Macrocyclic inhibitors of the malarial aspartic proteases plasmepsin I, II, and IV

Article abstract of DOI:10.1016/j.bmc.2005.11.003

The first macrocyclic inhibitor of the Plasmodium falciparum aspartic proteases plasmepsin I, II, and IV with considerable selectivity over the human aspartic protease cathepsin D has been identified. A series of macrocyclic compounds were designed and synthesized. Cyclizations were accomplished using ring-closing metathesis with the second generation Grubbs catalyst. These compounds contain either a 13-membered or a 16-membered macrocycle and incorporate a 1,2-dihydroxyethylene as transition state mimicking unit. The binding mode of this new class of compounds was predicted with automated docking and molecular dynamics simulations, with an estimation of the binding affinities through the linear interaction energy (LIE) method.

Full text of DOI:10.1016/j.bmc.2005.11.003

Bioorganic & Medicinal Chemistry 14 (2006) 2197–2208
Macrocyclic inhibitors of the malarial aspartic proteases
plasmepsin I, II, and IV
a
b
b
c
´
´
Karolina Ersmark, Martin Nervall, Hugo Gutierrez-de-Teran, Elizabeth Hamelink,
Linda K. Janka,d Jose C. Clemente,d Ben M. Dunn,d Adolf Gogoll,e Bertil Samuelsson,c
b
a,
*
˚
Johan Aqvist and Anders Hallberg
aDepartment of Medicinal Chemistry, Uppsala University, BMC, PO Box 574, SE-751 23 Uppsala, Sweden
bDepartment of Cell and Molecular Biology, Uppsala University, BMC, PO Box 596, SE-751 24 Uppsala, Sweden
cMedivir AB, Lunastigen 7, SE-141 44 Huddinge, Sweden
dDepartment of Biochemistry and Molecular Biology, University of Florida College of Medicine,
BMC, PO Box 100245 Gainesville, FL 32610-0245, USA
eDepartment of Organic Chemistry, Uppsala University, BMC, PO Box 599, SE-751 24 Uppsala, Sweden
Received 1 September 2005; accepted 2 November 2005
Available online 22 November 2005
Abstract—The first macrocyclic inhibitor of the Plasmodium falciparum aspartic proteases plasmepsin I, II, and IV with
considerable selectivity over the human aspartic protease cathepsin D has been identified. A series of macrocyclic compounds
were designed and synthesized. Cyclizations were accomplished using ring-closing metathesis with the second generation
Grubbs catalyst. These compounds contain either a 13-membered or a 16-membered macrocycle and incorporate a 1,2-dihydr-
oxyethylene as transition state mimicking unit. The binding mode of this new class of compounds was predicted with
automated docking and molecular dynamics simulations, with an estimation of the binding affinities through the linear
interaction energy (LIE) method.
ꢀ 2005 Elsevier Ltd. All rights reserved.
1. Introduction
and animal models.5,6 The precise role of each of
the ten plasmepsins found in the P. falciparum gen-
ome is not yet fully understood.4 However, four of
them, plasmepsin (Plm) I, II, IV, and the closely
related histo-aspartic protease (HAP), have been
demonstrated to be active in the food vacuolar
hemoglobin catabolism.7 Of these, the first character-
ized, Plm I and II, have so far been most extensively
investigated as targets for inhibitor design.6,8 Recent
studies employing parasite knockout clones suggest
that efficient antiparasitic agents should target not
only one but several of the food vacuole
plasmepsins.9,10
Malaria has in recent years re-emerged as one of the
most serious infectious diseases in the world. This is
primarily due to the rapid spread of parasite drug
resistance, in particular among the most lethal spe-
cies affecting humans, Plasmodium falciparum.1 Each
year approximately 300–500 million clinical cases
are reported and 1–3 million people, mostly children
in Africa, die as a result of the disease.2 Several at-
tempts are being made to develop new antimalarial
drugs with novel mechanisms of action.3 The malar-
ial aspartic proteases, termed plasmepsins, have been
identified as one class of suitable targets.4 Proof of
concept that these enzymes are essential for parasite
survival has been provided both from cell culture
We have previously identified linear Plm I and II inhib-
itors, which incorporate a mannitol-derived 1,2-dihydr-
oxyethylene unit, as
a common transition state
isostere.11–13 In this series of compounds, a number of
nanomolar activity inhibitors with a considerable dis-
crimination versus the most homologous human aspar-
tic protease cathepsin D (Cat D) as a characteristic
feature were discovered.
Keywords: Malaria; Plasmepsin; Protease inhibitors and molecular
dynamics.
*
Corresponding author. Tel.: +46 18 471 42 84; fax: +46 18 471 44
74; e-mail: anders.hallberg@orgfarm.uu.se
0968-0896/$ - see front matter ꢀ 2005 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmc.2005.11.003

2198
K. Ersmark et al. / Bioorg. Med. Chem. 14 (2006) 2197–2208
Aspartic proteases usually bind substrates/inhibitors
adopting an extended b-strand conformation.14 By en-
abling preorganization of the inhibitor in a bioactive
conformation, a higher affinity should be achieved due
to a reduced loss of entropy upon binding.14 Macrocyc-
lization has previously proven to be an effective con-
straining device in aspartic protease inhibitors.15–19
Additionally, constrained compounds with amide bonds
demonstrate a higher resistance to proteolytic cleavage,
and macrocyclic compounds often exhibit improved cell
permeability compared to their acyclic counterparts.16,20
To the best of our knowledge, only two plasmepsin
inhibitors with a macrocyclic structural element have
been reported.19,21 A statine-based inhibitor containing
a bridge between the P1 and P3 side chains that forms
a 15-membered macrocycle exhibited a high potency
against Plm II (1, Fig. 1). On the contrary, an inhibitor
with a P2-P30 20-membered macrocycle originating from
the same type of statin scaffold lost potency for Plm II
(2, Fig. 1).19 However, more recently the latter inhibitor
was shown to inhibit Plm IV efficiently.21 Unfortunate-
ly, none of these macrocyclic inhibitors was selective
over the human Cat D.
a 16-membered ring with a P1–P20/P2–P10 connection.
The smaller 13-membered rings were designed to be
accommodated in the unobstructed S1–S3 cleft, while
the 16-membered rings were included to address the pre-
vious observation that a large cycle spanning over both
the prime and non-prime sides is well accommodated
in Plm IV but is far less potent against Plm II.19,21 The
binding features of the new macrocyclic inhibitors were
investigated by molecular modeling.
Herein we report that RCM with the second generation
Grubbs catalyst delivers the first macrocyclic plasmepsin
inhibitors with significant selectivity over Cat D.
2. Results
2.1. Chemistry
Scheme 1 outlines the synthetic route to the unsaturated
and saturated macrocyclic target compounds 5–8. The
intermediate compounds 12 and 13 were synthesized
essentially following a previously reported strategy.22
Alkylation of L-mannaro-1,4:3,6-di-c-lactone was
accomplished using simultaneous addition of 1.5 equiv
benzyl trichloroacetimidate and 1.5 equiv allyl trichloro-
acetimidate yielding approximately a 1:1:1 mixture of
the diallylated11 and the dibenzylated23 bislactones
together with the unsymmetrically allylated and benzy-
lated bislactone, 9. Mono-opening of bislactone 9 was
accomplished with (1S,2R)-1-amino-2-indanol using 2-
hydroxypyridine as catalyst to give the mono-lactones
10 and 11, which could not be completely separated.
No obvious preference in the opening of the two lac-
tones of 9 was observed. Approximately a 1:1 ratio of
compounds 10 and 11 was formed as deduced from
LC–MS. Finally, the second lactone ring was opened
using excess octenamine (14) to afford a mixture of the
two intermediate compounds 12 and 13 in 73% yield.
As a consequence of 12 and 13 being inseparable on
the LC systems available, these isomers were used as a
mixture in the subsequent RCM. A second generation
Grubbs catalyst24 was employed in the RCM, which
was conducted in 1,2-dichloroethane at reflux. The four
resulting macrocycle diastereomers, i.e., the E and Z iso-
We felt encouraged to investigate whether non-prime side
or prime/non-prime side macrocyclizations of the previ-
ously reported prototype C2-symmetric inhibitors encom-
passing amino indanol structures in both the P2 and P20
positions would deliver, first of all, inhibitors of Plm I,
II, or IV, and secondly, inhibitors of these proteases that
were devoid of Cat D affinity. Compounds 3 and 4 in
Table 1 were used as starting structures and ring-closing
metathesis (RCM) rather than amide bond formation
was employed to avoid incorporation of a new hydrogen
bond-accepting/donating amide in the macrocycle. Two
ring sizes were explored: a 13-membered ring connecting
the P1 and P2 positions or the P10 and P20 positions,
depending on the binding mode to the plasmepsin, and
O
N
H
O
O
O
O
H
N
H
N
H
N
1
N
H
N
H
N
H
NH2
mers of 5 and 7, were separated by LC. H NMR was
O
OH
O
O
used to determine the E:Z ratios. To identify the 13-
membered macrocycle (5) and the 16-membered macro-
cycle (7), C–H connectivity information from HMBC
NMR experiments was used. Thus, starting with H-1
(Scheme 1), identified by its connectivity to the hydrox-
yindan moiety, H-3 can be located via C–H long-range
couplings. Proton H-1 (denoted in 5 and 7, Scheme 1)
has a long-range coupling to the carbonyl carbon 2. This
carbon couples with proton 3, which in turn has a long-
range coupling to either carbon 4. Depending on macro-
cycle, this carbon has either long-range couplings to aro-
matic protons (5) or to olefinic protons (7). The E- and
Z-configurations of the macrocycle double bond were
assigned based on 3J couplings between the olefinic pro-
tons (15.3 Hz for E- and 10.3 Hz for Z-configuration),
and by the detection of NOEs from CH2–CH@CH
protons to either both (for E) or only one (for Z) of
1. Ki Plm II, 0.2 nM; Ki Cat D, 0.26 nM.19
O
O
O
O
H
N
H
N
H
N
N
H
N
H
N
H
NH2
O
OH
O
O
HN
O
2. Ki Plm II, 1500 nM;19 Ki Plm IV, 0.52 nM;21
Ki Cat D, 1.4 nM.19
Figure 1. Macrocyclic plasmepsin inhibitors previously reported in the
literature.

K. Ersmark et al. / Bioorg. Med. Chem. 14 (2006) 2197–2208
Table 1. Plasmepsin inhibitory activity of the prepared macrocyclic compounds
2199
Compound
Structure
Enzyme Ki (nM)
Plm I
Plm II
Plm IV
Cat D
Exp
Calcd
Exp
Calcd
OH
O
OH
O
H
3
160a
96b
240b
>5000c
nd
>2000a
N
N
H
O
OH O
OH
OH
O
OH
O
H
4
25a
85a
99a
nd
nd
>2000a
N
N
H
O
OH O
OH
OH
O
O
OH
H
N
(E)-5
500
380
140
6800
25 · 104
>5900
>5900
>5900
>5900
N
H
O
OH O
OH
O
O
OH
H
N
6
180
120
260
200
60
N
H
O
OH O
OH
O
OH
O
H
N
N
H
(E)-7
>2900
1100
2970
2000
2910
O
OH O
OH
O
OH
O
H
N
N
H
8
>2900
>4800
2910
2200
1460
O
OH O
nd, not determined.
a Data from Ref. 11.
b Data from Ref. 12.
c Data from Ref. 13.
the olefinic protons. Furthermore, Z isomers showed an
NOE between allylic protons on either side of the double
bond. The saturated macrocycles 6 and 8 were finally
obtained by hydrogenation on Pd/C.
2.2. Enzyme inhibition
Macrocycles 5–8 were evaluated for their ability to
inhibit the parasite Plm I, II, IV, and the human

2200
K. Ersmark et al. / Bioorg. Med. Chem. 14 (2006) 2197–2208
OH
O
H
OH
O
O
N
HO
O
a
b
O
O
c
O
+
O
O
O
O
O
O
O
OH
O
O
O
N
H
O
HO
O
O
HO
10
11
9
OH
OH
O
OH
O
O
OH
O
H
N
H
N
+
N
N
H
H
O
OH O
O
OH O
13
12
d
4
4
OH
OH
O
O
OH
O
OH O
H
H
1
2
2
N
N
+
3
3
N
N
1
H
H
O
OH
O
O
OH
O
7, mixture of E:Z (~5:3 ratio)
5, mixture of E:Z (~19:1 ratio)
e
e
OH
OH
O
O
OH
OH
O
O
H
H
N
N
N
N
H
H
O
OH
O
O
OH O
6
8
Scheme 1. Reagents and conditions: (a) 1.5 equiv benzyl trichloroacetimidate, 1.5 equiv allyl trichloroacetimidate, BF3 · Et2O, dioxane; (b) (1S,2R)-
1-amino-2-indanol, 2-hydroxypyridine, CH2Cl2; (c) 7-octenamine (14), CH2Cl2, reflux; (d) Grubbs catalyst second generation,24 1,2-dichloroethane,
reflux; (e) H2, Pd/C, MeOH.
Cat D. The results from the enzyme assays are presented
as Ki values in Table 1.
The 13-membered macrocycles 5 and 6 were generally
much more active against the plasmepsins than the larg-
er 16-membered macrocycles 7 and 8 with the ring span-
ning from the prime to the non-prime side. In fact, the
16-membered macrocycles were devoid of any inhibitory
potency in the Plm I assay, and only the unsaturated 7
exhibited any potency in the Plm II assay. Overall, the
Compounds 3 and 4, used as starting points in the de-
sign, were included for comparison.11,12 Since sufficient
amounts of the pure Z-5 and Z-7 isomers could not be
isolated, only the E isomers were assessed.

K. Ersmark et al. / Bioorg. Med. Chem. 14 (2006) 2197–2208
2201
affinities toward Plm IV were low, except for the saturat-
ed 13-membered macrocycle 6. The 13-membered satu-
rated inhibitor 6 was significantly more potent than
the unsaturated 5 in all three Plm I, II, and IV assays.
In fact, compound 6 was the most potent inhibitor iden-
tified in this series, with Ki values of 180, 120, and
200 nM in the Plm I, II, and IV assays, respectively.
Notably, none of the macrocycles (5–8) was active in
the Cat D assay.
S2, in such a way that the amide bond present in the ring
can interact with both Gly216 and the flap loop Ser79.
The conformation of the 16-membered ring forces the
two hydroxyls from the 1,2-dihydroxyethylene in oppo-
site directions thereby allowing only the hydroxyl group
on the non-prime side to form hydrogen bonds with the
catalytic aspartic residues. The other hydroxyl group is
hence strained to interact with the flap loop without
the ability to establish any stable H-bond network. This
arrangement of the hydroxyls stands out against our
previously reported model of how the hydroxyl groups
are oriented when they are not constrained in a macro-
cyclic system.11,13 Further, the indanol moiety is posi-
tioned in the S20 pocket and the benzyloxy side chain
is accommodated in the S1–S3 pocket. In all cases, the
docking runs converged to the aforementioned position.
The stability of this binding mode for compounds (E)-7
and 8 was confirmed by MD simulations. We observed
frequent hydrogen bonding between the non-prime
hydroxyl and the aspartates, as well as for the amide
bond (with Gly36) and the hydroxyl group of the inda-
nol moiety (with Thr192) on the prime side. The prime
hydroxyl, which is pointing to the flap loop, is also in-
volved in an internal hydrogen bond with the oxygen
of the amide integrated in the macrocycle. This amide
2.3. Computational modeling of inhibitor binding
Compounds 5–8 were investigated by computational
procedures in order to elucidate the binding mode and
to gain some insight into the molecular determinants
of ligand binding. A docking exploration for each com-
pound, in both Plm II and Plm IV enzymes, was followed
by molecular dynamics (MD) coupled to an estimation
of the free energy of binding with the linear interaction
energy (LIE) method25 for selected docking poses.
The docking exploration showed one consistent solution
for (E)-7 and 8, the 16-membered macrocycles, in both
enzymes investigated (see Figs. 2A and B). The solution
positions the 16-membered ring stretching from S10 to
Figure 2. Binding modes of the macrocyclic compounds shown as snapshots from MD-simulations. (A and B) The 16-membered macrocyclic
compound 7 in complex with Plm II and Plm IV, respectively. Analogously, (C and D) represent the 13-membered macrocycle 6 in complex with
Plm II and Plm IV, respectively.

2202
K. Ersmark et al. / Bioorg. Med. Chem. 14 (2006) 2197–2208
interacts with Thr217 through the NH group, after some
rearrangement of the ring. In all four cases (compounds
(E)-7 and 8 with both Plm II and Plm IV, see Table 2),
the moderate binding predicted is in excellent agreement
with the experimental results.
For the 13-membered macrocycles, compounds (E)-5
and 6, the docking solutions were less consistent. In
the case of Plm II, the outcome of the docking was
reasonably convergent but for Plm IV there emerged
several solutions. As shown in Figure 3, the proposed
docking poses placed the macrocycle both in the
S1–S3 binding site and in the S10–S20 site, while the
docking position of the amino indanol and benzyloxy
substituents was also not converged. Subsequently, only
those solutions that showed consistency between the two
enzymes were selected for further investigations. How-
ever, alternate binding modes from the Plm IV docking
results, shown in Figure 3, were also examined by molec-
ular dynamics simulations, which showed that none of
them were structurally stable and that they yielded unfa-
vorable free energies of binding (data not shown). The
selected binding mode, depicted in Figures 2C and D,
was consistent in the two enzymes and placed the 13-
membered cycles of (E)-5 and 6 in the S1–S3 cleft. The
two central hydroxyls were positioned at hydrogen-
bond distances to the catalytic aspartic acids in a man-
ner consistent with previous studies.11–13 The amino
indanol group was found in the S20 site, analogously
to what was found for compounds (E)-7 and 8. Finally,
the benzyloxy side chain was in this case placed into the
S10 pocket as opposed to (E)-7 and 8, where the
benzyloxy substituent was placed in S1. Molecular
dynamics showed high stability for the interactions
involved in the aforementioned position for both (E)-5
and 6 when bound to Plm II. The same stability was
found as well for the saturated compound 6 in Plm
IV, with binding free energy estimates in excellent agree-
ment with experimental measurements (see Table 2).
However, for the unsaturated compound (E)-5, the
binding conformation adopted in Plm IV by the more
rigid macrocycle forced the hydroxyl groups to be
positioned in a non-optimal geometry for the hydrogen
bonding to the catalytic aspartates. This caused a
dramatic decrease in the predicted affinity, an observa-
tion that is in agreement with the experimental evidences
although somewhat exaggerated.
Figure 3. Alternative docking poses found for compound 6 into Plm IV
by GOLD. The pose selected for further evaluations, here shown in thick
yellow sticks, was also obtained in Plm II docking experiments.
3. Discussion
Two of our previously reported linear inhibitors, 3 and
4, originating from mannitol and containing a 1,2-
dihydroxyethylene motif, are shown in Table 1.12 Within
this class of inhibitors several compounds with high
affinity for both Plm I and II, and with a remarkable
selectivity over the human Cat D were identified.12 By
utilizing the ring-closing metathesis (RCM) methodolo-
gy these inhibitors were constrained, providing either a
13-membered (unsaturated 5 and saturated 6) or a 16-
membered (unsaturated 7 and saturated 8) macrocycle.
As compared to 3 and 4, the new compounds are lacking
one of the two indanol groups, and in relation to 4 one
of the benzyloxy side chains. As can be seen in Table 1,
the smaller 13-membered macrocycles (E)-5 and 6 were
in general significantly more potent against Plm I, II,
and IV than the 16-membered cycles (E)-7 and 8 bridg-
ing the prime to the non-prime side. According to previ-
ous observations,19,21 this effect might be attributed to
the variation on position 78 at the tip of the flap; Plm
I and II both possessing a sterically demanding valine,
which in Plm IV is replaced by a glycine. In addition,
it was evident that the constrained macrocycles imposed
Table 2. Experimental and calculated free energies of binding for compounds 5–8 in both Plm II and Plm IV
Enzyme
Compound
DGbind, exp(kcal/mol)a
DGbind, calcd (kcal/mol)
Ligand–surrounding interactions (kcal/mol)b
vdW
l–s
el
vdW
l–s
el
hV iw
l–s
hV
ip
hV ip
lꢀs
hV
iw
Plm II
(E)-5
6
ꢀ8.8
ꢀ9.5
ꢀ8.2
>ꢀ7.3
ꢀ7.1
ꢀ9.2
ꢀ7.8
ꢀ7.8
ꢀ9.4 1.2
ꢀ9.1 0.8
ꢀ7.6 1.4
ꢀ7.6 1.0
ꢀ4.9 0.7
ꢀ9.9 1.3
ꢀ7.6 1.6
ꢀ8.0 0.6
ꢀ66.3 0.3
ꢀ64.0 0.2
ꢀ64.7 2.9
ꢀ65.7 0.3
ꢀ72.1 0.9
ꢀ74.8 0.3
ꢀ72.5 0.8
ꢀ75.8 0.6
ꢀ84.3 1.8
ꢀ85.3 0.5
ꢀ83.5 0.6
ꢀ80.1 1.7
ꢀ67.6 0.1
ꢀ82.0 2.1
ꢀ79.3 2.3
ꢀ75.8 0.6
ꢀ42.7 0.2
ꢀ43.7 0.6
ꢀ43.8 0.3
ꢀ43.7 0.2
ꢀ42.7 0.2
ꢀ43.7 0.6
ꢀ43.8 0.3
ꢀ43.7 0.2
ꢀ68.7 1.5
ꢀ69.0 1.4
ꢀ72.0 1.9
ꢀ69.1 1.0
ꢀ68.7 1.5
ꢀ69.0 1.4
ꢀ72.0 1.9
ꢀ69.1 1.0
(E)-7
8
Plm IV
(E)-5
6
(E)-7
8
a The experimental binding free energy calculated from experimentally determined Kiꢀs using DGb0ind;exp ¼ RT ln Ki.
b The calculated average electrostatic hVeli and non-polar hVvdWi energies for ligand–surrounding (l–s) interactions. The subscripts p and w denote
simulations of the ligand in complex with the protein and free in water, respectively.

K. Ersmark et al. / Bioorg. Med. Chem. 14 (2006) 2197–2208
2203
an unfavorable position of the crucial hydroxyl groups
in the central transition state mimicking core unit. The
saturated 13-atom macrocycle 6 constitutes the most po-
tent plasmepsin inhibitor obtained in the series. This
inhibitor was about three times as active against Plm I
and II, and over 30 times as active against Plm IV as
the corresponding unsaturated inhibitor 5, possibly
reflecting an increased flexibility upon saturation
enabling a proper adaptation of the compound in the
enzyme active site. Additionally, inhibitor 6 was over
25-fold more potent against Plm IV and almost
equipotent against Plm I and II compared to the parent
linear inhibitor 3. Considering that the P2 side chain is
omitted from the 13-membered inhibitors and that a
considerable selectivity versus the human Cat D could
be achieved, this suggests macrocyclization as a viable
strategy for further optimization.
solutions for the saturated and unsaturated compounds
are almost identical, after some MD we observed a con-
formational rearrangement of the unsaturated com-
pound. The new, slightly modified position remained
stable for the rest of the data collection trajectory and
had a pronounced effect on the interface between the
inhibitor and the enzyme close to the catalytic residues,
as shown in Figure 4. Compared to Plm II, the S1–S3
pocket in Plm IV has more restricted geometrical
bounds, identified by the mutation at position 114 from
Thr in Plm II to the more bulky Ile in Plm IV. For the
saturated compound we did not observe significant
differences between the two enzymes, indicating that a
flexible ring is more likely to adapt to the geometrical
restraints of the enzyme.
The initial docking results of the 16-membered ring
compounds initially pointed to the hypothesis that
Val78 in Plm II constitutes a larger steric obstacle than
Gly78 in Plm IV. However, after the equilibration peri-
od in the MD simulations, both binding sites seem to
accommodate the macrocycle equally well, as shown in
Figures 2A and B. In fact, the loss in affinity due to
the S10–S2 macrocyclization is equivalent in both en-
zymes. The initially more closed surface of Plm II is
compensated by a relaxation of the backbone of the flap
loop, and an almost perfect superposition of the ligand
can be observed if the two complexes are compared. It
should be noted, however, that the shape complementar-
ity is not optimal in either of the enzymes for (E)-7 and
8, leading to structural fluctuations of the macrocyclic
rings. We even observe some partial flap loop opening
during our simulations due to such fluctuations. A sec-
ond common factor responsible for the weaker binding
of the larger cycles is found in the orientation of the
1,2-hydroxyethylene hydroxyls enforced by the confor-
mational restraints of the macrocycle. Only one of the
two core hydroxyl groups is able to interact with the cat-
alytic aspartic acid residues, while the other is left to
interact with the loop and to make an internal hydrogen
bond with an amide bond of the macrocycle. This inter-
action profile appears less favorable compared to the
Docking experiments with Plm II and Plm IV indicated
that the 16-membered macrocycles in (E)-7 and 8 are
aligned along the S10–S2 subsites, while the 13-mem-
bered cycles of (E)-5 and 6 are located in the S1–S3 cleft.
The feasibility of these docking poses was confirmed by
binding free energy estimations using the linear interac-
tion energy approach, obtaining a reasonable agreement
between calculated and experimental binding constants.
It is interesting to note that the introduction of the dou-
ble bond in the 13 member macrocycle (compounds (E)-
5 and 6) leads to a 30-fold drop in affinity for Plm IV, a
trend that is reproduced in our computational model.
The common hypothesis that rigid analogues, through
a lower loss of conformational entropy upon binding,
have a free energy advantage relative to their flexible
counterparts26 is not obeyed in this case. On the con-
trary, our simulations show that the double bond pres-
ent in compound (E)-5 leads to a non-optimal match
to the Plm IV pocket compared to the more flexible
inhibitor 6, as highlighted in Figure 4. The difference be-
tween the two related compounds, although overesti-
mated by the binding affinity calculations, originates
from an unfavorable arrangement of the E isomer ring
in the S1–S3 pocket (see Fig. 4). Even if the docking
Figure 4. Corss-eyed stereo view of (E)-5 (light blue) and 6 (magenta) in complex with Plm IV. Relevant snapshots were extracted from MD
trajectories. Only residues in close contact with the ligands are depicted, and water molecules have been removed. The conformation adopted by the
cycle upon binding is not the same in the two compounds, leading to different geometry of the protein–ligand interactions.

2204
K. Ersmark et al. / Bioorg. Med. Chem. 14 (2006) 2197–2208
hydrogen bonding pattern observed in the 13-membered
ring compounds, consistent with previously reported
inhibitors.11–13
(4.6 · 100 mm)) column, with a Finnigan AQA quadro-
pole mass spectrometer, at a flow rate of 4 mL/min,
using different gradients of CH3CN in 0.05% aqueous
formic acid. Preparative RP-LC–MS was performed
on a Gilson HPLC system with either a C8 (Agilent
Technologies Zorbax SB-C8 (5 lm, 21.2 · 150 mm)) col-
umn or a C18 (YMC ODS-AQ (5 lm, 20 · 50 mm)) col-
The previously reported inhibitors comprising a macro-
cycle element did not exhibit any selectivity for the
plasmepsins over the human aspartic protease Cat
D.19 Since the linear 1,2-dihydroxyethylene-based inhib-
itors have shown a considerable selectivity over Cat D,
they appeared to provide good starting points in the
search for selective macrocyclic inhibitors. In agreement
with this hypothesis, none of the macrocyclic 1,2-dihydr-
oxyethylene compounds inhibited Cat D.
umn, with
a
Finnigan AQA quadropole mass
spectrometer, at a flow rate of 15 mL/min, using differ-
ent gradients of CH3CN in 0.05% aqueous formic acid.
5.1.2. 2-O-Allyl-5-O-benzyl-L-mannaro-1,4:3,6-di-c-lac-
tone (9). L-1,4:3,6-Mannarodilactone (1.0 g, 5.7 mmol)
was dissolved in dry dioxane (130 mL) under a nitrogen
atmosphere. To the stirred solution were added benzyl
trichloroacetimidate (2.2 g, 8.6 mmol), allyl trichloro-
acetimidate (1.8 g, 8.6 mmol), and a catalytic amount
of BF3 · OEt2 (590 lL). The solution was stirred at
room temperature for 3 h. After filtration through a sil-
ica pad, the filtrate was concentrated under reduced
pressure. The crude product was washed with diethyl
ether (2 · 30 mL) to give a mixture of 9 together with
the two dialkylated molecules 2,5-di-O-allyl-L-manna-
ro-1,4:3,6-di-c-lactone and 2,5-di-O-benzyl-L-mannaro-
1,4:3,6-di-c-lactone (in a total of 1.2 g). In general, this
mixture was used in the next step without further purifi-
cation. A small portion (100 mg) was purified on RP-LC
(C8 column, 30 min gradient of 35-80% CH3CN in
4. Conclusion
Inhibitors encompassing 13-membered and 16-mem-
bered macrocyclic ring structures and that inhibit P. fal-
ciparum Plm I, II, and IV have been synthesized.
Cyclization was accomplished by means of RCM using
Grubbs second generation catalyst. The binding mode
of this new class of compounds was predicted by molec-
ular modeling, leading to a consistent interpretation of
the experimental results. One saturated 13-membered
macrocyclic compound (6) was found with an equally
high affinity for all three plasmepsins (Plm I, II, and
IV). This compound exerted no activity in the Cat D as-
say. To the best of our knowledge, these compounds are
the first macrocyclic plasmepsin inhibitors that exhibit a
high selectivity for the plasmepsins over the human
Cat D.
0.05% aqueous formic acid) to yield pure compound 9
19
D
(29 mg), which was analyzed: ½aꢁ ꢀ159.8 (c 1.00,
1
CH3CN); H NMR (399.8 MHz; CD3OD/CDCl3, 2:7)
d 7.38–7.25 (m, 5H), 5.89 (ddt, J = 17.2, 10.3, 6.1, Hz,
1H), 5.32 (dm, J = 17.2 Hz, 1H), 5.25 (dm,
J = 10.3 Hz, 1H), 4.98 (dd, J = 3.2, 4.7 Hz, 1H), 4.89
(dd, J = 3.2, 4.8 Hz, 1H), 4.84 (d, J = 12.0 Hz, 1H),
4.79 (d, J = 12.0 Hz, 1H), 4.39 (d, J = 4.7 Hz, 1H),
4.36 (d, J = 4.8 Hz, 1H), 4.33–4.19 (m, 2H); 13C NMR
(100.5 MHz; CD3OD/CDCl3, 2:7) d 170.8 (2C, accord-
ing to HMBC), 135.7, 132.7, 128.6, 128.5, 128.3, 119.7,
74.0, 73.92, 73.89, 73.8, 72.9, 72.2. Anal. (C16H16O6)
C, H.
5. Experimental
5.1. Chemistry
5.1.1. General information. NMR spectra were recorded
on a JEOL JNM-EX400 spectrometer (1H at 399.8 MHz
and 13C at 100.5 MHz), a Varian UNITY INOVA spec-
trometer (1H at 499.9 MHz), a Varian UNITY spec-
trometer (1H at 399.5 MHz) or on a Varian Mercury
Plus spectrometer (1H at 300.0 MHz, 13C at
75.4 MHz). Chemical shifts are reported as d values
(ppm) indirectly referenced to TMS via the solvent resid-
ual signal. Two-dimensional (2D) HMBC spectra were
recorded on a Varian UNITY INOVA spectrometer
(1H at 499.9 MHz) or on a Varian Mercury Plus spec-
trometer (1H at 300.0 MHz) using standard pulse
sequences. NOE effects were recorded on a Varian UNI-
TY INOVA spectrometer (1H at 499.9 MHz) or on a
Varian UNITY spectrometer (1H at 399.5 MHz) using
a NOESY experiment with a mixing time of 0.8 s. Opti-
cal rotations were obtained on a Perkin-Elmer 241
polarimeter. Specific rotations ([a]D) are reported in
deg/dm, and the concentration (c) is given in g/100 mL
in the specified solvent. Elemental analyses were per-
formed by Analytische Laboratorien, Lindlar, Germa-
ny. Flash chromatography was performed on Merck
silica gel 60, 0.04–0.063 mm. Analytical RP-LC–MS
was performed on a Gilson HPLC system with a C18
(Merck KgaA Chromolith Performance RP-18e
5.1.3. 7-Octenamine (14). 8-Bromo-1-octene (2.5 g,
13.1 mmol) and sodium azide (936 mg, 14.4 mmol) were
dissolved in EtOH (16 mL) and stirred at reflux for 16 h.
The solvent was removed under reduced pressure. The
crude product was partitioned between CH2Cl2
(100 mL) and H2O (100 mL). The aqueous layer was
extracted with CH2Cl2 (2 · 100 mL) and the combined
organic layers were dried (Na2SO4) and concentrated.
A solution of the crude azide (2.0 g, 13.1 mmol) in dry
diethyl ether (30 mL) was added drop-wise to a stirred
mixture of LiAlH4 (993 mg, 26.2 mmol) in dry diethyl
ether (50 mL) at 0 ꢁC. The mixture was allowed to attain
room temperature, stirred for 3 h, and then quenched by
the addition of H2O (2 mL), 15% NaOH (2 mL), and
more H2O (6 mL), followed by filtration. The mixture
was diluted with H2O (80 mL) and diethyl ether
(30 mL). The organic layer was separated and the aque-
ous layer was extracted with diethyl ether (2 · 100 mL)
and CH2Cl2 (3 · 100 mL). The combined organic layers
were dried (Na2SO4) and concentrated to yield the
amine 14 (1.2 g, 72%).27

K. Ersmark et al. / Bioorg. Med. Chem. 14 (2006) 2197–2208
2205
5.1.4. Mixture of (2R)-1N-(2-hydroxyindan-1-yl)-2-benzyl-
oxy-2-[(2R,3R,4R)-4-(2-propenyloxy)-3-hydroxy-5-oxo-tet-
rahydrofuran-2-yl]-ethanamide (10) and (2R)-1N-(2-
hydroxyindan-1-yl)-2-(2-propenyloxy)-2-[(2R,3R,4R)-
4-benzyloxy-3-hydroxy-5-oxo-tetrahydrofuran-2-yl]-ethana-
mide (11). The bislactone mixture of 9 together with 2,
5-di-O-allyl-L-mannaro-1,4:3,6-di-c-lactone and 2,5-di-
O-benzyl-L-mannaro-1,4:3,6-di-c-lactone) (2.0 g) and
2-hydroxypyridine (540 mg, 5.7 mmol) were dissolved
in dry CH2Cl2 (140 mL) and stirred at room tempera-
ture for 30 min. The solution was cooled to 0 ꢁC and
(1S,2R)-1-amino-2-indanol (846 mg, 5.7 mmol) was
added rapidly. The reaction mixture was allowed to
attain room temperature and stirred overnight for
12 h. The solvent was removed under reduced pressure
and subsequent flash chromatography (CHCl3) gave a
mixture of the two variants of mono-opened bislac-
tones 10 and 11 (501 mg). As 10 and 11 could not
be completely separated from each other, the mixture
was used in the next step without further purification.
5.1.7. (2R,3R,4R)-2-Benzyloxy-3,4-dihydroxy-N-[(1S,2R)-
2-hydroxyindan-1-yl]-4-[(2R,11E)-3-oxo-1-oxa-4-aza-cyclo-
18
D
tridec-11-en-2-yl]-butyramide ((E)-5). ½aꢁ ꢀ24.8 (c 0.79,
1
CHCl3); H NMR (499.9 MHz; acetone-d6/C6D6, 2:1) d
7.85 (d, J = 8.7 Hz, 1H), 7.46–7.25 (m, 9 H), 5.84 (dddm,
J = 15.3, 7.2, 7.2 Hz, 1H), 5.73 (dddm, J = 7.0, 7.0,
15.3 Hz, 1H), 5.57 (dd, J = 5.1, 8.7 Hz, 1H), 5.24 (s, 1H),
5.98 (s, 1H), 4.84 (d, J = 11.4 Hz, 1H), 4.80 (d,
J = 11.4 Hz, 1H), 4.73 (ddm, J = 5.1, 5.1 Hz, 1H), 4.52 (s,
1H), 4.45 (d, J = 7.3 Hz, 1H), 4.35 (ddm, J = 12.6, 7.2 Hz,
1H), 4.29 (dm, J = 7.3 Hz, 1H), 4.22 (dm, J = 7.7 Hz, 1H),
4.13 (d, J = 7.7 Hz, 1H), 4.06 (ddm, J = 12.6, 7.2 Hz, 1H),
3.67 (m, 1H), 3.17 (dd, J = 5.1, 16.2 Hz, 1H), 3.08–2.98
(m, 2H), 2.20–2.08 (m, 2H), 1.77 (m, 1H), 1.56–1.34 (m,
7H). A NOESY (499.9 MHz) experiment confirmed the E
geometry of the double bond, and an HMBC
(499.9 MHz) experiment confirmed the size of the macrocy-
cle; 13C NMR (75.4 MHz; CDCl3) d 173.0, 172.7, 141.0,
140.1, 138.6, 137.0, 128.9, 128.6, 128.55, 128.50, 127.3,
126.3, 125.6, 124.3, 81.4, 80.5, 74.1, 73.9, 72.9, 71.9, 71.8,
58.2, 39.6, 38.4, 31.3, 26.3, 25.9, 25.3, 23.6. Anal
(C31H42N2O7 Æ 1/3H2O) C, H, N.
5.1.5. Mixture of (2R,3R,4R,5R)-2-benzyloxy-3,4-dihy-
droxy-N1-[(1S,2R)-2-hydroxyindan-1-yl]-N6-(7-octene)-
5-(2-propenyloxy)-hexanamide (12) and (2R,3R,4R,5R)-
5-benzyloxy-3,4-dihydroxy-N1-[(1S,2R)-2-hydroxyindan-1-
yl]-N6-(7-octene)-2-(2-propenyloxy)-hexanamide (13). The
mixture of the two variants of mono-opened bislactone
10 and 11 (167 mg, 0.37 mmol) was dissolved in dry
CH2Cl2 (4 mL) and 7-octenamine (235 mg, 1.8 mmol)
was added. The reaction mixture was stirred at reflux
for 3 h. The solvent was removed under reduced pressure
and subsequent purification by RP-LC–MS (C8 column,
30 min gradient of 30–90% CH3CN in 0.05% aqueous for-
mic acid) gave a mixture of the two diamides 12 and 13
(157 mg, 73%). As 12 and 13 could not be separated from
each other, the mixture was used in the next step without
further purification.
5.1.8. (2R,3R,4R)-2-Benzyloxy-3,4-dihydroxy-N-[(1S,2R)-
2-hydroxyindan-1-yl]-4-[(2R,11Z)-3-oxo-1-oxa-4-aza-cyclo-
tridec-11-en-2-yl]-butyramide ((Z)-5). 1H NMR (499.9
MHz; CD3OD/CDCl3, 3:1) d 7.40–7.14 (m, 9H), 5.57
(dm, J = 10.3 Hz, 1H), 5.47 (dm, J = 10.3 Hz, 1H), 5.36 (d,
J = 5.0 Hz, 1H), 4.67 (d, J = 11.4 Hz, 1H), 4.64 (d,
J = 11.4 Hz, 1H), 4.59 (ddd, J = 1.7, 5.0, 5.1 Hz, 1H) 4.18–
4.14 (m, 3H), 4.07 (dd, J = 2.1, 6.6 Hz, 1H), 3.98 (dd,
J = 2.1, 7.6 Hz, 1H), 3.95 (d, J = 7.6 Hz, 1H), 3.35 (m, 1H),
3.23–3.12 (m, 2H), 2.94 (dd, J = 1.7, 16.6 Hz, 1H), 2.22 (m,
1H), 2.02–1.94 (m, 1H), 1.75 (m, 1H), 1.50–1.24 (m, 7H). A
NOESY (499.9 MHz) experiment confirmed the Z geometry
of the double bond, and an HMBC (499.9 MHz) experiment
confirmed the size of the macrocycle.
5.1.6. (2R,3R,4R)-2-Benzyloxy-3,4-dihydroxy-N-[(1S,2R)-
2-hydroxyindan-1-yl]-4-[(2R)-3-oxo-1-oxa-4-aza-cyclotridec-
11-en-2-yl]-butyramide (5) and (2R,3R,4R,5R)-5-benzyloxy-
3,4-dihydroxy-6-oxo-1-oxa-7-aza-cyclohexadec-14-ene-
2-carboxylic acid [(1S,2R)-2-hydroxyindan-1-yl]-amide
(7). The mixture of the diamides 12 and 13 (58 mg,
0.10 mmol) was dissolved in dry 1,2-dichloroethane
(20 mL) and a second generation Grubbs catalyst (benzyl-
idene[1,3-bis(2,4,6-trimethylphenyl)-2-imidazolidinylid-
ene]dichloro-(tricyclohexylphosphine)ruthenium) (8.5 mg,
0.010 mmol) was added. Since the reaction was slow, addi-
tional amounts of second-generation Grubbs catalyst were
added (4.2 mg, 0.0050 mmol and 4.2 mg, 0.0050 mmol after
2 and 4 h, respectively). The reaction mixture was stirred at
reflux for 6 h. The solvent was removed under reduced
pressure and subsequent purification on RP-LC–MS
(C18 column, 30 min gradient of 25–65% CH3CN in
0.05% aqueous formic acid) gave the E- and Z-diastereo-
mers of the two different macrocycles 5 and 7 (in a total of
26.3 mg, 48%) as white solids. (E/Z)-5. 1H NMR
(499.9 MHz; CD3OD/CDCl3, 3:1) of the LC–MS purified
mixture was used to estimate the E/Z ratio of 5 to ꢂ19:1.
5.1.9. (2R,3R,4R,5R,14E)-5-Benzyloxy-3,4-dihydroxy-6-
oxo-1-oxa-7-aza-cyclohexadec-14-ene-2-carboxylic acid
18
[(1S,2R)-2-hydroxyindan-1-yl]-amide ((E)-7). ½aꢁ +11.6
D
(c 0.69, CH2Cl2); 1H NMR (499.9 MHz; CD3OD/CDCl3,
3:1) d 7.39–7.18 (m, 9H), 5.70 (dddm, J = 7.0, 7.0,
15.3 Hz, 1H), 5.58 (dddm, J = 6.0, 6.0, 15.3 Hz, 1H),
5.37 (d, J = 5.0 Hz, 1H), 4.63–4.57 (m, 3H), 4.22 (ddd,
J = 1.2, 6.0, 12.5 Hz, 1H), 4.07–4.03 (m, 2H), 4.01 (d,
J = 6.8 Hz, 1H), 3.96 (dd, J = 2.0, 6.8 Hz, 1H), 3.89
(ddd, J = 1.1, 6.0, 12.5 Hz, 1H), 3.46 (ddd, J = 3.2, 8.3,
13.6 Hz, 1H), 3.16 (dd, J = 5.1, 16.6 Hz, 1H), 3.08 (ddd,
J = 3.2, 7.8, 13.6 Hz, 1H), 2.95 (dd, J = 1.8, 16.6 Hz,
1H), 2.10–2.04 (m, 2H), 1.61 (m, 1H), 1.51–1.25 (m,
7H). A NOESY (399.5 MHz) experiment confirmed the
E geometry of the double bond, and an HMBC
(499.9 MHz) experiment confirmed the size of the macro-
cycle; 13C NMR (75.4 MHz; CD3OD/CDCl3, 3:1) d174.1,
173.7, 141.7, 141.4, 138.2, 136.4, 129.37, 129.28, 129.0,
128.9, 127.8, 126.7, 126.1, 125.1, 80.9, 79.8, 74.0, 73.7,
72.7, 72.4, 71.8, 58.3, 40.6, 39.6, 32.1, 29.5, 29.3, 28.7,
27.7. Anal (C31H42N2O7 Æ 1/3H2O) C, H, N.
1
(E/Z)-7. H NMR (499.9 MHz; CD3OD/CDCl3, 3:1) of
5.1.10. (2R,3R,4R,5R,14Z)-5-Benzyloxy-3,4-dihydroxy-6-
oxo-1-oxa-7-aza-cyclohexadec-14-ene-2-carboxylic
the LC–MS purified mixture was used to estimate the E/Z
ratio of 7 to ꢂ5:3.
acid
[(1S,2R)-2-hydroxyindan-1-yl]-amide ((Z)-7). 1H NMR

2206
K. Ersmark et al. / Bioorg. Med. Chem. 14 (2006) 2197–2208
(499.9 MHz; CD3OD/CDCl3, 3:1) d 7.39–7.18 (m, 9H),
5.61–5.51 (m, 2H), 5.37 (d, J = 5.0 Hz, 1H), 4.67–4.57
(m, 3H), 4.29 (ddm, J = 8.2, 10.1 Hz, 1H), 4.08 (dd,
J = 1.6, 7.7 Hz, 1H), 4.03–3.97 (m, 4H), 3.95 (dd,
J = 1.6, 7.6 Hz, 1H), 3.58 (ddd, J = 3.2, 9.5, 13.4 Hz,
1H), 3.17 (dd, J = 5.0, 16.6 Hz, 1H), 3.02 (ddd, J = 3.3,
6.3, 13.4 Hz, 1H), 2.95 (dd, J = 1.5, 16.6 Hz, 1H), 2.27
(m, 1H), 2.01 (m, 1H), 1.63–1.24 (m, 8H). A NOESY
(499.9 MHz) experiment confirmed the Z geometry of
the double bond, and an HMBC (499.9 MHz) experiment
confirmed the size of the macrocycle.
80.5 (2C, according to HMBC), 74.5, 73.7, 72.5, 71.4,
71.2, 58.2, 40.7, 39.6, 30.2, 30.0, 29.6, 28.8, 28.1, 26.7,
26.5. Anal (C31H42N2O7 Æ 2H2O) C, H, N.
5.1.13. Enzyme assays. Plm II was prepared according to
Westling et al.28, and the expression and purification of
Plm I will be published elsewhere (manuscript in prepara-
tion). The Plm IV assay was performed as described else-
where.29 Human liver Cat D was purchased from Sigma–
Aldrich, Sweden. The activities of Plm I, Plm II, and Cat
D were measured essentially as described earlier,30 using a
total reaction volume of 100 lL. The concentration of
pro-Plm II was 3 nM, the amount of Plm I was adjusted
to give a similar catalytic activity, and 50 ng/mL pro-
Cat D was used. The pro-sequence of Plm II was cleaved
off by preincubation in assay reaction buffer (100 mM
sodium acetate buffer (pH 4.5), 10% glycerol, and 0.01%
Tween 20) at room temperature for 40 min, and Cat D
was activated by incubation in the same reaction buffer
at 37 ꢁC for 20 min. The reaction was initiated by the
addition of 3 lM substrate (DABCYL-Glu-Arg-Nle-
Phe-Leu-Ser-Phe-Pro-EDANS, AnaSpec Inc, San Jose,
CA, USA) and hydrolysis was recorded as the increase
in fluorescence intensity over a 10-min time period, during
which the rate increased linearly with time. Stock solu-
tions of inhibitors in DMSO were serially diluted in
DMSO and added directly before addition of the sub-
strate, giving a final DMSO concentration of 1%. IC50 val-
ues were obtained by assuming competitive inhibition and
fitting the Langmuir isotherm (vi/vo = 1/(1 + [I]/IC50)) to
the dose–response data (Grafit), where vi and vo are the
initial velocities of the inhibited and uninhibited reac-
tions, respectively, and [I] is the inhibitor concentration.31
Ki was subsequently calculated using Ki = IC50/(1 + [S]/
Km)32 and a Km value was determined according to
Michaelis–Menten.
5.1.11. (2R,3R,4R)-2-Benzyloxy-3,4-dihydroxy-N-[(1S,2R)-
2-hydroxyindan-1-yl]-4-[(2R)-3-oxo-1-oxa-4-aza-cyclotri-
dec-2-yl]-butyramide (6). The mixture of E and Z isomers
of 5 (9.5 mg, 0.017 mmol) was dissolved in MeOH
(4 mL) and palladium 10 wt% (dry basis, 50% water)
on activated carbon (2 mg) was added. The reaction
mixture was stirred under H2 at room temperature for
3 h. The palladium on carbon was filtered off and the fil-
trate was concentrated under reduced pressure. Purifica-
tion using RP-LC (C8 column, 30 min gradient of 30–
80% CH3CN in 0.05% aqueous formic acid) gave the sat-
18
D
urated macrocycle 6 (5.9 mg, 62%) as a white solid: ½aꢁ
1
+20.3 (c 0.81, CHCl3); H NMR (300.0 MHz; CD3OD/
CDCl3, 3:1) d 7.38–7.16 (m, 9H), 5.35 (d, J = 5.1 Hz,
1H), 4.69 (d, J = 11.4 Hz, 1H), 4.64 (d, J = 11.4 Hz,
1H), 4.56 (ddd, J = 1.7, 5.1, 5.1 Hz, 1H), 4.18 (d,
J = 6.3 Hz, 1H), 4.03 (dd, J = 2.3, 6.3 Hz, 1H), 3.99
(dd, J = 2.3, 7.2 Hz, 1H), 3.85 (d, J = 7.2 Hz, 1H), 3.74
(m, 1H), 3.58–3.36 (m, 2H), 3.26–3.10 (m, 2H), 2.93
(dd, J = 1.7, 16.7 Hz, 1H), 1.78–1.1.62 (m, 2H), 1.57–
1.23 (m, 12H); 13C NMR (75.4 MHz; CD3OD/CDCl3,
3:1) d 174.0, 173.6, 141.43, 141.35, 138.1, 129.3, 129.2,
128.94, 128.92, 127.8, 126.0, 125.1, 82.1, 81.4, 74.0,
73.6, 71.74, 71.66, 71.5, 58.3, 40.5, 39.1, 29.2, 26.9,
25.9, 25.5, 23.9, 23.6. One 13C signal is missing in the ali-
phatic area due to overlapping resonances. When chang-
ing solvent to pure CDCl3, all of the aliphatic signals
were split into separate signals. Anal (C31H42N2O7) C,
H, N.
5.1.14. Computational modeling. Compounds 5–8 were
built in an extended conformation with the program
ViewerPro,33 while the coordinates of the plasmepsins
investigated were retrieved from the Bookhaven Data-
bank34 (codes 1LF2 and 1LF3 for Plm II and 1LS5
for Plm IV) with subsequent addition of hydrogens.
Regarding the catalytic aspartates, Asp34 was consid-
ered protonated and Asp214 negatively charged, accord-
ing to our previous works.11 Docking was performed
with GOLD v2.2,35 allowing full flexibility for the li-
gand, while keeping the protein fixed. The following
protocol was used to explore the three crystallographic
structures mentioned above: 20 independent runs of
the docking algorithm were performed with each com-
pound, using the default genetic algorithm (GA) search
parameters but increasing the number of operators that
are applied over the course of a GA run to 300.000.
Both the Goldscore and Chemscore scoring functions
implemented in GOLD36 were used in parallel assays
on each compound. The automated assignment of pro-
tein and ligand atom types was checked in order to en-
sure the correct treatment of the potential hydrogen
bond donors and acceptors.
5.1.12.
(2R,3R,4R,5R)-5-Benzyloxy-3,4-dihydroxy-6-
oxo-1-oxa-7-aza-cyclohexadec-2-carboxylic acid [(1S,2R)-
2-hydroxyindan-1-yl]-amide (8). The mixture of E and Z
isomers of 7 (8.6 mg, 0.016 mmol) was dissolved in
MeOH (4 mL) and palladium 10 wt% (dry basis, 50%
water) on activated carbon (2 mg) was added. The reac-
tion mixture was stirred under H2 at room temperature
for 3 h. The palladium on carbon was filtered off and the
filtrate was concentrated under reduced pressure. Purifi-
cation using RP-LC (C8 column, 30 min gradient of 30–
80% CH3CN in 0.05% aqueous formic acid) gave the
saturated macrocycle 8 (5.2 mg, 60%) as a white solid:
18
D
½aꢁ +9.8 (c 1.2, CH2Cl2); 1H NMR (300.0 MHz;
CD3OD) d 7.45–7.17 (m, 9H), 5.41 (d, J = 4.8 Hz,
1H), 4.66–4.55 (m, 3H), 4.10 (dd, J = 1.3, 9.1 Hz, 1H),
4.01–3.93 (m, 2H), 3.90 (d, J = 8.8 Hz, 1H), 3.73 (m,
1H), 3.62 (ddd, J = 3.1, 8.9, 13.3 Hz, 1H), 3.46 (m,
1H), 3.19 (dd, J = 5.0, 16.6 Hz, 1H), 3.02–2.91 (m,
2H), 1.67–1.24 (m, 14H); 13C NMR (75.4 MHz;
CD3OD/CDCl3, 3:1) d 174.64, 174.60, 141.7, 141.3,
138.1, 129.34, 129.28, 129.0, 128.9, 127.8, 126.1, 125.1,
Binding free energies were estimated with the linear
interaction energy (LIE) method, described in detail

K. Ersmark et al. / Bioorg. Med. Chem. 14 (2006) 2197–2208
2207
2
elsewhere.25,37 Basically, this approach estimates the
ligand free energy of binding from the difference in the
ligand surrounding energies in both its bound and free
states. The relationship between ligand surrounding
energies and free energy of binding is given by the
equation
10 kcal/(mol A ) force constant. MD followed for more
than 1.5 ns under the same conditions as for the bound
state, but keeping the initial position of the central atom
of the ligand fixed, to ensure a homogeneous solvation.
˚
ꢀ
ꢁ
ꢀ
ꢁ
Acknowledgments
vdW
l–s
el
l–s
DGbind ¼ aD V
þ bD V
vdW
l–s
el
We thank the Swedish Foundation for Strategic Re-
search (SSF/RAPID) and the Swedish Research Council
(VR) for financial support.
where V
and V denote, respectively, the Lennard-
l–s
Jones and electrostatic potentials between the ligand
and its surroundings (l–s), which are evaluated as energy
averages (denoted by hi) from separate MD simulations
of the free (solvated in water) and bound states. The dif-
ference (D) between such averages for each type of po-
tential is scaled by different coefficients (see Ref. 37),
giving the polar (electrostatic) and non-polar (van der
Waals) contributions to the free energy. For the non-po-
lar contribution, this coefficient has been empirically set
to a = 0.181, while for the polar contribution the scaling
factor is dependent on the chemical nature of the ligand.
For ligands with at least two hydroxyl groups, like the
ones handled in this study, the value is b = 0.33.
References and notes
1. White, N. J. J. Clin. Invest. 2004, 113, 1084.
2. Breman, J. G. Am. J. Trop. Med. Hyg. 2001, 64, 1.
3. Wiesner, J.; Ortmann, R.; Jomaa, H.; Schlitzer, M. Angew
Chem., Int. Ed. 2003, 42, 5274.
4. Coombs, G. H.; Goldberg, D. E.; Klemba, M.; Berry, C.;
Kay, J.; Mottram, J. C. Trends Parasitol. 2001, 17, 532.
5. Francis, S. E.; Gluzman, I. Y.; Oksman, A.; Knicker-
bocker, A.; Mueller, R.; Bryant, M. L.; Sherman, D. R.;
Russell, D. G.; Goldberg, D. E. EMBO J. 1994, 13, 306.
6. Boss, C.; Richard-Bildstein, S.; Weller, T.; Fischli, W.;
Meyer, S.; Binkert, C. Curr. Med. Chem. 2003, 10, 883.
7. Banerjee, R.; Liu, J.; Beatty, W.; Pelosof, L.; Klemba, M.;
Goldberg, D. E. Proc. Natl. Acad. Sci. U.S.A. 2002, 99,
990.
8. Berry, C. Curr. Opin. Drug Dis. Dev. 2000, 3, 624.
9. Omara-Opyene, A. L.; Moura, P. A.; Sulsona, C. R.;
Bonilla, J. A.; Yowell, C. A.; Fujioka, H.; Fidock, D. A.;
Dame, J. B. J. Biol. Chem. 2004, 279, 54088.
10. Liu, J.; Gluzman, I. Y.; Drew, M. E.; Goldberg, D. E.
J. Biol. Chem. 2005, 280, 1432.
11. Ersmark, K.; Feierberg, I.; Bjelic, S.; Hulten, J.; Samuels-
Molecular dynamics were done using the program Q,38
with the version of the OPLS all atom forcefield imple-
mented there.39 A 20 A sphere of simulation was chosen,
˚
centered on the transition state mimic hydroxyl oxygen
interacting with the catalytic aspartates. The sphere
was solvated with TIP3P waters40 with the water surface
subjected to radial and polarization restraints41 in order
to mimic bulk water at the sphere boundary. Non-bond-
˚
ed interaction energies were measured up to a 10 A cut-
off, except for the ligand atoms where no cutoff was
used. Beyond the cutoff, long-range electrostatics were
treated with the local reaction field (LRF) multipole
expansion method.42 For the ligand–protein simulation,
an equilibration phase was performed before the data
collection phase. The equilibration protocol started with
a 1000 step MD with very short time step (0.2 fs) at 1 K,
coupled to a strong bath (0.2 fs bath coupling) thus sim-
ilar to energy minimization. Then the system was grad-
ually heated up to 300 K, relaxing the bath coupling to
10 fs and increasing the timestep to 1.5 fs. During this
equilibration process, on Plm IV, hydrogen bonding
constraints were imposed between the central hydroxyls
and the aspartates (only one hydroxyl could be used for
the 16-membered ring compounds). The need of these
constraints is related to the non-optimal geometry that
the 1LS5 structure shows for the catalytic aspartates.
Unrestricted molecular dynamics followed then for
1.8–2.5 ns, making use of the SHAKE algorithm for
the treatment of the bond lengths43 and collecting ener-
gies at regular intervals of 15 fs. The MD was run until
the energies collected showed stability for a period not
shorter than 600 ps (collection period). Stability was ad-
dressed by comparing the average values of the first and
second halves of the collection period.
˚
son, B.; Aqvist, J.; Hallberg, A. Bioorg. Med. Chem. 2003,
11, 3723.
12. Ersmark, K.; Feierberg, I.; Bjelic, S.; Hamelink, E.;
Hackett, F.; Blackman, M. J.; Hulten, J.; Samuelsson,
˚
B.; Aqvist, J.; Hallberg, A. J. Med. Chem. 2004, 47, 110.
13. Ersmark, K.; Nervall, M.; Hamelink, E.; Janka, L. K.;
Clemente, J. C.; Dunn, B. M.; Blackman, M. J.; Samu-
˚
elsson, B.; Aqvist, J.; Hallberg, A. J. Med. Chem. 2005, 48,
6090.
14. Fairlie, D. P.; Tyndall, J. D. A.; Reid, R. C.; Wong, A. K.;
Abbenante, G.; Scanlon, M. J.; March, D. R.; Bergman,
D. A.; Chai, C. L. L.; Burkett, B. A. J. Med. Chem. 2000,
43, 1271.
15. Weber, A. E.; Halgren, T. A.; Doyle, J. J.; Lynch, R. J.;
Siegl, P. K.; Parsons, W. H.; Greenlee, W. J.; Patchett, A.
A. J. Med. Chem. 1991, 34, 2692.
16. Loughlin, W. A.; Tyndall, J. D. A.; Glenn, M. P.; Fairlie,
D. P. Chem. Rev. 2004, 104, 6085.
17. Ghosh, A. K.; Devasamudram, T.; Hong, L.; DeZutter,
C.; Xu, X.; Weerasena, V.; Koelsch, G.; Bilcer, G.; Tang,
J. Bioorg. Med. Chem. Lett. 2005, 15, 15.
18. Tyndall, J. D. A.; Fairlie, D. P. Curr. Med. Chem. 2001, 8,
893.
19. Silva, A. M.; Lee, A. Y.; Gulnik, S. V.; Maier, P.; Collins,
J.; Bhat, T. N.; Collins, P. J.; Cachau, R. E.; Luker, K. E.;
Gluzman, I. Y.; Francis, S. E.; Oksman, A.; Goldberg, D.
E.; Erickson, J. W. Proc. Natl. Acad. Sci. U.S.A. 1996, 93,
10034.
20. Tyndall, J. D. A.; Reid, R. C.; Tyssen, D. P.; Jardine, D.
K.; Todd, B.; Passmore, M.; March, D. R.; Pattenden, L.
K.; Bergman, D. A.; Alewood, D.; Hu, S.-H.; Alewood, P.
The MD sampling of the free ligand was done using the
˚
same 20 A water sphere of solvation. The solvent was
equilibrated with a 10 ps MD at 300 K, in which the
heavy atoms of the ligand were restrained through a

2208
K. Ersmark et al. / Bioorg. Med. Chem. 14 (2006) 2197–2208
F.; Birch, C. J.; Martin, J. L.; Fairlie, D. P. J. Med. Chem.
2000, 43, 3495.
21. Li, T.; Yowell, C. A.; Beyer, B. B.; Hung, S.-H.; Westling,
J.; Lam, M. T.; Dunn, B. M.; Dame, J. B. Mol. Biochem.
Parasitol. 2004, 135, 101.
31. Copeland, R. A., Ed.; Enzymes: A Practical Introduction
to Structure, Mechanism, and Data Analysis New York,
1996.
32. Cheng, Y.; Prusoff, W. H. Biochem. Pharmacol. 1973, 22,
3099.
22. Oscarsson, K.; Lahmann, M.; Lindberg, J.; Kangasmetsa,
J.; Unge, T.; Oscarson, S.; Hallberg, A.; Samuelsson, B.
Bioorg. Med. Chem. 2003, 11, 1107.
23. Alterman, M.; Bjoersne, M.; Muehlman, A.; Classon, B.;
Kvarnstroem, I.; Danielson, H.; Markgren, P.-O.; Nill-
roth, U.; Unge, T.; Hallberg, A.; Samuelsson, B. J. Med.
Chem. 1998, 41, 3782.
24. Anon. Synlett 2003, 423.
˚
25. Aqvist, J.; Medina, C.; Samuelsson, J. E. Protein Eng.
1994, 7, 385.
33. ViewerPro 5.0, Accelrys Inc., San Diego, CA, 2002.
34. Westbrook, J.; Feng, Z. K.; Chen, L.; Yang, H. W.;
Berman, H. M. Nucleic Acids Res. 2003, 31, 489.
35. Jones, G.; Willett, P.; Glen, R. C.; Leach, A. R.; Taylor,
R. J. Mol. Biol. 1997, 267, 727.
36. Verdonk, M. L.; Cole, J. C.; Hartshorn, M. J.; Murray, C.
W.; Taylor, R. D. Prot. Struct. Funct. Genet. 2003, 52, 609.
˚
37. Hansson, T.; Marelius, J.; Aqvist, J. J. Comput. Aided
Mol. Des. 1998, 12, 27.
38. Marelius, J.; Kolmodin, K.; Feierberg, I.; Aqvist, J.
˚
26. Andrews, P. R. In The Practice of Medicinal Chemistry;
Wermuth, C. G., Ed.; Academis Press: 1996; p. 345.
27. Yoshimura, N.; Tamura, M. Eur. Pat. Appl. 69339, 1983.
28. Westling, J.; Cipullo, P.; Hung, S.-H.; Saft, H.; Dame, J.
B.; Dunn, B. M. Protein Sci. 1999, 8, 2001.
J. Mol. Graph Modelling 1999, 16, 213.
39. Jorgensen, W. L.; Maxwell, D. S.; TiradoRives, J. J. Am.
Chem. Soc. 1996, 118, 11225.
40. Jorgensen, W.; Chandrasekhar, J.; Madura, J.; RW, I.;
Klein, M. J. Chem. Phys. 1983, 79, 926.
29. Dame,J.B.;Yowell,C.A.;Omara-Opyene,L.;Carlton,J.M.;
Cooper, R. A.; Li, T. Mol. Biochem. Parasitol. 2003, 130, 1.
30. Haque, T. S.; Skillman, A. G.; Lee, C. E.; Habashita, H.;
Gluzman, I. Y.; Ewing, T. J. A.; Goldberg, D. E.; Kuntz,
I. D.; Ellman, J. A. J. Med. Chem. 1999, 42, 1428.
41. King, G.; Warshel, A. J. Chem. Phys. 1989, 91, 3647.
42. Lee, F. S.; Warshel, A. J. Chem. Phys. 1992, 97,
3100.
43. Ryckaert, J. P.; Ciccotti, G.; Berendsen, H. J. C. J. Compt.
Phys. 1977, 23, 327.