Starving the malaria parasite: Inhibitors active against the aspartic proteases plasmepsins I, II, and IV

Article abstract of DOI:10.1002/anie.200504119

Clamping down: A new class of aspartic protease inhibitors that target the malarial protease family Plasmepsin are reported. These ligands utilize a novel "diamine clamp" to engage the catalytic dyad. They are potent inhibitors of plasmepsins I, II, and IV, while retaining good selectivity against the closely related human cathepsins D and E. (Chemical Equation Presented) .

Full text of DOI:10.1002/anie.200504119

Communications
year.[1] The plasmepsins (PMs) are a family of Plasmodium
aspartic proteases that digest human hemoglobin and deliver
amino acids that are required for growth.[2,3] In the search for
targets for new antimalarial therapies, the mutually redun-
dant proteolytic activity of plasmepsins for hemoglobin[4,5]
requires inhibitors that are broadly active against all hemo-
globin-degrading plasmepsins (PMI, PMII, PMIV, and
HAP)[6] while remaining inactive against the most closely
related human aspartic proteases (cathepsins D and E;
hCatD and E). Among the plasmepsins, only the structure
[2,7–9]
of PMII has been extensively characterized.
As a result
of the similar sequence identity among the plasmepsins, we
hypothesized that the renin-like flap-open conformation
(Figure 1a), observed in two[9,10] of the fourteen known
structures of PMII, would be accessible by each of the
other three members of the Plasmepsin family. On this basis,
we initiated a structure-based design program to explore the
activity and selectivity of Plasmepsin inhibitors that target the
flap pocket.
Enzyme Inhibitors
DOI: 10.1002/anie.200504119
Starving the Malaria Parasite: Inhibitors Active
against the Aspartic Proteases Plasmepsins I, II,
and IV**
Fraser Hof, Andri Schütz, Christoph Fäh,
Solange Meyer, Daniel Bur, Jun Liu,
Daniel E. Goldberg, and François Diederich*
The most dangerous of the malaria-causing parasites, Plas-
modium falciparum, infects 300–660 million people each
[*] A. Schütz, C. Fäh, Prof. Dr. F. Diederich
Laboratorium für Organische Chemie
ETH Zürich
Hönggerberg, HCI, 8093 Zürich (Switzerland)
Fax: (+41)1-632-1109
E-mail: diederich@org.chem.ethz.ch
Dr. F. Hof
Department of Chemistry
University of Victoria
P.O. Box 3065, STN CSC, Victoria, BC, V8W3V6 (Canada)
Figure 1. Plasmepsin II and the discussed inhibitors. a) Schematic
diagram of the active site showing the catalytic dyad, the S1/S3
pocket, and the flap pocket. b) A bicyclic amine that engages the
catalytic aspartate dyad of Plasmepsin II. c) A “diamine clamp”
designed to provide extra hydrogen bonding to the catalytic residues.
Dr. S. Meyer, Dr. D. Bur
Actelion Pharmaceuticals
Gewerbestrasse 16, 4123 Allschwil (Switzerland)
J. Liu, Prof. D. E. Goldberg
Howard Hughes Medical Institute
Departments of Medicine and Molecular Microbiology
Washington University
660 South Euclid Avenue, St. Louis, MO 63110-1093 (USA)
Our earliest inhibitors (Figure 1b) displayed moderate
activity (IC50 = 3000–35000 nm; IC50 = concentration of inhib-
itor at which 50% of the maximum initial rate is observed) for
PMII. [11–13] Herein we report the creation and elaboration of a
new core consisting of a bicyclic diamine that forms a
hydrogen-bonded clamp around the catalytic aspartate resi-
dues (Figure 1c). The exo amino group additionally provides
an attachment point for a pocket-directed binding substitu-
[**] F.H. thanks the Novartis Foundation and the Human Frontier
Science Program for financial support. This research was supported
by the ETH Research Council. We thank Anna Oksman for technical
assistance.
Supportinginformation for this article is available on the WWW
under http://www.angewandte.org or from the author.
2138
ꢀ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2006, 45, 2138 –2141

Angewandte
Chemie
ent. The sulfone moiety in the b position serves to decrease
the basicity of the amino groups that bind to the catalytic
dyad, and can also be adorned as a second binding vector .
Our first generation of diamine-clamp inhibitors (e.g.
(Æ )-1, Figure 2b) targeted the S1/S3 pocket with a sulfone-
angle constraint within our molecular-mechanics-based mod-
eling program[15] exposed the flaw in our original design;
when the dihedral angle is properly set, the rigid aryl sulfone
linkage directs the naphthyl substituent directly into Ile23
and Phe120, thus producing significantly unfavorable steric
clashes (Figure 2b).
Subsequent modeling by using a properly fixed dihedral-
angle constraint (qC1-S-C2-C3 = 908) suggested a means by which
we could exploit the rigid conformation of the aryl sulfone
moiety. By using the enantiomer of the original aminosulfone
core, but swapping the flap-binding element onto the endo
sulfone vector and the S1/S3 binding element onto the more
flexible exo amino vector, we arrived at
a pseudo-
enantiomeric family of structures that simultaneously
engaged the catalytic dyad, flap pocket, and S1/S3 pocket
with the appropriate binding elements (diamine clamp,
hydrophobic chain, and aryl group, respectively; Figure 3).
Figure 2. Conformational analysis of a first-generation diamine-clamp
inhibitor. a) Relative energy while driving the dihedral angle qC1-S-C2-C3 of
an aryl sulfone from 0–1808. DFT-B3LYP calculations carried out using
Spartan (6-31G* basis set, MP2 correlation).[19] A molecular model and
a Newman projection of the optimal angle (qC1-S-C2-C3 =908) are shown
at the right. b) MOLOC-generated models of inhibitor 1 within the
active site of PM II. In the original, unconstrained model (blue), the
naphthyl residue is allowed to comfortably fit within the S1/S3 binding
pocket, but the model in which the conformational preference of the
aryl sulfone is taken into account by fixingof the qC1-S-C2-C3 at 908
(yellow) produces steric clashes with Ile123 and Phe120 of the
enzyme. The enantiomer of the inhibitor that is predicted by modeling
to bind into the active site is shown.
Figure 3. a) A pseudoenantiomeric second generation of inhibitors
arisingfrom the swap of flap and S1/S3 substituents between the
amine and sulfone functionalities.
The target compounds were synthesized as shown in
Scheme 1.[16] The key exo amino substituent is attached
through a diastereoselective Michael-type addition of alkyl
amines to the vinyl sulfones (Æ )-2 and (Æ )-3 to give the
desired exo-amino/endo-sulfone isomers (Æ )-4 and (Æ )-5,
respectively. Subsequent functional group manipulation pro-
vided a second generation of diamine clamp inhibitors
consisting of (Æ )-6, (Æ )-7 (via (Æ )-8), (Æ )-9, (Æ )-10 (via
(Æ )-11), and (Æ )-12.
linked naphthyl substituent (for the synthesis, see the
Supporting Information). Regardless of the identity of the
exo amine substituent, the ability of these ligands to inhibit
PMII was unexpectedly poor. IC values of 9500–60000 nm
The in vitro activities of (Æ )-7, (Æ )-9 and (Æ )-10, and
(Æ )-12 for PMII showed that the affinity of these ligands was
significantly improved, with IC50 values ranging from 130 to
400 nm (Table 1, column 3[17]). The presence of the hexyl chain
that, according to our modeling, occupies the flap pocket
(compare compounds (Æ )-6 and (Æ )-7), is responsible for the
potency (> 100-fold difference for PMII) as well as the
selectivity ((Æ )-7 shows IC50(hCat D)/IC50(PMII) > 15,
whereas (Æ )-6 shows essentially no discrimination). The
aryl substituent directed into the S1/S3 pocket can be
decreased in size (see compound (Æ )-12) without affecting
activity (IC50(PMII) = 210 nm) or selectivity (IC50(hCat D)/
IC50(PMII) > 13). Furthermore, an increase in size, for
example, through introduction of a biaryl substituent (as in
50
were observed and led us to re-evaluate our design. In
particular, we were concerned that the aryl sulfone linkage
might be too rigid to allow for the optimal orientation of the
naphthalene binding element within the S1/S3 pocket (known
to prefer aryl binding partners[3]). The density functional
theory (DFT) dihedral-angle-driving calculation for the C1-S-
C2-C3 dihedral angle (Figure 2a) continually varied between
0 and 1808. This revealed an unexpectedly strong preference
for the orientation in which the p orbital of the ipso carbon
=
atom bisects the two S O bonds of the sulfone moiety. This
conformational preference is also evident in a survey of aryl
sulfone crystal structures in the Cambridge Structural Data-
base (CSD).[14] The incorporation of this preference as an
Angew. Chem. Int. Ed. 2006, 45, 2138 –2141
ꢀ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
2139

Communications
(hCat D)/IC50(PMIV) = 90 and IC50-
(hCat E)/IC50(PMIV) = 173). Compound
(Æ )-6, which lacks an appropriate flap
substituent, is inactive in all assays. It can
therefore be seen that the hexylphenyl
substituent is responsible for the potent
activity against PMI, II, and IV. This
suggests that the flap-open conformation is
likely to be accessible by each of these
three enzymes, and that the binding energy
provided by occupation of the mostly
hydrophobic flap pockets of PMI, II, and
IV is significant.
To further clarify the binding mode of
these inhibitors, compound (Æ )-5 was
resolved into its enantiomers (+)-5 and
(À)-5 by preparative chiral HPLC.[18] The
enantiomers of 5 were then subsequently
deprotected to give the optically pure
inhibitors (+)-12 and (À)-12. The IC50
values of (À)-12 for PMII (45 n m) and
PMIV (10 n m) are, within error, twofold
better than those of the racemate, and (+)-
12 is practically inactive against PMII and
IV (IC50 = 3260 and 33900 nm, respec-
tively). Modeling of the two enantiomers
show that the better-binding enantiomer
((À)-12) fits nicely within the active site
and simultaneously engages the catalytic
dyad, flap pocket, and the S1/S3 pocket.
The other enantiomer, however, can be
docked into the active site only in a less-
preferred binding orientation (see the
Supporting Information). On the basis of
this information, the absolute configura-
tion of the active enantiomer ((À)-12) was
tentatively assigned as shown in Figure 4.
Finally, compounds (Æ )-7, (Æ)-9, (Æ )-
10, and (Æ )-12 were tested for parasite-
growth inhibition in cultures of parasite-
infected red blood cells (Table 1, column
7). The growth of the parasites is indicated
Scheme 1. Synthesis of second-generation diamine-clamp inhibitors from vinyl sulfone precursors
(Æ)-2 and (Æ)-3. a) 2-Naphthylmethylamine, 958C, 16 h; 81%. b) n-C6H13B(OH)2, Pd(OAc)2, 2-
(dicyclohexylphosphanyl)biphenyl, K3PO4, PhMe, 808C, 16 h; 65%. c) BBr3, CH2Cl2, À788C, 15 min;
85–100%. d) 3-Bromobenzylamine, 708C, 1 h; 65%. e) 2-Methoxyphenylboronic acid, Pd(OAc)2, 2-
(dicyclohexylphosphanyl)biphenyl, K3PO4, DMF, 908C, 16 h; 76%. f) BBr3, CH2Cl2, À788C. Ar=aryl
group, Boc=tert-butyloxycarbonyl, DMF=N,N-dimethylformamide
Table 1: Inhibition of plasmepsins and human cathepsins by “diamine-clamp” inhibitors targeting the
flap pocket.
Compound
PM I[a,b]
PM II[a,b]
PM II[a,c]
PM IV[a,c]
hCatD[a,c]
hCat E[a,c]
Culture[d]
(Æ)-6
n.d.
100
210
290
150
n.d.
n.d.
n.d.
370
2100
930
670
n.d.
n.d.
14800
130
390
400
210
17300
50
180
110
30
18000
2030
1410
1500
2800
n.d.
30500
7050
5910
5400
5200
n.d.
n.d.
5500
3700
2200
7100
n.d.
(Æ)-7
(Æ)-9
(Æ)-10
(Æ)-12
(À)-12
(+)-12
45
3260
10
33900
n.d.
n.d.
n.d.
[a] IC50 values (nm) are averages of two to three repetitions of a fluorogenic proteolysis assay (see the
[9,20]
SupportingInformation).
The estimated error for this assay is Æ50%. [b] Values determined at
Washington University: [enzyme]=1 nm, [substrate]=1 mm, 100 mm acetate buffer, pH 5.1. [c] Values
determined at Actelion Pharmaceuticals: [enzyme]=1 nm, [substrate]=1 mm, 50 mm acetate buffer,
pH 5.0, 12.5% glycerol, 10% Me2SO, 0.1% BSA. [d] Culture IC50 values (nm) obtained by triplicate
analysis of the dose-dependent 3H-hypoxanthine uptake by cultured Plasmodium falciparum (3D7)
parasite (see the SupportingInformation). [21] Estimated errors are Æ10%. n.d.=not determined.
compounds (Æ )-9 and (Æ )-10), leads to an insignificant
dropoff in activity (IC50(PMII) = 390 and 400 nm, respec-
tively).
A selection of compounds was tested against a wide panel
of plasmepsins (Table 1). Although these compounds were
designed based upon a three-dimensional structure of PMII,
all of the noncontrol compounds were found to be more-
potent inhibitors of PMI and IV, than of PMII. These
compounds retain little affinity for hCat D and E, which
means that either there are significant differences in the
structure of the flap pocket, or that it is missing altogether.
The ligands with the biphenyl S1/S3 substituents ((Æ )-9 and
(Æ )-10) are the most active against the human enzymes,
whereas that with the smaller S1/S3 substituent (bromoaryl
compound (Æ )-12) displays the best overall selectivity (IC50-
Figure 4. Absolute configuration of the active enantiomer (À)-12 (left),
proposed on the basis of molecular modelingwithin the active site of
PM II (right). Several residues have been removed from the model to
permit visualization of the flap pocket, occupied by the n-hexyl chain
of the inhibitor. This image was prepared using MSMS and Chi-
mera.[22,23]
2140
www.angewandte.org
ꢀ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2006, 45, 2138 –2141

Angewandte
Chemie
3
Gluzman, T. J. A. Ewing, D. E. Goldberg, I. D. Kuntz, J. A.
Ellman, J. Med. Chem. 1999, 42, 1428 – 1440.
by uptake of H-hypoxanthine, which is measured by scintil-
lation counting on isolated cells after a period of incubation
with the inhibitor (see the Supporting Information). The
culture IC50 values range from 2200 to 7100 nm. Although the
new diamine-clamp compounds are significantly less active
in culture than in vitro, these results demonstrate that they
are sufficiently active against the Plasmepsin family to shut
down parasite growth.
[8] N. Khazanovich Bernstein, M. M. Cherney, C. A. Yowell, J. B.
Dame, M. N. G. James, J. Mol. Biol. 2003, 329, 505 – 524.
[9] L. Prade, A. F. Jones, C. Boss, S. Richard-Bildstein, S. Meyer, C.
Binkert, D. Bur, J. Biol. Chem. 2005, 280, 23837 – 23843.
[10] J. Yuvaniyama, A. DꢀArcy, C. Öfner, H. Lötscher, D. Bur, G. E.
Dale, H. P. Märki, F. K. Winkler, R. P. Moon, R. G. Ridley in
Symposium on “Parasite proteases as drug targets”, University of
Glasgow, UK, 2000.
Taken together, these results suggest that a conformation
with a flap-open pocket is accessible for each member of the
Plasmepsin family, and that engaging this pocket with a
complementary hydrophobic substituent provides significant
binding energy. These compounds all display high selectivity
against the human cathepsins D and E, suggesting that this
design principle holds promise for the production of useful
broad-spectrum plasmepsin inhibitors that remain innocuous
to the most closely related human enzymes. The bicyclic
diamine clamp introduced in this work shows potent inhib-
ition of targeted structures while allowing for significant
discrimination between different classes of aspartic proteases.
We are currently working to improve the in vivo biological
activity in an effort to develop the diamine-clamp inhibitors
as a new class of antimalarials.
[11] D. A. Carcache, S. R. Hörtner, A. Bertogg, F. Diederich, A.
Dorn, H. P. Märki, C. Binkert, D. Bur, Helv. Chim. Acta 2003, 86,
2192 – 2209.
[12] D. A. Carcache, S. R. Hörtner, P. Seiler, F. Diederich, A. Dorn,
H. P. Märki, C. Binkert, D. Bur, Helv. Chim. Acta 2003, 86, 2173 –
2191.
[13] D. A. Carcache, S. R. Hörtner, A. Bertogg, C. Binkert, D. Bur,
H. P. Märki, A. Dorn, F. Diederich, ChemBioChem 2002, 3,
1137 – 1141.
[14] The conformational restriction of aryl sulfones was first brought
to our attention in the form of a CSD survey carried out by Dr.
Klaus Müller and F. Hoffman-La Roche, which was presented
during a seminar series at the ETH Zürich in 2004. See the
Supporting Information.
[15] All enzyme–inhibitor complexes were modeled using the MAB
force field as implemented in the program MOLOC: a) P. R.
Gerber, K. Müller, J. Comput.-Aided Mater. Des. 1995, 9, 251 –
268; b) Gerber Molecular Design (http://www.moloc.ch).
[16] All new compounds were fully characterized by m.p., IR, 1H and
13C NMR, MS, and HRMS or elemental analysis.
Received: November 19, 2005
Published online: February 27, 2006
[17] Inhibition studies for PMII were, as indicated in Table 1 and the
Supporting Information, carried out in two different laboratories
under different conditions, and as such, the slightly different
values for PMII inhibition are not unexpected. The results in
Table 1, column 3, which were obtained under conditions
identical to those used for PMIV and both human cathepsins,
are considered for the purposes of discussions in the text relating
to selectivity.
Keywords: aspartic proteases · drugdesign · inhibitors ·
malaria · medicinal chemistry
.
[1] R. W. Snow, C. A. Guerra, A. M. Noor, H. Y. Myint, S. I. Hay,
Nature 2005, 434, 214 – 217.
[2] A. M. Silva, A. Y. Lee, S. V. Gulnik, P. Majer, J. Collins, T. N.
Bhat, P. J. Collins, R. E. Cachau, K. E. Luker, I. Y. Gluzman,
S. E. Francis, A. Oksman, D. E. Goldberg, J. W. Erickson, Proc.
Natl. Acad. Sci. USA 1996, 93, 10034 – 10039.
[18] Baseline separation was achieved by using a semipreparative
Chiralpak AD-H column and eluting with hexane/iPrOH/
HNMe2 (80:20:0.1). This separation was conducted at Chiral
Technologies Europe, 67404 Illkirch, France.
[3] C. Boss, S. Richard-Bildstein, T. Weller, W. Fischli, S. Meyer, C.
Binkert, Curr. Med. Chem. 2003, 10, 883 – 907.
[4] A. L. Omara-Opyene, P. A. Moura, C. R. Sulsona, J. A. Bonilla,
C. A. Yowell, H. Fujioka, D. A. Fidock, J. B. Dame, J. Biol.
Chem. 2004, 279, 54088 – 54096.
[19] Spartanꢀ04, Wavefunction, Inc., Irvine, USA, 2004.
[20] K. E. Luker, S. E. Francis, I. Y. Gluzman, D. E. Goldberg, Mol.
Biochem. Parasitol. 1996, 79, 71 – 78.
[21] I. Y. Gluzman, S. E. Francis, A. Oksman, C. E. Smith, K. L.
Duffin, D. E. Goldberg, J. Clin. Invest. 1994, 93, 1602 – 1608.
[22] M. F. Sanner, A. J. Olson, J. C. Spehner, Biopolymers 1996, 38,
305 – 320.
[5] J. Liu, I. Y. Gluzman, M. E. Drew, D. E. Goldberg, J. Biol. Chem.
2005, 280, 1432 – 1437.
[6] A. Nezami, T. Kimura, K. Hidaka, A. Kiso, J. Liu, Y. Kiso, D. E.
Goldberg, E. Freire, Biochemistry 2003, 42, 8459 – 8464.
[7] a) O. A. Asojo, S. V. Gulnik, E. Afonina, B. Yu, J. A. Ellman,
T. S. Haque, A. M. Silva, J. Mol. Biol. 2003, 327, 173 – 181;
b) T. S. Haque, A. G. Skillman, C. E. Lee, H. Habashita, I. Y.
[23] Molecular-graphics images were produced by using the Chimera
package from the Computer Graphics Laboratory, University of
California, San Francisco (supported by NIH P41 RR-01081):
C. C. Huang, G. S. Couch, E. F. Pettersen, T. E. Ferrin, Pac.
Symp. Biocomput. 1996, 1, 724; http://www.cgl.ucsf.edu/chimera.
Angew. Chem. Int. Ed. 2006, 45, 2138 –2141
ꢀ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
2141