then incubated with serial dilutions
of mouse serum in PBS containing 0.05% Tween-20 and 0.5% milk
powder (dilution buffer) for 2 h at 37 °C. After washing, the wells
were incubated with alkaline phosphatase conjugated-goat anti-mouse
IgG (γ-chain specific) antibodies (62 ng/mL in dilution buffer) for 2 h
at 37 °C and then washed. Phosphatase substrate (p-nitrophenyl
phosphate) (1 mg/mL) in buffer (Na2CO3 (1.4 g/L) NaHCO3 (3 g/L)
and MgCl2‚6H2O (0.2 g/L), pH 9.6) was added and incubated at room
temperature. After 1 h the absorbance was measured at 405 nm (Figure
7).
Multiple-Antigen-Peptides. The peptide H4-(Lys)2-Lys-Ala-Ala-
Gln(Mtt)-Tyr(tBu)-Ile-Lys(Boc)-Ala-Asn(Mtt)-Ser(tBu)-Lys(Boc)-Phe-
Ile-Gly-Ile-Thr(tBu)-Glu(OtBu)-Leu-Ala was assembled on Tentagel-S
AC resin using standard methods.17 The loop 2 (60 mg, 3 equiv) was
then coupled to this resin (47 mg, with 0.025 mmol free amino groups)
using O-(benzotriazol-1-yl)-N,N,N′,N′-tetramethyluronium hexafluoro-
phosphate (HBTU) (28 mg, 3 equiv) and 1-hydroxybenzotriazole
(HOBt) (9.4 mg, 3 equiv) in N-methylpyrrolidone (0.6 mL) and
diisopropylethylamine (19.4 µL) overnight at room temperature. The
product 9 was cleaved from the resin with TFA-triisopropylsilane-
H2O (95:2.5:2.5) and purified by reverse-phase HPLC (see above for
conditions) (5.2 mg). EI-MS: 5475.0 (100, [M + H]+). The peptide
10 was assembled by standard solid-phase methods,17 and purified in
the same way.
Immunofluorescence Assays. IFA were performed as described
elsewhere.36 Briefly, air-dried unfixed P. falciparum salivary gland
sporozoites (strain NF54) attached to microscope glass slides were
incubated in a moist chamber for 20 min at 37 °C with serum diluted
in PBS. The slides were then washed five times with PBS containing
0.1% bovine serum albumin (PBS-BSA) and dried. FITC-labeled goat
anti-mouse IgG (Fab-specific) antibodies (Sigma), diluted to 46 µg/
mL in PBS (pH 7.2) containing 0.1 g/L Evans-Blue (Merck, Germany),
were added. After incubation for 20 min at 37 °C the slides were again
washed five times with PBS-BSA, dried, and covered with glycerol
and a cover-slide. A Leitz Dialux 20 microscope using 12.5/18 ocular
and a 40x/1.30 oil fluorescence 160/0.17 objective was used to detect
fluorescence staining at 495 nm excitation and 525 nm emission
wavelengths. For competition experiments, serum was diluted 1:10
and incubated with cyclic peptides at final concentrations of 500 µg/
mL or 50 µg/mL for 1 h at 37 °C prior to staining. Both concentrations
yielded comparable results for each peptide tested.
1
NMR and Structure Calculations. For the peptides 2 and 8, H
2D ROESY, TOCSY, and DQF-COSY spectra were recorded in H2O/
D2O (90:10), pH 5, at 600 MHz (Bruker AMX600 spectrometer),
typically at a concentration of ≈20 mg/mL. The water signal was
presaturated, and 2D spectra were acquired with 2048 × 512 points,
except for determination of coupling constants from an E.COSY
spectrum measured with 8192 × 512 points. NMR spectra were
processed and analyzed using Felix software (MSI, San Diego).
To derive NOE distance restraints it was assumed that the initial
rate approximation is valid and that each peptide rotates as a single
isotropic rotor. The NOE connectivities for 2 were determined from
a series of ROESY spectra with mixing times of 40, 80, 120, and 200
ms and for 8 using 50, 100, 150, and 250 ms mixing times. Cross-
peak volumes were determined by integration, and the build-up curves
were checked to ensure a smooth exponential increase in peak intensity
for all NOEs used in deriving distance restraints, as a selection criterion
for cross-peak intensities due to direct NOEs and not the product of,
for example, TOCSY-type coherence pathways combined with cross-
relaxation.
The relative cross-peak volumes were assumed to be proportional
to r-6 and were used to derive distance restraints for simulated annealing
(SA) calculations and MD simulations. In the case of peptide 2, the
ROESY series was measured at 300 and at 285 K. The same distance
restraints were determined (within a margin of ≈10%) at both
temperatures, so the minor conformer of 2 should make a negligible
contribution to the NOEs measured for the major conformer (see Results
sectionsexchange rates). The minor conformer was not investigated
further. SA calculations were performed to derive average solution
structures, using methods described in detail elsewhere.21
Acknowledgment. This work was supported by grants from
the Swiss National Science Foundation and the Roche Research
Foundation. We thank Prof. van Gunsteren for providing the
GROMOS96 suit of programs for computer simulations.
JA980444J
(35) Kensil, C. R.; Soltysik, S.; Wheeler, D. A.; Wu, J. Y. AdV. Exp.
Med. Biol. 1996, 404, 165-172.
(36) Boulanger, N.; Matile, H.; Betschart, B. Acta Trop. 1988, 45, 55-
65.
(37) Kraulis, P. J. J. Appl. Cryst. 1991, 24, 946-950.