вход по аккаунту


An Old Target Revisited Two New Privileged Skeletons and an Unexpected Binding Mode For HIV-Protease Inhibitors.

код для вставкиСкачать
Structure Determination
An Old Target Revisited: Two New Privileged
Skeletons and an Unexpected Binding Mode For
HIV-Protease Inhibitors**
Edgar Specker, Jark Bttcher, Hauke Lilie,
Andreas Heine, Andreas Schoop, Gerhard Mller,
Nils Griebenow, and Gerhard Klebe*
proper recognition and arrest of their substrates. Furthermore, both skeletons were designed with the impetus to give
easy access to multiple decorations using side chains tailored
to systematically explore and specifically address the various
substrate recognition pockets exhibited by this class of
Up to now a plethora of privileged building blocks have
been incorporated successfully into HIV-protease inhibitors
(Figure 1). The first generation of inhibitors comprises motifs
HIV protease is probably the most vigorously studied
enzymatic target for therapeutic intervention in the short
history of structure-based drug design.[1] For over a decade,
attempts to find new efficient inhibitors have attracted and
absorbed the curiosity, talents, and energy of many scientists.[2] No other protein has been scrutinized as often in
crystal structures of protein–ligand complexes; in the publicly
available Protein Data Bank more than 200 individual
structures have been deposited.[3] As a result of these
impressive efforts, nine approved drugs have been launched
that improve both the quality of life and life expectancy for an
increasing number of HIV-infected patients.[4] However, the
initial euphoria surrounding these major achievements has
decreased due to the occurrence of viral resistance to the
prescribed therapy based on these drugs.[5] As a consequence,
novel inhibitors with alternative frameworks are being sought
to circumvent resistance development.
Can we expect any new inhibitory principles to be
discovered in the case of such an exhaustively studied
member of the aspartyl protease family? In an attempt to
develop novel privileged building blocks to address the
conserved binding epitope of aspartyl protease, we have
designed and synthesized two novel parent structures, the
aminohydroxysulfones and pyrrolidinemethaneamines. Both
skeletons were designed to address, across the entire family of
aspartyl proteases, the conserved catalytic diad and the
peptide recognition motif, which features the proteases for
Figure 1. Privileged skeletons frequently used to address the conserved
catalytic center in aspartyl protease. a) Classical transition-state analogues usually incorporated into peptidomimetics; b) newly discovered
scaffolds that address aspartyl proteases, in particular HIV protease.
[*] Dr. E. Specker,+ J. Bttcher,+ Dr. A. Heine, Prof. G. Klebe
Institut fr Pharmazeutische Chemie
Philipps-Universitt Marburg
Marbacher Weg 6, 35032 Marburg (Germany)
Fax: (+ 49) 6421-282-994
Dr. H. Lilie
Institut fr Biotechnologie
Martin-Luther-Universitt Halle-Wittenberg
Kurt-Mothes-Strasse 3, 06120 Halle/Saale (Germany)
Dr. A. Schoop
Boehringer Ingelheim, Vienna (Austria)
Dr. G. Mller
Axxima Pharmaceuticals AG, Mnchen (Germany)
Dr. N. Griebenow
Bayer AG, Elberfeld (Germany)
[+] Equally contributing authors
[**] The authors are grateful to Bayer AG, Wuppertal (Germany), for
financial support. The clone of the protease was kindly provided by
Prof. Helena Danielson, University of Uppsala, Department of
2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
classified as transition-state analogues, which closely resemble the tetrahedral intermediate formed along the reaction
pathway. The second generation of lead compounds departed
from closely imitating the transition state, and they frequently
tried to incorporate the position of the structural water.[6, 7]
This water mediates in case of the substrate analogue
inhibitors the protein–ligand interaction. In the late 1990s
researchers at Roche discovered by screening studies the
piperidine moiety as a novel privileged skeleton for addressing the catalytic center of aspartyl proteases.[8]
Our rational design of the aminohydroxysulfones was
inspired by four skeletons of already known inhibitors
targeting distinct aspartyl proteases (Scheme 1). Our aminohydroxysulfones are derived from the hydroxyethyleneamine
core of the HIV-protease inhibitor amprenavir (1),[9] by
replacing the sec-butyl-substituted nitrogen atom in the a-
DOI: 10.1002/anie.200462643
Angew. Chem. Int. Ed. 2005, 44, 3140 –3144
Scheme 1. Design of the aminohydroxysulfones based on the structural elements of known aspartyl protease inhibitors.
position with a carbon atom. The related sulfone core of the
renin inhibitor 2[10] lacks a substitution in a-position to
address the P1’ pocket. In the aminohydroxysulfones the
central scaffolds of the cathepsin D inhibitor 3[11] and the bsecretase inhibitor 4[12] are changed from an amide bond and a
keto group, respectively, to a methylenesulfone moiety. To
assess the potential of the aminohydroxysulfones as novel
lead structures for the inhibition of aspartyl proteases, we
decorated the central core with the previously optimized
substituents found in the inhibitors 1–4, which address the S2–
S2’ subsites.
In our design of the novel pyrrolidinemethaneamines we
intended to combine two key structural elements present in
the classical peptidomimetic substrate-analogue inhibitors
such as pepstatin (5, Scheme 2)[13, 14] and the newly discovered
type of cyclic amines (see Figure 1) to directly address the
pivotal position between both catalytic aspartates.
The hydroxy group of statin directly coordinates to the
two opposing aspartic acids. We decided to replace this
privileged core fragment by a pyrrolidine moiety capable of
being symmetrically extended towards the C- and N-terminal
Angew. Chem. Int. Ed. 2005, 44, 3140 –3144
subsites by using molecular scaffolds frequently incorporated
into substrate-analogue inhibitors. As apparent from the
multiple crystal structures with such peptidomimetics (e.g.
pepstatin 5, Scheme 2), important hydrogen bonds are formed
by the peptide-like backbone of the inhibitor to the flap
regions of these proteases. In the case of HIV protease, this
contact is mediated by a water molecule in hydrogen bonds
with the amide functions of Ile A50 and Ile B50; for cathepsin D the inhibitor is directly hydrogen-bonded to the amide
functions of Gly 79 and Asp 87. To evaluate the inhibitory
potential of the new pyrrolidinemethaneamines we decorated
the central pyrrolidine moiety with side chains already
optimized for the HIV-protease inhibitor amprenavir (1).
Diastereoselective syntheses yielded racemic mixtures of the
designed target compounds. The synthesis and characterization of the new inhibitors will be described elsewhere.
In the series of aminohydroxysulfones the racemate of 6
(Figure 2) exhibited a Ki value of 80 nm for HIV-protease
inhibition. Resolution of the racemic mixture revealed the
more potent enantiomer to have a Ki value of 45 nm. To
validate whether the binding of 6 follows the initial design
2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Table 1: Data collection and refinement statistics for the complexes with
6 and 7.
Resolution []
Space group
Cell dimensions []
Highest resolution shell []
Measured reflections
Independent reflections
Completeness of data [%]
Rsym [%]
Refined residues
Refined ligand atoms
Refined water molecules
Refined glycerol molecules
Refined Cl ions
Resolution in refinement []
Rcryst (F > 4sFo ; Fo)
Rfree (F > 4sFo ; Fo)
Mean B-factor [2]
(peptide chain A; B)
Main chain [2]
Side chains [2]
Ligand [2]
Water [2]
Glycerol molecules [2]
Cl ions [2]
Ramachandran plot
Most favored geometry [%]
Additionally allowed [%]
Generously allowed [%]
Disallowed [%]
HIV1 complex
with 6
HIV1 complex
with 7
a = 57.9
b = 85.8
c = 46.8
101 892
24 462
97.6 (79.2)[a]
12.8 (1.7)[a]
7.8 (45.2)[a]
16.7; 20.1
21.0; 24.7
17.2; 14.6
a = 51.9
b = 57.7
c = 62.2
109 029
29 849
97.5 (77.3)[a]
18.6 (1.6)[a]
6.3 (54.0)[a]
16.7; 18.1
22.0; 23.6
16.7; 19.0
12.8; 11.4
22.1; 18.0
20.6; 22.5
[a] Values in parentheses refer to the shell of highest resolution.
Scheme 2. a) Hydrogen bonds of pepstatin (3) in the active site of HIV
protease; b) hydrogen bonds of peptstatin (3) in the active site of cathepsin D; c) a pyrrolidinemethaneamine unit and its hydrogen bonds in
the active site of an aspartyl protease.
Figure 2. Structural formulae of (S,R,S)-6 and (R,R)-7 and the Ki values for inhibition of HIV1 protease.
concept and its stereochemistry, we determined the crystal
structure of the complex with HIV1 protease (Table 1).
2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Figure 3 a shows the binding mode of the more potent
enantiomer, (S,R,S)-6, superimposed with the complex of
amprenavir (1). The latter inhibitor served as one of the
references in our design concept. Both inhibitors bind with
their central OH group between the two aspartates, and either
their carbonyl or sulfonyl groups form hydrogen bonds to the
structural water of the protease. The S1 pocket (Leu A23,
Pro A81, Val A82, Ile A84) is occupied nearly identically by
the benzyl and bromobenzyl moieties in 1 and 6, respectively.
The bulky meta-bromine substituent in 6 extends into the
adjacent solvent environment, thus not provoking any
induced-fit adaptations. Similarly the S2’ subsite accommodates the para-aminophenyl (1) and the para-fluorophenyl (6)
groups in nearly indistinguishable fashion. However, the
fluorine substituent in 6 cannot form a hydrogen bond with
the adjacent Asp 30 as is observed in the complex with 1. The
sec-butyl substituents at the amino group of 1 and on the a-C
atom of the sulfone group of 6 orient their terminal methyl
groups in very similar fashion into the S1’ site (Leu A23,
Gly A27, Ile B81, Val B82, Ile B84) and achieve comparable
occupancy. The two inhibitors differ in the size of their
substitutents hosted in the S2 subsite (Ala B28, Val B32,
Ile B47, Ile A50). Amprenavir (1) exhibits a terminal tetrahy-
Angew. Chem. Int. Ed. 2005, 44, 3140 –3144
Figure 3. a) Superposition of the binding mode of (S,R,S)-6 with the
structure of 1 in complex with HIV1 protease. b) Binding mode of 7
with the protease showing the pyrrolidino nitrogen in the pivotal position between the two aspartates; the difference electron density (contour level 1s) is indicated as blue chicken wire. c) Superposition of the
complex of 6 and 7 depicting the strong distortion of the flap region
for the latter complex. d) Ligands 6 and 7 are mutually aligned to
show the addressing of subsites S2–S2’ of the protease. Sections of
the complex with 6 (e) and 7 (f) showing the occupation of S2 to S2’
by the side chains of the inhibitor; protein surface in white, ligand
surface in orange.
drofuran moiety, whereas 6 is decorated with a 2,6-dimethylphenoxy portion. The latter sterically more demanding
substituent requires more space in the S2 pocket[15] and
extends beyond the region filled by 1 into the bordering
solvent. Nevertheless, it achieves a better occupation of this
subpocket. Despite these differences, both complexes show
hardly any conformational differences and no induced-fit
adaptations of the enzyme.
The aminohydroxysulfone moiety of 6, developed from a
rather conservative design concept from already known
inhibitors, exhibits the expected binding mode. The 75-fold
decrease in affinity relative to that of amprenavir is supposedly due to the bromine atom of the meta-bromobenzyl
portion, which remains partly exposed to solvent in the bound
state. This incomplete desolvation is possibly detrimental to
binding. Furthermore, most likely the close contact of the
para-aminophenyl moiety in 1 is more favorable than the
short distance of the negatively polarized para-fluorophenyl
group in 6 to the carboxy group of Asp 30.
While our rather conservative design of the aminohydroxysulfone template closely resembles the underlying
parent substrate structure of an extended peptide chain with
protruding side chains, the pyrrolidinemethaneamine core
exhibits little if any structural resemblance to a peptide.
Angew. Chem. Int. Ed. 2005, 44, 3140 –3144
Docking suggested the binding to be similar to that of the
classical substrate-like inhibitors; however, significant crowding in the central region was indicated. Among the derivatives
synthesized, 7 (Figure 2), specifically decorated to address
HIV protease, shows a Ki value of 1.5 mm as a racemate.
Cocrystallization of HIV protease with racemic 7 revealed
that the R,R enantiomer binds to the enzyme (Table 1). As
expected, the nitrogen atom of the pyrrolidine ring, likely
protonated in the bound state, occupies the pivotal position
between the two aspartates. As expected, it is a perfect
substitute for the hydroxy group of, for example, 6 (Figure 3 b). Even though the side chains of (R,R)-7 are similar to
those of (S,R,S)-6, the pyrrolidine-based inhibitor adopts a
new and surprising binding mode. First of all, it repells the
structural water from the binding site and forms one direct
hydrogen bond to the backbone NH of Ile A50 with one of its
sulfoxy oxygens. The carbonyl group of the amide functionality remains unsatisfied without forming any further polar
contacts to the enzyme. As a consequence, the NH backbone
functionality of Ile B50 finds a hydrogen-bonding partner in
the carbonyl group of the amide bond of Ile A50 of the
neighboring polymer chain. This additional contact to the
neighboring flap loop stabilizes the dimer in the flap region.
Possibly this surprising further hydrogen bond is a surrogate
for the missing contact originally formed by the bridging
structural water which has been repelled from the complex.
This unusual geometry of the flap parallels a significant
distortion (up to 4.9 ) of the polymer chain in this region
compared to, for example, the complex formed with 6
(Figure 3 c).
The decorating side chains of 7 were designed to address,
like 6, the four subsites S2–S2’. However, the overall
orientation of the inhibitor in the complex is rotated relative
that of 6. To some extent similarity in the occupation of the S1
and S2’ subsites (Figure 3 d–f) can be recognized. The S1
pocket (Leu A23, Val A82, Val A84, Gly B48, Gly B49) accommodates the sec-butyl group of 7 and the meta-bromobenzyl
moiety of 6 in a similar fashion. Likewise, the phenylsulfonamide substitutent is hosted by the S2’ pocket (Ala A28,
Val A32, Ile A84, Ile B50). In case of 6, this site is filled by the
para-fluorophenyl group. The remaining two decorations, the
N-benzyl- and dimethylphenoxy group, are placed next to
each other; they extensively occupy the S1’ pocket and
penetrate into the bordering solvent environment. The Nbenzyl substituent barely permeates into the S2 pocket, which
remains virtually unoccupied. The site accommodating the
latter two side chains of 7 is bound by the residues Ile A50,
Val B32, Ile B47, Ile B54, Thr B80, Pro B81, Val B82, and
Ile B84.
Interestingly enough 6 and 7 crystallize in two distinct
space groups, P21212 and P212121, respectively. Accordingly, it
is questionable whether the observed differences are provoked by crystal packing. However, in structures of other
complexes found in either space groups, distortions of the flap
region similar to that observed for 7 were not observed. Next,
we compared the size of the ligand-surface portion that is
buried upon complex formation.[16] For 1 and 6 95 % of the
ligand surface is in close contact with the protein, whereas for
7 only 85 % of the surface is buried upon complexation. This
2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
significantly smaller value for 7 is in agreement with the
solvent exposure of the bulky dimethylphenoxy moiety in S1’.
The latter extends with one of its faces significantly into the
neighboring solvent area; inhibitor 6 places a short sec-butyl
group into this pocket largely avoiding solvent exposure.
Two new skeletons have been discovered to successfully
address the conserved binding motif of aspartyl proteases: the
aminohydroxysulfones and the pyrrolidinemethanamines.
Crystal structure analyses of the new privileged scaffolds
decorated with side chains intended for HIV-protease inhibition reveal their binding modes. The binding mode of the
aminohydroxysulfone was not surprising. However, the
pyrrolidine derivative adopts a surprising and rather unexpected binding mode. Detailed analysis of its bound conformation and the contacts formed with the protein will
provide further insights to explain the adopted binding mode.
Furthermore, it will guide the structure-based optimization of
this new inhibitor skeleton to improve its binding affinity.
Supposedly, the new inhibitor skeleton also provides challenging opportunities to combat resistance. The new type of
inhibitor achieves good inhibition although one subsite is left
virtually unoccupied. Thus, the virus has at least one less
option for mutational adaptation. Furthermore, the induced
perturbation in the flap region also occurs in an area most
likely out of range for the virus to mutate without significant
loss in functionality of its protease.
The present study is a lesson in drug design. The binding
mode of 6,[17] derived by a rather conservative design concept,
was correctly predicted including the assignment of stereocenters. Side chains transferred from other potent inhibitors
were actually found in the expected subsites. The opposite has
to be conceived for 7. In rational design of 7 the hydroxy
function in the transition-state analogue was correctly
replaced by a pyrrolidino nitrogen. Furthermore, the inhibitor
was equipped symmetrically with two polar acceptor groups
to address the flap water. However, this water is repelled from
the complex. The side chains were selected analogously to
those of 6, but they actually accommodate different subsites.
It has to be admitted that design and docking were unable to
predict the correct binding mode. Nevertheless, a potent class
of ligands has been created. While the binding geometry of
the heterocyclic scaffolds matches the prediction, the
mounted side chains deviate substantially from the expected
orientation. Without crystallography we would never have
detected this discrepancy. With some good luck, an elaborate
and tediously collected, however nonconclusive structure–
activity relationship could have been indicative. This clearly
points to the limits of computer design and stresses the
importance of experiments. Only if both go hand in hand, can
successful structure-based design be accomplished. The two
novel scaffolds can be added to the toolbox of medicinal
chemistry. They meet the stringent definition of privileged
structures: they address the target family-wide molecular
recognition commonalities and simultaneously offer a modular chemistry for peripheral decoration with required special
Keywords: drug design · HIV protease · inhibitors · proteins ·
structure determination
[1] R. E. Babine, S. L. Bender, Chem. Rev. 1997, 97, 1359 – 1472.
[2] S. Thaisvivongs, J. W. Strohbach, Biopolym. Pep. Sci. 1999, 51,
51 – 58.
[3] J. Vondrasek, A. Wlodawer, Proteins Struct. Funct. Genet. 2002,
49, 29 – 31.
[4] D. D. Richman, Nature 2001, 410, 995 – 1001.
[5] L. Romano, G. Venturi, S. Giomi, L. Pippi, P. E. Valensin, M.
Zazzi, J. Med. Virol. 2002, 66, 143 – 150.
[6] E. De Clercq, Int. J. Biochem. Cell Biol. 2004, 36, 1800 – 1822.
[7] A. Brik, C.-H. Wong, Org. Biomol. Chem. 2003, 1, 5 – 14.
[8] E. Vieira, A. Binggeli, V. Breu, D. Bur, W. Fischli, R. Gller, G.
Hirth, H. P. Mrki, M. Mller, C. Oefner, M. Scalone, H. Stadler,
M. Wilhelm, W. Wostl, Bioorg. Med. Chem. Lett. 1999, 9, 1397 –
[9] E. E. Kim, C. T. Baker, M. D. Dwyer, M. A. Murcko; B. G. Rao,
R. D. Tung, M. A. Navia, J. Am. Chem. Soc. 1995, 117, 1181 –
[10] G. Bolis, A. K. L. Fung, J. Greer, H. D. Kleinert, P. A. Marcotte,
T. J. Perun, J. J. Plattner, H. H. Stein, J. Med. Chem. 1987, 30,
1729 – 1737.
[11] E. C. Lee, E. K. Kick, J. A. Ellman, J. Am. Chem. Soc. 1998, 120,
9735 – 9747.
[12] A. K. Ghosh, G. Bilcer, C. Harwood, R. Kawahama, D. Shin,
K. A. Hussain, L. Hong, J. A. Loy, C. Nguyen, G. Koelsch, J.
Ermolieff, J. Tang, J. Med. Chem. 2001, 44, 2865 – 2868.
[13] P. M. D. Fitzgerald, B. M. McKeever, J. F. van Middlesworth,
J. P. Springer, J. C. Heimbach, C-T. Leu, W. K. Herber, R. A. F.
Dixon, P. L. Darke, J. Biol. Chem. 1990, 265, 14209 – 14219.
[14] E. T. Baldwin, T. N. Bhat, S. Gulnik, M. V. Hosur, D. Sowder,
R. E. Cachau, J. Collins, A. M. Silva, J. W. Erickson, Proc. Natl.
Acad. Sci. USA 1993, 90, 6796 – 6799.
[15] P. L. Beaulieu, P. C. Anderson, D. R. Cameron, G. Croteau, V.
Gorys, C. Grand-Matre, D. Lamarre, F. Liard, W. Paris, L.
Plamondon, F. Soucy, D. Thibeault, D. Wernic, C. Yoakim, J.
Med. Chem. 2000, 43, 1094 – 1108.
[16] M. L. Connolly, J. Appl. Crystallogr. 1983, 16, 548 – 558.
[17] The structure data of 6 and 7 are available from the Protein Data
Bank (PDB) under the codes 1XL2 and 1XL5, respectively.
Received: November 17, 2004
Published online: April 8, 2005
2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2005, 44, 3140 –3144
Без категории
Размер файла
479 Кб
hiv, two, mode, old, target, skeleton, unexpected, new, inhibitors, privilege, revisited, protease, binding
Пожаловаться на содержимое документа