Báo cáo khoa học: Optimization of P1–P3 groups in symmetric and asymmetric HIV-1 protease inhibitors pptx

Optimization of P1–P3 groups in symmetric and asymmetric
HIV-1 protease inhibitors
Hans O. Andersson
1
, Kerstin Fridborg
1
, Seved Lo¨ wgren
1
, Mathias Alterman
2
, Anna Mu¨ hlman
4
,
Magnus Bjo¨ rsne
4
, Neeraj Garg
2
, Ingmar Kvarnstro¨m
3
, Wesley Schaal
2
, Bjo¨ rn Classon
4
,
Anders Karle
´
n
2
, U. Helena Danielsson
5
,Go¨ ran Ahlse

´
n
5
, Ullrika Nillroth
5
, Lotta Vrang
6
,BoO
¨
berg
6
,
Bertil Samuelsson
4
, Anders Hallberg
2
and Torsten Unge
1
1
Institute of Cell and Molecular Biology, Uppsala, University, Sweden;
2
Department of Organic Pharmaceutical Chemistry,
Uppsala University, Sweden;
3
Department of Chemistry, Linko
¨
ping University, Sweden;
4
Department of Organic Chemistry,
Stockholm University, Sweden;

5
Department of Biochemistry, Uppsala University, Sweden;
6
Medivir AB,
Lunastigen 7, Huddinge, Sweden
HIV-1 protease is an important target for treatment of
AIDS, and efficient drugs have been developed. However,
the resistance and negative side effects of the current drugs
has necessitated the development of new compounds with
different binding patterns. In this study, nine C-terminally
duplicated HIV-1 protease inhibitors were cocrystallised
with the enzyme, the crystal structures analysed at 1.8–2.3 A
˚
resolution, and the inhibitory activity of the compounds
characterized in order to evaluate the effects of the individual
modifications. These compounds comprise two central
hydroxy groups that mimic the geminal hydroxy groups of a
cleavage-reaction intermediate. One of the hydroxy groups is
located between the d-oxygen atoms of the two catalytic
aspartic acid residues, and the other in the gauche position
relative to the first. The asymmetric binding of the two
central inhibitory hydroxyls induced a small deviation from
exact C2 symmetry in the whole enzyme–inhibitor complex.
The study shows that the protease molecule could accom-
modate its structure to different sizes of the P2/P2¢ groups.
The structural alterations were, however, relatively conser-
vative and limited. The binding capacity of the S3/S3¢ sites
was exploited by elongation of the compounds with groups
in the P3/P3¢ positions or by extension of the P1/P1¢ groups.
Furthermore, water molecules were shown to be important

binding links between the protease and the inhibitors. This
study produced a number of inhibitors with K
i
values in the
100 picomolar range.
Keywords: AIDS; drug; HIV; protease; X-ray.
An absolute necessity for the assembly and production of
infectious HIV-1 particles is the proteolytic processing of the
gag and gag-pol polyproteins into functional enzymes and
structural proteins [1–3]. The pol-gene-encoded protease,
which is responsible for this key function, has been selected
as a target for intervention of the HIV-1 infection with
antiviral drugs [4–7]. Numerous competitive inhibitors of
the protease have been prepared [8,9]. The Food and Drug
Administration (FDA) has approved six inhibitors: ampre-
navir, indinavir, lopinavir, nelfinavir, ritonavir and saqui-
navir [10]. The side effects of these inhibitors and the clinical
emergence of resistant mutants in HIV-1 means that new
protease inhibitors need to be developed [11,12].
The HIV-1 protease is a C2 symmetric homodimer
[13,14]. The protein monomer consists of 99 amino acids.
The active site, with the two catalytic aspartate residues
Asp25 and Asp125, is located at the interface between the
two monomers. Two b-hairpin structures, called ÔflapsÕ,are
positioned over the active site. They undergo structural
changes on binding of the inhibitor molecule. In the
unliganded protease structure, the conformation of
the flaps is open, thereby exposing the active site, whereas
in the ligand complex, the flaps form a roof over the active
site and the ligand. The flaps cover to a large extent the

bound ligand. This arrangement is advantageous for the
design of inhibitors, because it offers a large number of tight
interactions between the enzyme and the inhibitor. The
active site contains eight C2-symmetric subsites (S4, S3, S2,
S1, S1¢,S2¢,S3¢, and S4¢) [15]. These are the binding sites for
the P4, P3, P2, P1, P1¢,P2¢,P3¢,andP4¢ residues of an
octapeptide substrate [16]. Thus the N-terminal and
C-terminal parts of a bound substrate, or the corresponding
parts of the inhibitor, will interact with structurally similar
subsites. To exploit the C2 symmetry of the protease–
substrate complex, N-terminally or C-terminally duplicated
C2-symmetric inhibitors have been designed [17–20]. The
finding that a point mutation could completely abolish the
inhibitory activity of the symmetric compounds highlights
the weakness of this type of compound [21]. Drug-resistant
Correspondence to T. Unge, Institute of Cell and Molecular Biology,
BMC, Box 590, Uppsala University, SE-751 24, Uppsala, Sweden.
Tel.: + 46 18 471 49 85,
e-mail:
Enzyme: HIV-1 protease, POL_HV1B1 (P03366) (EC 3.4.23.16).
Note: The refined co-ordinates of HIV-1 protease in complex with
compounds 1–9 have been deposited in the RCSB Protein Data Bank
under the file names, 1EBW, 1EBY, 1EBZ, 1D4I, 1D4H, 1D4J,
1EC1, 1EC2 and 1EC3.
(Received 9 December 2002, revised 18 February 2003,
accepted 21 February 2003)
Eur. J. Biochem. 270, 1746–1758 (2003) Ó FEBS 2003 doi:10.1046/j.1432-1033.2003.03533.x
forms of the protease have been studied with respect to
kinetic and resistance properties [22]. New generations of
mainly asymmetric compounds have been developed with

high inhibitory activity against resistant variants of the
protease [23–25].
We here report crystallographic studies of C-terminally
duplicated C2-symmetric and asymmetric inhibitors in
complex with the HIV-1 protease. These compounds,
including the C2-symmetric ones, were found to bind in
an asymmetric fashion. K
i
values in the 100 picomolar range
were obtained for a number of these inhibitory compounds
[26,27].
Materials and methods
Expression of HIV-1 protease
The plasmid pBH10 containing the pol gene of the HIV-1
BH10 isolate was a gift from R. Gallo (National Cancer
Institute, Bethesda, MD, USA). The protease gene was
isolated by PCR with the upstream primer GAACA
TATGGCCGATAGACAAGGAACTGTATCC and the
downstream primer AGGGGATCCCTAAAAATTTAA
AGTGCAACCAATCTG. The annealing site for the
upstream primer corresponds to 12 amino acids before the
protease sequence. These extra amino acids were added to
make autolytic processing of the precursor protein possible,
enabling confirmation that the N-terminus was correct.
Through PCR, the protease DNA fragment was provided
with an NdeI restriction site at the 5¢ end and a BamHI site
at the 3¢ end. These sites were used for the ligation to the
pET11a expression vector. The Escherichia coli strains XL-1
and HB101 were used as hosts for cloning.
Protein was expressed in the E. coli strain BL21

(DE3). Bacteria were grown in Luria–Bertani medium
to an D
550
of 1.0 before induction with 0.5 m
M
isopropyl
thio-b-
D
-galactoside. Cells were harvested after 3 h of
induction.
Purification of HIV-1 protease
The chromatographic steps were performed at 5 °C. SDS/
PAGE was used after each chromatographic step to
monitor the purification. Cells were suspended in lysis
buffer (20 m
M
Tris/HCl, pH 7.5, 10 m
M
dithiothreitol,
1m
M
phenylmethanesulfonyl fluoride) and lysed in a
French press. The lysate was centrifuged for 30 min at
12 100 g. The insoluble inclusion body fraction, which
contained more than 90% of the expressed material, was
dissolved in buffer (8
M
urea, 20 m
M
Tris/HCl, pH 8.5,

10 m
M
NaCl, 10 m
M
dithiothreitol, 1 m
M
EDTA) and
centrifuged for 1 h at 48 200 g.
The supernatant was applied to a POROS Q column
(Roche). The flow-through fraction was collected and
diluted to a final protein concentration of 0.3 mgÆmL
)1
.
Refolding was performed by dialysis against 20 m
M
sodium
phosphate buffer, pH 6.5, containing 10 m
M
dithiothreitol
and 1 m
M
EDTA. The refolded protein was diluted with an
equal volume of 50 m
M
Mes, pH 6.5, containing 1 m
M
dithiothreitol and 1 m
M
EDTA, applied to a POROS HS
column (Roche), and eluted with a linear gradient of

0–0.6
M
NaCl in Mes buffer. The pooled fractions were
precipitated with (NH
4
)
2
SO
4
. The precipitate was collected
by low-speed centrifugation and dissolved in 50 m
M
Mes,
pH 6.5, containing 10 m
M
dithiothreitol, 100 m
M
2-mercap-
toethanol and 1 m
M
EDTA. The solution was desalted on a
PD-10 column (AP Biotech AB, Uppsala, Sweden) and
concentrated by ultrafiltration with Centricon Centrifugal
Filter Units to 2 mgÆmL
)1
.
Enzyme activity and inhibition studies
Enzyme activity/inhibition studies were performed as des-
cribed by Nillroth et al. [28]. The method includes active-site
titrations. Briefly, a fluorimetric assay was used to determine

the effects of the inhibitors on HIV-1 protease. This assay
used an internally quenched fluorescent peptide substrate,
DABSYL-c-Abu-Ser-Gln-Asn-Tyr-Pro-Ile-Val-Gln-
EDANS (Bachem, Bubendorf, Switzerland). The measure-
ments were performed in 96-well plates with a Fluoroscan
plate reader (Labsystems, Helsinki, Finland). Excitation
and emission wavelengths were 355 nm and 500 nm,
respectively.
Anti-HIV activity was assayed in vitro in MT4 cells using
the vital dye XTT to monitor the cytopathogenic effects
[29].
Crystallization
Crystallization was performed at 4 °C with the hanging
drop vapour-diffusion method. Protease (5 lL) at a
concentration of 2.0 mgÆmL
)1
in buffer consisting of
50 m
M
Mes, pH 6.5, 10 m
M
dithiothreitol and 1 m
M
EDTA was mixed with an equal volume of the reservoir
solution. The reservoir solution contained 50 m
M
Mes,
pH 5.5, and 0.5
M
NaCl. The drops were microseeded

after 2 days with seeds from protease/inhibitor crystals
belonging to space group P2
1
2
1
2. Crystals appeared after
1 week, and grew to a final size of 0.3 · 0.3 · 0.05 mm in
3–4 weeks.
Data collection and processing
X-ray data were recorded on MAR-imaging plates on the
synchrotron beam lines 9.5 DRAL at Daresbury, UK,
DL41 and DW32 at Lure, France, and I711 at MAX-lab
Lund, Sweden. The programs
DENZO
and
SCALEPACK
were
used for processing and scaling [30,31]. A summary of data
collection statistics is given in Table 1.
Structure refinement
Refinement was performed using the program package
CNS
[32]. The protease model co-ordinates from 1AJV were used
for molecular replacement calculations. The starting model
was refined with rigid-body refinement and simulated
annealing. The difference Fourier map (F
o
–F
c
) clearly

showed the position and orientation of the inhibitor
together with a large number of water molecules. The
inhibitor was built into the electron density with the help of
the program
O
[33]. Water molecules were added to the
structures determined from the difference Fourier maps at
chemically acceptable sites. Only solvent molecules with B
valueslessthan50A
˚
2
were accepted. Several cycles of
Ó FEBS 2003 Optimization of HIV-1 protease inhibitors (Eur. J. Biochem. 270) 1747
minimization, simulated annealing, and B-factor refinement
were performed for each complex, accompanied by manual
rebuilding. The R
crystal
and R
free
factors were used to
monitor the refinement [34,35]. The refinement statistics are
showninTable1.
Graphics
All the figures were drawn with the programs
SWISS
-
PDB-
VIEWER
[36] ( and
POV

-
RAY
( />Results and Discussion
Inhibitor properties
The linear C-terminally duplicated inhibitors in this study
encompass a central six-carbon skeleton derived from
L
-mannaric acid (Table 2). Five of the inhibitors [1,2,7–9]
are chemically C2 symmetric. Seven of these nine
compounds exhibit K
i
values in the nanomolar or low-
nanomolar range, with antiviral effects (ED
50
) ranging from
>75 l
M
(compound 6) to 0.04 l
M
[7].
Crystallographic calculations
The crystal structures of the nine inhibitors in complex
withHIV-1proteaseweredeterminedtohighresolution
(Table 1). All the complexes were crystallized in the
orthorhombic space group P2
1
2
1
2. The asymmetric unit
contained the whole protease molecule. This means that

even though the crystal packing contains twofold axes,
these do not impose twofold symmetry on the inhibitor–
protease complex. In the database of HIV and simian
immunodeficiency virus protein–complex structures
( there are examples of
structures in which the inhibitors are uniquely oriented,
but also structures in which the corresponding electron
density represents two orientations of the inhibitor [37,38].
For the structures presented here, the electron densities of
the inhibitors, especially the densities of the side groups of
the asymmetric compounds, indicate a unique orientation
of the protease–inhibitor complex in the crystal lattice
(Fig. 1).
The orientation of the two central hydroxy groups
influences the positioning of the inhibitor in the active site
and thereby the structure of the protease.
Table 1. Crystallographic structure determination statistics. R
merge
¼ SjI
i
)<I>/SI
i
,whereI
i
is an observation of the intensity of an individual
reflection and <I> is the average intensity over symmetry equivalents. R
crystal
¼ SjjF
o
j)jF

c
jj/SjF
o
j,whereFoandFc are the observed and
calculated structure factor amplitudes, respectively. R
free
is equivalent to R
crystal
but calculated for a randomly chosen set of reflections that were
omitted from the refinement process. Ideal parameters are those defined by Engh & Huber [55].
123456789
Data collection details
Space group P2
1
2
1
2P2
1
2
1
2P2
1
2
1
2P2
1
2
1
2P2
1

2
1
2P2
1
2
1
2P2
1
2
1
2P2
1
2
1
2P2
1
2
1
2
Wavelength (A
˚
) 0.920 1.386 0.970 1.375 1.375 1.375 0.970 0.958 0.958
No. of crystals 2 14647364
Cell dimensions (A
˚
)
a ¼ 59.04 59.15 58.84 58.62 58.46 58.78 58.12 58.48 58.89
b ¼ 86.83 86.98 86.66 86.32 86.31 86.52 86.11 86.67 86.70
c ¼ 46.98 47.16 46.87 46.69 46.57 46.58 46.11 46.52 46.70
d

min
(A
˚
) 1.80 2.30 2.00 1.81 1.81 1.81 2.10 2.00 1.90
No. of observations 42993 28849 55336 95666 61625 92684 35374 44926 62391
No. of unique reflections 16613 10685 15626 21735 20047 21687 13120 15612 18784
Completeness (%) 69.7 93.8 95.9 97.5 90.5 97.2 93.1 92.9 95.7
R
merge
(%) 9.4 4.6 7.5 11.1 8.9 11.2 7.5 11.6 8.1
Reflections I >2 r (%) 63.0 84.1 83.4 84.6 79.7 80.5 77.1 72.5 90.4
Reflections I >2 r in
highest resolution shell (%)
9.4 74.2 70.1 60.8 52.5 51.1 63.1 61.0 79.2
Bin resolution (A
˚
) 1.86–1.80 2.40–2.30 2.07–2.00 1.87–1.81 1.87–1.81 1.87–1.81 2.18–2.10 2.07–2.00 1.97–1.90
Refinement statistics
Resolution range (A
˚
) 24.6–1.81 24.0–2.30 24.6–2.01 27.9–1.81 22.6–1.81 15.0–1.81 24.3–2.10 24.7–2.00 24.3–2.10
R
cryst
(%) 19.0 18.1 18.0 18.8 19.4 20.1 20.4 20.7 20.4
R
free
(%) 21.8 20.0 20.3 222.4 23.4 23.1 23.7 23.9 23.7
No. of atoms 1677 1662 1647 1697 1694 1655 1676 1688 1688
Mean B factors (A
˚

2
)
Protein 19.6 20.2 20.2 20.2 20.4 25.6 18.3 17.4 23.4
Inhibitor 16.8 12.9 18.1 16.5 20.0 23.8 14.0 13.4 23.3
All 20.4 20.5 21.0 21.1 21.9 26.3 19.8 18.7 19.8
Deviation from ideality
Bond lengths (A
˚
) 0.006 0.008 0.006 0.006 0.006 0.006 0.007 0.007 0.007
Angles (°) 1.3 1.2 1.3 1.2 1.2 1.2 1.4 1.3 1.4
Dihedrals (°) 25.7 25.4 25.5 25.4 25.3 25.2 25.4 25.4 25.4
Impropers (°) 0.79 0.77 0.76 0.79 0.82 0.85 0.77 0.75 0.77
1748 H. O. Andersson et al.(Eur. J. Biochem. 270) Ó FEBS 2003
The inhibitors in this study (excep compound 4) contain
two central hydroxy groups that mimic the geminal hydroxy
groups of an intermediate in the cleavage reaction [39]. A
completely symmetric arrangement of these inhibitors in the
active site would require a twofold symmetrical arrange-
ment of the two hydroxyls and consequently identical
binding patterns to the catalytic residues (Asp25/Asp125).
However, a symmetrical arrangement of the central
hydroxyls was not found for either the symmetric or the
asymmetric compounds. Instead, one of the hydroxy groups
was placed at hydrogen-bonding distance (2.7 A
˚
) between
the d-oxygen atoms of the two aspartic acid residues (Figs 1
Table 2. Inhibitor structure, enzyme inhibition and antiviral activity in MT4-cell culture. ED
50
values for reference substances tested in the same

assay: ritonavir (ED
50
0.06 l
M
), indinavir (ED
50
0.06 l
M
), saquinavir (ED
50
0.01 l
M
) and nelfinavir (ED
50
0.04 l
M
). Compound 4 was synthesized
with only one central hydroxy group.
Compound no. A B C K
i
(n
M
)ED
50
(l
M
)
1
0.90 1.24
2 0.20 0.11

3 0.40 0.15
4 1.40 1.49
5 0.10 3.39
6 4.40 >75
7 1.20 0.04
8 0.10 2.47
9 0.92 0.75
Ó FEBS 2003 Optimization of HIV-1 protease inhibitors (Eur. J. Biochem. 270) 1749
and 2), and the other hydroxy group was in gauche position
relative to the first, hydrogen-bonded to one of the
aspartates (2.8 A
˚
) and had van der Waals interaction with
Cb of Ala28 (Ala128) at a distance of 3.9 A
˚
.Thesame
arrangement has been observed for other C2-symmmertric
diol-containing inhibitors [40]. As a direct consequence of
this arrangement, the twofold symmetry axis of the inhibitor
will not coincide with the protease twofold axis. Despite the
ability of the protease to adapt to differences in the size of
the inhibitor side groups, a comparison between the left and
right binding sites reveals small but significant differences in
bond lengths. This is exemplified by the observation that the
distance between the P1 benzyl group of compound 1 to
Arg8 was 4.0 A
˚
whereas the distance from P1¢ to Arg108
was 3.6 A
˚

. The significance of this discrepancy is apparent
from a comparison with the HIV-1 protease–inhibitor
structure 4phv, which contains an inhibitor homologous to
compound 2 but with only one central hydroxy group and
which is also shorter by one carbon in the central part of the
inhibitor [41]. In this structure, the distances between the P1/
P1¢ benzyl groups and Arg8/Arg108 were 4.0 and 3.9,
respectively. Thus, like the natural protein substrate, the
inhibitors, including the symmetric ones, tend to bind in an
asymmetric fashion [40,42]. Compounds 3, 5 and 6 were
made in order to exploit the asymmetric binding.
Electron density maps revealed a preference for the same
arrangement of the central hydroxyls for compounds 1, 3
and 5–8 (see Figs 1, 2 and 4). Compound 4, with only one
central hydroxyl, had it arranged as in compounds 1,3 and
5–8. Preference for the opposite arrangement was found for
compound 9 (see Fig. 6). However, the central hydroxyls
were not completely uniquely arranged in any of the
structures of compounds 1–3 or 5–9. Degrees of uniqueness
varied from 75 to 90%. For the asymmetric compound 3,
the density indicated a 90% unique orientation (Fig. 1).
Small but significant deviations from exact C2 symmetry
were also observed in parts of the protease structure
connected with movement of the flaps. Application of a
Fig. 1. Orientation of the inhibitor in the active site and arrangement of the central vicinal hydroxyls (stereo view). The figure shows the structure of the
asymmetric compound 3 as it is arranged in the HIV-1 protease active site. The electron density map indicates a unique orientation of the inhibitor
and the whole protease–inhibitor complex in the crystal lattice. The density also indicates an  90% unique orientation of the central vicinal
hydroxy groups in the complex with this compound. The Fo–Fc electron density map was calculated at 2.0 A
˚
resolution with the inhibitor

compound omitted, and contoured at 2.5 r.
Fig. 2. Positioning of the inhibitor in the active site and hydrogen-bond network. One of the inhibitor hydroxyls has extensive contacts with the
catalytic Asp25/Asp125. The hydrogen-bond distances are short (2.7–2.8 A
˚
). The gauche hydroxy group is hydrogen-bonded to one of the catalytic
aspartate residues. Gly27/Gly127 contribute to the active-site hydrogen-bond network by donation of hydrogens via the main-chain amide groups.
These two compounds 1 and 2 represent the two groups of inhibitors in this study. Compound 1 (A) has 10 and compound 2 (B) has 8 hydrogen-
bond donors/acceptors. In the latter case, two water molecules remain co-ordinated to the G48/G148 carbonyl groups after complex formation.
1750 H. O. Andersson et al.(Eur. J. Biochem. 270) Ó FEBS 2003
least-squares superimpositioning of chain A on B showed
an average difference of 0.9 A
˚
between Ca atoms in the flap
region and in the b-strands with amino acids 15–18, 37–41,
59–64 and 71–74 (not shown). Consequently, significant
differences in the binding pattern were observed for the
chemically identical P1/P1¢,P2/P2¢ and P3/P3¢ groups. The
asymmetric positioning of the inhibitor in the protease
substrate-binding site, especially the interactions between
the inhibitor and the flaps, may be the major reason for
these structural differences. However, crystal-packing inter-
actions, which in this space group are different for the two
peptide chains, may contribute to the asymmetry of the
protease.
Interactions between the central inhibitor hydroxy
groups and the active-site aspartate residues
The oxygen atom of the hydroxy group, which is positioned
symmetrically above the catalytic Asp25/Asp125, is hydro-
gen-bonded to the carboxylate oxygens of the aspartates at
distances between 2.63 and 2.89 A

˚
(Fig. 2, Table 3). This
position is 1.4 A
˚
away from the position of the catalytic
water as suggested by the structure of the hydrated
difluoroketone inhibitor A79285 (1DIF) [39]. The active
site of the inhibitor complex is rich in hydrogen bonds. In
addition to the abundant hydrogen-bond network involving
the inhibitor, the main-chain amide nitrogen atoms of
Gly27 and Gly127 are hydrogen-bonded to the carboxylate
oxygens Asp25 OD1 and Asp125 OD1, respectively, at
distances of 2.8–2.9 A
˚
. The position and orientation of
G27/G127 is to a large extent determined by the stabilizing
hydrogen-bond network, which involves the second mem-
ber of the catalytic triad T26/T126 [43]. The close packing of
the central hydroxy group between the aspartic carboxylates
is a common property of not only the compounds in this
series but also of other linear inhibitors [44]. The positioning
of the inhibitor in the active site results in a tight interaction
between the inhibitor’s hydroxy group and the catalytic
residues. This close positioning of the inhibitor’s hydroxyl to
the carboxylate oxygens is not only caused by the attraction
between these groups, but also by the interactions between
the inhibitor’s side chains and the S1-S3 site residues. This
was revealed by a comparison with the position of a
homologous inhibitor that lacked the co-ordinating central
hydroxy group (unpublished data).

Binding contribution by the gauche hydroxy group
The central gauche hydroxy group of these compounds
mimics the gauche position of the hydrated peptide
carbonyl in a cleavage-reaction intermediate (Fig. 2). Its
position is an average distance of 0.3 A
˚
from the gauche
hydroxyl in compound A79285, which mimics a hydrated
peptide intermediate [39]. To evaluate the binding contri-
butions of this gauche hydroxy group, compound 4 was
synthesized, in which the gauche hydroxy group was
replaced with a hydrogen atom. Superimposition of the
compound 2 and 4 protease–inhibitor complexes with
lsq_explicit in the program
O
resulted in an r.m.s.d. value
of 0.18 A
˚
for the protein Ca atoms. The same magnitude-
of-distance deviations were observed between identical inhi-
bitor atoms. There were, however, asymmetric discrepancies
in the P1/P1¢ positions. The positions of P1 atoms in
compound 4 agreed within 0.1 A
˚
with the corresponding
atoms of compound 2, as opposed to atoms of P1¢,which
were within 0.3 A
˚
. This led us to conclude that the hydroxy
group could be exchanged for a hydrogen atom without any

major effects on the positioning of the inhibitor in the active
site. However, the modification had a negative effect on the
K
i
value, which was seven times higher for compound 4
(1.4 n
M
) than for compound 2 (0.2 n
M
). Thus, the gauche
hydroxy group contributes significantly to the binding
capability. The contribution, which is complex, includes
hydrogen-bonding to Asp25/Asp125, van der Waals
Table 3. Hydrogen bonds between the protease and the inhibitor com-
pounds 1, 2, 3 and 5.
Residue Atom Atom
Protease–inhibitor
distance (A
˚
)
Compound 1
25 Asp Od1 O27 2.71
25 Asp Od2 O27 2.89
27 Gly O N18 3.13
29 Asp N O24 2.88
48 Gly O N25 2.93
125 Asp Od1 O27 2.74
125 Asp Od2 O27 2.68
125 Asp Od2 O28 2.70
127 Gly O N8 3.12

129 Asp N O14 2.92
148 Gly O N15 2.96
Compound 2
25 Asp Od1 O6 2.64
25 Asp Od2 O6 2.85
27 Gly O N1 3.13
29 Asp N O50 3.02
125 Asp Od1 O6 2.90
125 Asp Od2 O6 2.65
125 Asp Od2 O8 2.71
127 Gly O N12 3.21
129 Asp N O60 3.07
129 Asp Od2 O60 2.86
Compound 3
25 Asp Od1 O6 2.75
25 Asp Od2 O6 2.83
27 Gly O N1 3.16
29 Asp N O46 2.89
48 Gly O N47 2.92
125 Asp Od1 O6 2.72
125 Asp Od2 O6 2.64
125 Asp Od2 O8 2.70
127 Gly O N12 3.13
129 Asp N O60 3.12
Compound 5
25 Asp Od1 O32 2.63
25 Asp Od2 O32 2.88
27 Gly O N44 3.19
125 Asp Od1 O32 2.82
125 Asp Od2 O02 2.66

125 Asp Od2 O32 2.63
127 Gly O N14 2.97
129 Asp N O24 3.10
129 Asp Od2 O24 2.97
Ó FEBS 2003 Optimization of HIV-1 protease inhibitors (Eur. J. Biochem. 270) 1751
interactions with Ala28, as well as energy contributions
from restriction of the rotational freedom around the C4
carbon.
Overall hydrogen-bond pattern between the inhibitors
and the protease
The compounds in this series contain several hydrogen-
bond donors and acceptors (Fig. 2 and Table 3). The
number of hydrogen bonds between the protease and the
inhibitors varies between 7 and 12 (Table 4). The fact that
these inhibitors are all based on the same skeleton is
reflected in the conserved hydrogen-bond pattern between
the different compounds (Table 3). The asymmetric substi-
tutions in the P2 positions do not significantly alter the
pattern of the retained groups. All polar groups in the
inhibitors, except the ether link in P1/P1¢, are involved in
hydrogen bonding to the protein, either directly or indirectly
through water molecules. Compared with the substrate-like
peptide bond in the P2/P3 positions of compounds 1, 3 and
7–9, substitutions with the indanyl and benzyl groups in P2
led to loss of the hydrogen bond to the carbonyls of Gly48.
As expected, because of its position close to the entrance of
the binding site, a water molecule co-ordinates Gly48 in
these complexes (Fig. 2B). According to a study by Ala
et al. [45], hydrogen bonds contribute less than hydrophobic
interactions to the binding energy. However, the overview

of the number of hydrogen bonds and other contacts in
Table 4 indicates that hydrogen bonds, as well as the close-
packing interactions, contribute to the inhibitory efficacy of
the compounds (Table 4). The variations in bond distances
indicate potential variation in energy content in the
established hydrogen bonds (Table 3). Hydrogen bonds to
co-ordinating water molecules are not included in Tables 3
and 4. These are discussed below.
Optimization of the P1/P1¢ side chains
The HIV-1 protease cleaves the precursor protein at nine
positions. In seven of these, the P1 or P1¢ amino acid is a
phenylalanine or tyrosine. By homology, it is natural to use
benzyl groups as P1/P1¢substituents on the C2/C5 carbons,
as is the case in a large number of inhibitors. In this series of
compounds, however, benzyloxy groups are used. Through
the ether linkage the side chain is elongated by 1.45 A
˚
.This
permits positioning of P1/P1¢ in a position that is almost
identical with that in the homologous compound L-700,417,
which has only one central hydroxyl and P1/P1¢ benzyl side
chains (4PHV [41]) (Fig. 3). Even though the ether linkage
increases the potential degrees of rotational freedom of the
P1/P1¢ side chains, this is not the outcome. Rather, the
inhibitor becomes more compact. The elongation enables
close packing of the side chain to the inhibitor backbone
and to P2/P2¢ [27]. The position of the P1/P1¢ side chains is
conserved among these compound complexes of com-
pounds 1–8. The oxygen atom in the ether linkage shows
only weak interaction with the protein. The closest atoms

are the Od2 atoms of Asp25/Asp125 at distances of 3.4/
3.7 A
˚
.TheP1/P1¢ benzyloxy side groups are within a 3.7-A
˚
radius of the S1/S1¢ site atoms O of Gly27/Gly127, CG of
Table 4. Summary of the inhibitor/protease interactions. Buried surface area was calculated with programs within the CNS package [32]. Hydrogen
bonds were calculated with a maximum distance of 3.5 A
˚
between acceptors and donors. An atom/pair distance of less than 3.9 A
˚
was used as
criterion for a contact.
Compound
Molecular
mass (Da)
Buried surface
area (A
˚
2
)
No. of hydrogen
bonds
No. of inhibitor/
protease contacts K
i
(n
M
)
1 642.791 1400.48 11 39 0.90

2 652.743 1426.56 10 52 0.20
3 633.740 1369.16 10 52 0.13
4 636.743 1394.59 8 48 1.40
5 610.705 1398.39 9 51 0.10
6 663.141 1391.54 7 52 4.40
7 778.977 1528.61 11 53 1.20
8 768.908 1572.11 11 47 0.10
9 768.908 1551.03 12 61 0.92
Fig. 3. Comparison between the 3D structures of our compound 2 and compound L-700, 417 (in blue) (4PHV [41]) (stereo view). The two compounds 2
and L-700, 417 have similar C2-symmetric scaffolds, but the scaffold of L-700, 417 is one carbon atom shorter. Through the elongation with an ether
linkincompound2,theP1/P1¢ benzyl groups superimpose well. The two compounds form notably compact structures when bound to the protease.
1752 H. O. Andersson et al.(Eur. J. Biochem. 270) Ó FEBS 2003
Pro81/Pro181, CG1 of Val82/Val182, and CD1 of Ile84/
Ile184. Within a distance of 4.0 A
˚
is CD2 of Leu23/Leu123.
Two water molecules in each side of the protease molecule
interacts with P1/P1¢. In the complex with compound 9, the
benzyloxy group is shifted upward by  1.5 A
˚
on one of the
sides leading to loss of the interaction with O of Gly127 but
instead making contact with C and O of Gly149.
Optimization of the P2/P2¢ side chains
The chemical properties of the natural substrate P2/P2¢
amino-acid residues vary more than these of the P1/P1¢
residues. The S2/S2¢ sites contain the hydrophobic amino
acids Ala28/Ala128, Val32/Val132, Ile47/Ile147, Ile84/
Ile184 and Ile150/Ile50, as well as the polar Asp30/Asp130
[16]. The P2/P2¢ side chains of the compounds studied here

are manly hydrophobic. Thus, the lipophilic groups valinyl
(the side chain of valine) [3,7–9], isoleucinyl (the side chain
of isoleucine) [1], indanyl [2–6], benzyl [5] and 2-chloro-6-
fluorobenzyl [6] were explored (Table 2). The interacting
S2/S2¢ ligands were Ala28/Ala128, Asp30/Asp130, Val32/
Val132, Ile47/Ile147 and Ile84/Ile184. Even though the sizes
of the P2/P2¢ side chains differ significantly and penetrate
the binding site with a difference of 2.1 A
˚
, the positions of
the contacting S2/S2¢ amino acids are relatively conserved
except for those in the 30s and 80s loops. The side chain of
Asp30 and Val32 moves as much as 2.0 A
˚
and 0.6 A
˚
,
respectively, to accommodate the benzyl and 2-chloro-6-
fluorobenzyl groups of compounds 5 and 6 (Fig. 4). Fig. 5
shows a summary of the shifts in the Ca positions. In
addition to the shifts around amino-acid position 30,
significant shifts of the order of 0.2–0.7 A
˚
are observed
around positions 18, 67 and 81. Only the peptide chain
harbouring the S2 site was used in the calculations.
The change in position of Asp30 leads to small changes
in the hydrogen-bond networks involving this residue. A
comparison with compounds 1, 2 and 3 indicate how the
side chains valinyl, isoleucinyl and indanyl gradually fill

out the S2/S2¢ sites, resulting in the expected improvements
in the K
i
values (Table 2). However, the chlorine-substi-
tuted and fluorine-substituted benzyl group of compound
6 created a P2 group that was too large, as reflected by the
high K
i
value of 4.4 n
M
. Because of steric hindrance by the
main-chain carbonyl group of Gly48, the Cl atom had to
be positioned ÔinwardÕ against the S2 site and in close
contact with Cb of V28 and Cd of I84. The close packing
against A28 forces this P2 side chain upward by 0.6 A
˚
.As
a consequence, the hydrogen bond with the inhibitor
backbone amide and Gly27 is broken and replaced by
co-ordination of a water molecule. This water molecule is
co-ordinated by the NH group of the inhibitor, O of
Gly27, and N of Asp29. The close packing between the
chlorine atom and Cd of I84 displaces the phenyl ring
 2A
˚
closer to the 30s loop compared with the phenyl
group of compound 5 and the six-carbon ring of the
indanyl group of compound 2. This leads to repositioning
of the 30s loop and Asp30 by 0.4 and 1.0 A
˚

, respectively,
compared with their positions with the smaller P2¢ groups.
The electronegative fluorine is within van der Waals radius
(3.3 A
˚
) of the similarly electronegative carbonyl oxygen of
Gly84. The most serious problems with this P2 group are
the breaking of the hydrogen bond and repulsion between
the dipoles.
In compound 5, one of the indanyl groups was exchanged
for a benzyl group to investigate the requirements for
optimal asymmetric binding to the S2/S2¢ site. The benzyl
group is coplanar with the six-carbon ring of the indanyl
group but its position is shifted by  0.3 A
˚
. Thus, the
hydroxy group is not necessary for the orientation of the
plane of the aromatic moiety. A bound water molecule
(B ¼ 39.6) positioned in contact with the phenyl planes of
the P1¢ and P2¢ side chains and 0.5 A
˚
from the position of
the indanyl oxygen atom replaces the function of the
indanyl hydroxy group as hydrogen-bond donor (Fig. 6).
Interestingly, compounds 5 and 2 have about the same K
i
values, which indicates the value of the bridging water
molecule for specific and efficient interactions between
ligand and enzyme.
Calculation of accessible area in the protease buried by

the compounds in the complex showed that compound 1,
with a relatively low molecular mass, buried an area of
equivalent size to the bigger compound 2 (Table 4). A
comparison of the number of contacts and hydrogen bonds
for these compounds revealed inefficient utilization of area
in the case of compound 1. This was also reflected in a
higher K
i
value for compound 1.
Fig. 4. Accommodation of S2/S2¢ residue to different P2/P2¢ groups (stereo view). Whereas the positions and orientations of Ala28, Val32, Ile47,
Ile84 are conserved, Asp30 and Val32 as well as the entire loop containing these residues adopt to the different P2 side chains as much as 2.1 A
˚
for
the side chain and 0.2–0.7 A
˚
for the main-chain atoms. Colour code: light blue (1), magenta (2), dark blue (3), black (5) and red (6).
Ó FEBS 2003 Optimization of HIV-1 protease inhibitors (Eur. J. Biochem. 270) 1753
Optimization of the P3/P3¢ side chains
In addition to extension of the inhibitor by addition of a
new peptide bond and P3/P3¢ ligands, the S3/S3¢ sites can
be reached by substituents on the P1/P1¢ ligands. To test
this, compounds 7–9 were synthesized. Compounds 7 and
8 are substituted with thienly and pyridyl groups at the
para position of the P1/P1¢ benzyl groups, and inhibitor 9
contains pyridyl groups in the P3/P3¢ positions (Fig. 7).
The structures of compounds 7 and 8 have previously
been briefly described by Alterman et al. [46]. Compound
8 has a significantly better binding parameter
(K
i

¼ 0.1 n
M
) than compounds 7 and 9 (1.2 n
M
and
0.92 n
M
, respectively) (Table 2). The electron density of
the thienyl ring of compound 7 was low, which indicated
undefined binding, whereas the densities for the pyridyl
rings of compounds 8 and 9 were well defined. The
significantly lower K
i
value for compound 8 is explained
by the van der Waals interactions of the pyridyl rings with
Phe53/Phe153, Pro181/Pro81 and Gly48/Gly148, where
Pro packs against the plane of the pyridyl ring. Further-
more, the pyridyl nitrogens are co-ordinated to Arg8/108
via water molecules (Fig. 7). The pyridyl ring of com-
pound 9 interacts with Gly48/148 and Arg8/108. How-
ever, the orientations of the two pyridyl rings as well as
their binding patterns are different. There are not, as for
compound 8, any apparent co-ordinating ligands to the
pyridyl nitrogens. Only in one side of the S3 sites is it
possible to find water molecules within binding distances.
Loosely bound water molecules at the entrance to the S3/
S3¢ sites are displaced by the extending pyridyl and thienyl
rings.
Water molecules located in the active site
Several water molecules are located in the active site and

bound to the protein as well as to inhibitor compounds.
Similar to other linear HIV-1 protease inhibitors, in these
inhibitor complexes also a structural water molecule acts as
a link between the inhibitor and the flap residues Ile50/
Ile150. Considerable effort has been expended on designing
inhibitors that include this water molecule in their structure
[47–50]. However, in these compounds, co-ordinating
hydrogen-bond-accepting carbonyl groups were designed.
Fig. 5. Ca plot analysis [56] of the inhibitor complexes 1–9 (here labelled A-J). The mRMSD panel shows the combined distance deviation for all
pairs of A subunits (containing the S2 site). Values were calculated with LSQMAN [57].
1754 H. O. Andersson et al.(Eur. J. Biochem. 270) Ó FEBS 2003
The hydrogen-bond distances between the water molecule
and the Ile50/Ile150 N atoms are in the range 2.7–2.9 A
˚
.
The tetrahedral arrangement of the ligands to this water
molecule is slightly distorted. This water oxygen atom has a
low-temperature factor (15 A
˚
2
).
Two water molecules are positioned between the P1/P2
arms and hydrogen-bonded by the P3 carbonyl oxygen
atom and at the corresponding symmetry-related sites
(Fig. 2). Water molecules at these positions have been
found in several inhibitor complexes [51]. Also these water
molecules have low-temperature factors. Their role is not
clear, but their position, just below the inhibitor, indicates
that water molecules may serve as lubricant between
movable parts in a protein molecule, increase the promis-

cuity of the interactions, and add to the enthalpy energy
term by contributing additional bonds to the protein–
inhibitor complex [52–54]. In support of this role of water
molecules in enzyme–ligand complexes is the finding that
compounds 2 and 5 inhibit the protease activity with similar
efficacy although one of the indanyl groups of compound 2
was exchanged for a benzyl and a co-ordinating water
molecule in compound 5 (Fig. 6 and Table 2).
Conclusions
We have exploited the technique of structure-aided drug
design to improve the inhibitory efficacy of HIV-1 protease
lead compounds. The flexibility of the target molecule
complicates prediction of the effect of a modification of the
inhibitor and necessitates structural analysis of each
complex. In this case, the HIV-1 protease, the flexibility
was relatively conservative. The asymmetric binding of the
two central inhibitor hydroxyls to the active-site aspartates
induced a small deviation from the exact C2 symmetry in
the whole enzyme–inhibitor complex. This study shows
that, even without changing the chemistry of the inhibitor
scaffold but with modifications limited to the side groups,
several potent compounds can be designed. The most
active compounds in this series had the highest number of
contacts (bonds) between the protease and the inhibitor
Fig. 6. Value of a water molecule in the inter-
face between inhibitor and protein (stereo view).
A bound water molecule in the compound 5
protease complex (A) fulfils the function of the
indanyl hydroxyl in compound 2 (B) The two
compounds have comparable inhibitory

activity, 0.1 and 0.2 n
M
, respectively.
Ó FEBS 2003 Optimization of HIV-1 protease inhibitors (Eur. J. Biochem. 270) 1755
compound for a given area in the protease-active site. This
area is characterized as accessible area buried by the
compound in the complex. The study shows that these
parameters are useful guides in the optimization of
inhibitor compounds. Firmly bound water molecules were
found to be important components of the efficient
inhibitors. These results will be used in the development
of new HIV-1 protease inhibitors in general and new
compounds with significantly different resistance profiles in
particular.
Acknowledgements
We thank Mrs Terese Bergfors for reading the manuscript and
Dr Mats Sandgren for preparation of Fig. 5. We thank Professor
Sherry Mowbray and Professor Alwyn Jones for fruitful discussions.
This work was supported by the Swedish Medical Research Council
(MFR, K79-16X-09505-07A), the Swedish National Board for
Industrial and Technical Development (NUTEK), the Swedish
Research Council for Engineering Sciences (TFR), and Medivir
AB, Huddinge Sweden.
References
1. Kohl, N.E., Emini, E.A., Schleif, W.A., Davis, L.J., Heimbach,
J.C., Dixon, R.A., Scolnick, E.M. & Sigal, I.S. (1988) Active
human immunodeficiency virus protease is required for viral
infectivity. Proc. Natl Acad. Sci. USA 85, 4686–4690.
2. Peng, C., Ho, B.K., Chang, T.W. & Chang, N.T. (1989) Role of
human immunodeficiency virus type 1-specific protease in core

protein maturation and viral infectivity. J. Virol. 63, 2550–2556.
3. Gottlinger, H.G., Sodroski, J.G. & Haseltine, W.A. (1989) Role of
capsid precursor processing and myristoylation in morphogenesis
and infectivity of human immunodeficiency virus type 1. Proc.
NatlAcad.Sci.USA86, 5781–5785.
4. Debouck, C., Gorniak, J.G., Strickler, J.E., Meek, T.D., Metcalf,
B.W. & Rosenberg, M. (1987) Human immunodeficiency virus
protease expressed in Escherichia coli exhibits autoprocessing and
specific maturation of the gag precursor. Proc. Natl Acad. Sci.
USA 84, 8903–8906.
5. Graves, M.C., Lim, J.J., Heimer, E.P. & Kramer, R.A. (1988) An
11-kDa form of human immunodeficiency virus protease
expressed in Escherichia coli is sufficient for enzymatic activity.
Proc.NatlAcad.Sci.USA85, 2449–2453.
6. Hansen, J., Billich, S., Schulze, T., Sukrow, S. & Moelling, K.
(1988) Partial purification and substrate analysis of bacterially
expressed HIV protease by means of monoclonal antibody.
EMBO J. 7, 1785–1791.
7. Debouck, C. (1992) The HIV-1 protease as a therapeutic target for
AIDS. AIDS Res. Hum. Retroviruses 8, 153–164.
8. Wlodawer, A. & Vondrasek, J. (1998) Inhibitors of HIV-1 pro-
tease: a major success of structure-assisted drug design. Annu. Rev.
Biophys. Biomol. Struct. 27, 249–284.
9. Bishofberger, N. & O
¨
berg, B. (1999) Antivirals (HIV) editorial
overview. Curr. Opin. Anti-Infect. Invest. Drugs 1, 125–126.
10. Richman, D.D. (2001) HIV chemotherapy. Nature (london) 410,
995–1001.
Fig. 7. Extension of the inhibitors to the S3/S3¢

sites (stereo view). Extension to the S3/S3¢ sites
was accomplished in compound 8 by elonga-
tion of the P1/P1¢ benzyl groups with pyridyl
groups (A). In compound 9, the S3/S3¢ sites
were reached by elongation of the compound
with pyridyl groups at the P3 positions (B).
The van der Waals interaction with the S3 site
residues G48,G49, F53 and P181 is more ex-
tensive to the pyridyl group of compound 8
than to the pyridyl group in compound 9. In
addition, whereas in compound 9 the pyridyl
group only has van der Waals close packing
interactions with R108, the pyridyl group in
compound 8 is co-ordinated to R108 through
a hydrogen-bond network including water
molecules.
1756 H. O. Andersson et al.(Eur. J. Biochem. 270) Ó FEBS 2003
11. Carr, A. & Cooper, D.A. (1998) Images in clinical medicine.
Lipodystrophy associated with an HIV-protease inhibitor.
N. Engl. J. Med. 339, 1296.
12. Ridky, T. & Leis, J. (1995) Development of drug resistance to
HIV-1 protease inhibitors. J. Biol. Chem. 270, 29621–29623.
13. Wlodawer, A., Miller, M., Jaskolski, M., Sathyanarayana, B.K.,
Baldwin, E., Weber, I.T., Selk, L.M., Clawson, L., Schneider, J.
& Kent, S.B. (1989) Conserved folding in retroviral proteases:
crystal structure of a synthetic HIV-1 protease. Science 245,616–
621.
14. Navia, M.A., Fitzgerald, P.M., McKeever, B.M., Leu, C.T.,
Heimbach, J.C., Herber, W.K., Sigal, I.S., Darke, P.L. & Springer,
J.P. (1989) Three-dimensional structure of aspartyl protease from

human immunodeficiency virus HIV-1. Nature (London) 337,
615–620.
15. Schechter, I. & Berger, A. (1967) On the size of the active site in
proteases. I. Papain. Biochem. Biophys. Res. Commun. 27,157–
162.
16. Wlodawer, A. & Erickson, J.W. (1993) Structure-based inhibitors
of HIV-1 protease. Annu. Rev. Biochem. 62, 543–585.
17. Kempf, D.J., Norbeck, D.W., Codacovi, L., Wang, X.C.,
Kohlbrenner, W.E., Wideburg, N.E., Paul, D.A., Knigge,
M.F., Vasavanonda, S., Craig-Kennard, A. et al. (1990) Structure-
based, C2 symmetric inhibitors of HIV protease. J. Med. Chem.
33, 2687–2689.
18. Erickson, J., Neidhart, D.J., VanDrie, J., Kempf, D.J., Wang,
X.C., Norbeck, D.W., Plattner, J.J., Rittenhouse, J.W., Turon,
M., Wideburg, N. et al. (1990) Design, activity, and 2.8 A
˚
crystal
structure of a C2 symmetric inhibitor complexed to HIV-1 pro-
tease. Science 249, 527–533.
19. Erickson, J. & Kempf, D. (1994) Structure-based design of sym-
metric inhibitors of HIV-1 protease. Arch. Virol. Suppl. 9, 19–29.
20. Kempf, D.J. (1994) Design of symmetry-based, peptidomimetic
inhibitors of human immunodeficiency virus protease. Methods
Enzymol. 241, 334–354.
21. Ho, D.D., Toyoshima, T., Mo, H., Kempf, D.J., Norbeck, D.,
Chen, C.M., Wideburg, N.E., Burt, S.K., Erickson, J.W. & Singh,
M.K. (1994) Characterization of human immunodeficiency virus
type 1 variants with increased resistance to a C2-symmetric pro-
tease inhibitor. J. Virol. 68, 2016–2020.
22. Mahalingam,B.,Louis,J.M.,Reed,C.C.,Adomat,J.M.,Krouse,

J., Wang, Y.F., Harrison, R.W. & Weber, I.T. (1999) Structural
and kinetic analysis of drug resistant mutants of HIV-1 protease.
Eur. J. Biochem. 263, 238–245.
23. Weber, J., Mesters, J.R., Lepsik, M., Prejdova, J., Svec, M.,
Sponarova, J., Mlcochova, P., Skalicka, K., Strisovsky, K.,
Uhlikova, T., Soucek, M., Machala, L., Stankova, M., Vond-
rasek, J., Klimkait, T., Kraeusslich, H.G., Hilgenfeld, R. &
Konvalinka, J. (2002) Unusual binding mode of an HIV-1 pro-
tease inhibitor explains its potency against multi-drug-resistant
virus strains. J. Mol. Biol. 324, 739–754.
24. Yoshimura, K., Kato, R., Kavlick, M.F., Nguyen, A., Maroun,
V., Maeda, K., Hussain, K.A., Ghosh, A.K., Gulnik, S.V.,
Erickson, J.W. & Mitsuya, H. (2002) A potent human im-
munodeficiency virus type 1 protease inhibitor, UIC- 94003
(TMC-126), and selection of a novel (A28S) mutation in the
protease active site. J. Virol. 76, 1349–1358.
25. Hidaka, K., Kimura, T., Hayashi, Y., McDaniel, K.F., Dekhtyar,
T., Colletti, L. & Kiso, Y. (2003) Design and synthesis of pseudo-
symmetric HIV protease inhibitors containing a novel hydroxy-
methylcarbonyl (HMC)-hydrazide isostere. Bioorg.Med.Chem.
Lett. 13, 93–96.
26. Alterman, M., Bjorsne, M., Muhlman, A., Classon, B., Kvarn-
strom, I., Danielson, H., Markgren, P.O., Nillroth, U., Unge, T.,
Hallberg, A. & Samuelsson, B. (1998) Design and synthesis of new
potent C2-symmetric HIV-1 protease inhibitors. Use of 1-man-
naric acid as a peptidomimetic scaffold. J. Med. Chem. 41, 3782–
3792.
27. Alterman, M., Andersson, H.O., Garg, N., Ahlsen, G., Lovgren,
S.,Classon,B.,Danielson,U.H.,Kvarnstrom,I.,Vrang,L.,
Unge, T., Samuelsson, B. & Hallberg, A. (1999) Design and fast

synthesis of C-terminal duplicated potent C(2)-symmetric P1/P1¢-
modified HIV-1 protease inhibitors. J. Med. Chem. 42, 3835–3844.
28. Nillroth, U., Vrang, L., Markgren, P.O., Hulten, J., Hallberg, A.
& Danielson, U.H. (1997) Human immunodeficiency virus type 1
proteinase resistance to symmetric cyclic urea inhibitor analogs.
Antimicrob. Agents Chemother. 41, 2383–2388.
29. Weislow, O.S., Kiser, R., Fine, D.L., Bader, J., Shoemaker, R.H.
& Boyd, M.R. (1989) New soluble-formazan assay for HIV-1
cytopathic effects: application to high-flux screening of synthetic
and natural products for AIDS: antiviral activity. J. Natl Cancer
Inst. 81, 577–586.
30. Otwinowski, Z. & Minor, W. (1997) Processing of x-ray diffrac-
tion data collected in oscillation mode. Methods Enzymol. 276,
307–326.
31. Gewirth, D. & Otwinowski, Z. (1995) The SCALEPACK Manual,
available from />abs_denzo_suite.html
32. Bru
¨
nger, A.T., Adams, P.D., Clore, G.M., Delano, W.L., Gros,
P., Grosse-Kunstleve, R.W., Jiang, J S., Kuszewski, J., Nilges, N.,
Pannu,N.S.,Read,R.J.,Rice,L.M.,Simonson,T.&Warren,
G.L. (1998) Crystallography and NMR system (CNS): a new
software system for macromolecular structure determination. Acta
Crystallogr. D54, 905–921.
33. Jones,T.A.,Zou,J Y.,Cowan,S.W.&Kjeldgaard,M.(1991)
Improved methods for building protein models in electron density
maps and the location of errors in these models. Acta Crystallogr
A. 47, 110–119.
34. Bru
¨

nger, A.T., Kuriyan, J. & Karplus, M. (1987) Crystallographic
R factor refinement by molecular dynamics. Science 235, 458–460.
35. Kleywegt, G.J. & Bru
¨
nger, A.T. (1996) Checking your imagina-
tion: applications of the free R. value. Structure 4, 897–904.
36. Guex, N. & Peitsch, M.C. (1997) SWISS-MODEL and the Swiss-
PdbViewer: an environment for comparative protein modeling.
Electrophoresis 18, 2714–2723.
37. Vondrasek, J., van Buskirk, C.P. & Wlodawer, A. (1997) Data-
base of three-dimensional structures of HIV proteinases. Nat.
Struct. Biol. 4,8.
38. Wlodawer, A. & Gustchina, A. (2000) Structural and biochemical
studies of retroviral proteases. Biochim. Biophys. Acta 1477, 16–34.
39. Silva, A.M., Cachau, R.E., Sham, H.L. & Erickson, J.W. (1996)
Inhibition and catalytic mechanism of HIV-1 aspartic protease.
J. Mol. Biol. 255, 321–346.
40. Dreyer, G.B., Boehm, J.C., Chenara, B., DesJarlais, R.L., Hassell,
A.M., Meek, T.D., Tomaszek, T.A. Jr & Lewis, M. (1993) A
symmetric inhibitor binds HIV-1 protease asymmetrically. Bio-
chemistry 32, 937–947.
41. Bone, R., Vacca, J.P., Anderson, P.S. & Holloway, M.K. (1991)
X-ray crystal structure of the HIV protease complex with
L-700,417, an inhibitor with pseudo C
2
symmetry. J. Am. Chem.
Soc. 113, 9382–9384.
42. Turner, B.G. & Summers, M.F. (1999) Structural biology of HIV.
J. Mol. Biol. 285, 1–32.
43. Mager, P.P. (2001) The active site of HIV-1 protease. Med. Res.

Rev. 21, 348–353.
44. Ren, S. & Lien, E.J. (2001) Development of HIV protease
inhibitors: a survey. Prog. Drug Res. Spec. No 1–34.
45. Ala, P.J., DeLoskey, R.J., Huston, E.E., Jadhav, P.K., Lam, P.Y.,
Eyermann, C.J., Hodge, C.N., Schadt, M.C., Lewandowski, F.A.,
Weber, P.C., McCabe, D.D., Duke, J.L. & Chang, C.H. (1998)
Molecular recognition of cyclic urea HIV-1 protease inhibitors.
J. Biol. Chem. 273, 12325–12331.
Ó FEBS 2003 Optimization of HIV-1 protease inhibitors (Eur. J. Biochem. 270) 1757
46. Alterman,M.,Sjobom,H.,Safsten,P.,Markgren,P.O.,Daniel-
son, U.H., Hamalainen, M., Lofas, S., Hulten, J., Classon, B.,
Samuelsson, B. & Hallberg, A. (2001) P1/P1¢ modified HIV pro-
tease inhibitors as tools in two new sensitive surface plasmon
resonance biosensor screening assays. Eur. J. Pharm. Sci. 13, 203–
212.
47. Lam, P.Y.S., Jadhav, P.K., Eyermann, C.J., Hodge, C.N., Ru, Y.,
Bacheler, L.T., Meek, J.L., Otto, M.J., Rayner, M.M., Wong,
Y.N.,Chang,C H.,Weber,P.C.,Jackson,D.A.,Sharpe,T.R.&
Erickson-Viitanen, S. (1994) Rational design of potent, bioavail-
able, nonpeptide cyclic ureas as HIV protease inhibitors. Science
263, 380–384.
48. Lam, P.Y., Ru, Y., Jadhav, P.K., Aldrich, P.E., DeLucca, G.V.,
Eyermann, C.J., Chang, C.H., Emmett, G., Holler, E.R., Dane-
ker, W.F. et al. (1996) Cyclic HIV protease inhibitors: synthesis,
conformational analysis, P2/P2¢ structure-activity relationship,
and molecular recognition of cyclic ureas. J. Med. Chem. 39, 3514–
3525.
49. Hulten, J., Bonham, N.M., Nillroth, U., Hansson, T., Zuccarello,
G., Bouzide, A., Aqvist, J., Classon, B., Danielson, U.H., Karlen,
A., Kvarnstrom, I., Samuelsson, B. & Hallberg, A. (1997) Cyclic

HIV-1 protease inhibitors derived from mannitol: synthesis,
inhibitory potencies, and computational predictions of binding
affinities. J. Med. Chem. 40, 885–897.
50. Hulten, J., Andersson, H.O., Schaal, W., Danielson, H.U., Clas-
son,B.,Kvarnstrom,I.,Karlen,A.,Unge,T.,Samuelsson,B.&
Hallberg, A. (1999) Inhibitors of the C(2)-symmetric HIV-1 pro-
tease: nonsymmetric binding of a symmetric cyclic sulfamide with
ketoxime groups in the P2/P2¢ side chains. J. Med. Chem. 42,
4054–4061.
51. Wang, H. & Ben-Naim, A. (1996) A possible involvement of
solvent-induced interactions in drug design. J. Med. Chem. 39,
1531–1539.
52. Mattos, C. (2002) Protein–water interactions in a dynamic world.
Trends Biochem. Sci. 27, 203–208.
53. Bhat, T.N., Bentley, G.A., Boulot, G., Greene, M.I., Tello, D.,
Dall’Acqua, W., Souchon, H., Schwarz, F.P., Mariuzza, R.A. &
Poljak, R.J. (1994) Bound water molecules and conformational
stabilization help mediate an antigen–antibody association. Proc.
NatlAcad.Sci.USA91, 1089–1093.
54. Ladbury, J.E. (1996) Just add water! The effect of water on the
specificity of protein-ligand binding sites and its potential appli-
cation to drug design. Chem. Biol. 3, 973–980.
55. Engh, R.A. & Huber, R. (1991) Accurate bond and angle para-
meters for X-ray protein structure refinement. Acta Crystallogr. A
47, 392–400.
56. Kleywegt, G.J. & Jones, T.A. (1999) Software for handling macro
molecular envelopes. Acta Crystallogr. D55, 941–944.
57. Kleywegt, G.J. & Jones, T.A. (1997) Detecting folding motifs and
similarities in protein structures. Methods Enzymol. 277, 525–545.
1758 H. O. Andersson et al.(Eur. J. Biochem. 270) Ó FEBS 2003