From docking false-positive to active anti-HIV agent

Article abstract of DOI:10.1021/jm070683u

Virtual screening of the Maybridge library of ca. 70 000 compounds was performed using a similarity filter, docking, and molecular mechanics- generalized Born/surface area postprocessing to seek potential non-nucleoside inhibitors of human immunodeficienc

Full text of DOI:10.1021/jm070683u

5
324
J. Med. Chem. 2007, 50, 5324-5329
From Docking False-Positive to Active Anti-HIV Agent
†
†
†
‡
‡
‡
Gabriela Barreiro, Joseph T. Kim, Cristiano R. W. Guimar a˜ es, Christopher M. Bailey, Robert A. Domaoal, Ligong Wang,
‡
,†
Karen S. Anderson, and William L. Jorgensen*
Department of Chemistry, Yale UniVersity, New HaVen, Connecticut 06520-8107, and Department of Pharmacology,
Yale UniVersity School of Medicine, New HaVen, Connecticut 06520-8066
ReceiVed June 12, 2007
Virtual screening of the Maybridge library of ca. 70 000 compounds was performed using a similarity filter,
docking, and molecular mechanics-generalized Born/surface area postprocessing to seek potential non-
nucleoside inhibitors of human immunodeficiency virus-1 (HIV-1) reverse transcriptase (NNRTIs). Although
known NNRTIs were retrieved well, purchase and assaying of representative, top-scoring compounds from
the library failed to yield any active anti-HIV agents. However, the highest-ranked library compound,
oxadiazole 1, was pursued as a potential “near-miss” with the BOMB program to seek constructive
modifications. Subsequent synthesis and assaying of several polychloro-analogs did yield anti-HIV agents
with EC50 values as low as 310 nM. The study demonstrates that it is possible to learn from a formally
unsuccessful virtual-screening exercise and, with the aid of computational analyses, to efficiently evolve a
false positive into a true active.
Introduction
for lead generation by molecular docking have been intriguing,4
and it was decided to try this approach to seek novel NNRTIs.
The following report provides a case study on a common
dilemma in a virtual or high-throughput screening (HTS)
exercise. It is demonstrated that, with confidence in computed
structures and estimated activities, it is possible to convert a
false positive into an active agent.
The acquired immunodeficiency syndrome (AIDS) crisis
continues with ca. 40 million people infected by human
immunodeficiency virus (HIV) and 2.9 million HIV-related
1
deaths in 2006. The virally encoded proteins of HIV provide
chemotherapeutic targets for the treatment of infection by the
virus. A principal point of attack has been HIV reverse
transcriptase (RT), which is required for the conversion of the
viral genomic RNA to DNA and for successful infection of host
cells. This has led to the development and FDA approval of
two important classes of anti-HIV drugs: (i) nucleoside and
nucleotide RT inhibitors (NRTIs and NtRTIs, respectively), for
example, AZT, ddI, ddC, d4T, FTC, TDF, and 3TC, and (ii)
non-nucleoside RT inhibitors (NNRTIs), specifically, nevirapine,
delavirdine, and efavirenz. The NRTIs and NtRTIs are actually
faulty substrates that cause premature termination of the growing
DNA transcript, while NNRTIs are true inhibitors that bind to
Virtual Screening
A summary of the virtual screening effort is presented here,
5
with full details provided elsewhere. Virtual screening was
performed on roughly 70 000 compounds from the Maybridge
collection, which was supplemented with 26 known NNRTIs.
The screening protocol began with a similarity filter that
retrieves 60% of the known actives in the top 5% of the screened
library. The ca. 2000 library compounds that were most similar
to the known actives were then docked into the structure of
2
an allosteric pocket in the vicinity of the polymerase active site.
6
wild-type HIV-1 RT, taken from its 1rt4 crystal structure in
The therapeutic situation is challenged by rapid mutation of the
virus to yield resistant strains. This leads to a need for new
drugs with activity against at least parts of the spectrum of
variants, which are now clinically observed.
complex with a carboxanilide NNRTI, using Glide 3.5 with
standard precision. The top 500 compounds were then re-
docked and scored in Glide extra-precision (XP) mode. The top
7
1
00 of these were post-scored with a molecular mechanics-
Our own efforts have been directed toward the development
of NNRTIs with enhanced therapeutic spectra and auspicious
pharmacological properties. The approach to date has featured
focused synthetic organic chemistry and anti-HIV assaying
driven by automated procedures for the creation and evaluation
of virtual libraries, estimation of pharmacological properties,
and lead optimization featuring free-energy perturbation calcula-
generalized Born/surface area (MM-GB/SA) method that was
shown to provide high correlation between predicted and
observed activities for known NNRTIs. The MM-GB/SA
5
procedure included a conformational search for the ligand and
optimization of the complex with the all-atom optimized
8
potentials for liquid simulations (OPLS-AA) force field as well
as evaluation of the change in free energy of hydration using
3
tions to assess relative protein-ligand binding affinities. Highly
9
10
the GB/SA approach, as implemented in MacroModel.
potent and structurally diverse anti-HIV agents have been
The final top-20 ranked compounds consisted of 10 library
compounds and 10 known NNRTIs. The best ranked library
compound, 1 (S10087), was third overall, and the other top-20
library compounds were in two series containing a dihydroben-
zofuran or 1-phenyltetrazole. The known NNRTIs are 2
3
discovered; however, we continue to seek activity against an
ever-wider range of viral mutants and exploration of alternative
structural classes for NNRTIs. To this end, reported successes
*
Corresponding author. Phone: 203-432-3915. Fax: 203-432-6299.
(
SDABO-3w), 3 (UC781), 4 (TMC125), 5 (efavirenz) and three
E-mail: william.jorgensen@yale.edu.
3
†
‡
analogs, and three pyrimidine derivatives from our group. Six
of the top library compounds were purchased from Maybridge:
Department of Chemistry.
Department of Pharmacology.
1
0.1021/jm070683u CCC: $37.00 © 2007 American Chemical Society
Published on Web 10/06/2007

EVolVing False-PositiVes to ActiVe Anti-HIV Agents
Journal of Medicinal Chemistry, 2007, Vol. 50, No. 22 5325
1
and the KM and NRB compounds shown above. The two
NRB compounds were found to be impure and gave mass
spectra that are inconsistent with the illustrated structures. The
other four library compounds gave correct NMR and mass
spectra and were purified by high-performance liquid chroma-
tography, recrystallized, and submitted for assay. Activities
against the IIIB strain of HIV-1 were determined using MT-2
human T-cells; EC50 values are obtained as the dose required
to achieve 50% protection of the infected cells using the MTT
colorimetric method. The CC50 for inhibition of MT-2 cell
growth by 50% is obtained simultaneously.11 The four com-
pounds showed no anti-HIV activity up to their CC50 concentra-
tions of 61, 41, 75, and 3 µM, respectively.
Figure 1. Computed structure for 1 bound to HIV-1 RT from Glide
docking followed by energy minimization with MacroModel. Only
protein residues in the first layer around the ligand are illustrated. The
ligand is positioned as expected from known structures such as for
16,17
4
.
as illustrated in Figure 1 for 1, and the penalties for achieving
the docked conformations were not high.
Post Mortem. Several reasons can be suggested for the lack
of success with finding active compounds in the Maybridge
library: (a) remaining inaccuracies in the scoring including the
“conformer focusing” problem and lack of explicit representation
of the solvent, (b) the size of the screened library (only 70 000
compounds), (c) needing to dig deeper into the deck (purchase
more compounds), and (d) the sensitivity of activity to structure.
Concerning reason d, as demonstrated in our own NNRTI series,
the addition of a chloro and a methoxy group takes 6 from an
3
EC50 of 30 µM to 10 nM (8), while the addition of another
small group in the wrong place can be very detrimental (9 or
1
0).
In essence, our best effort was made to computationally screen
the Maybridge collection in search of NNRTI leads. The
screening protocol seemed well-founded in view of (a) the
testing of the MM-GB/SA procedure, (b) the fact that Glide-
docked structures (“poses”) for nine known NNRTIs agreed well
with crystallographic results, and (c) the fact that the NRB
compounds are close in structure to recently reported NNRTI
HTS-hits that feature N-phenyl triazoles and tetrazoles (Scheme
1
2
1
).
In fact, 16 compounds had been purchased from Maybridge
and assayed from a preliminary version of the same docking
exercise that did not include the MM-GB/SA post-scoring. These
compounds were also all inactive in the MT-2 assay, and
structure visualization indicated that there were generosities in
the evaluation of conformational energetics by Glide, including
some twisted groups such as amides that should be more-or-
less planar. This led to exploration of the post-scoring and to
the reporting of an analysis of the current problems for rank-
ordering of compounds by docking owing to the uncertainties
in evaluating conformational energetics and “conformer focus-
ing”, basically the need to evaluate the overall change in
conformational energetics for the ligand and protein between
the unbound and bound states.13 In the end, the structures from
the docking plus MM-energy minimization appear reasonable,
Thus, it is expected that there can be “near-misses” from
virtual or experimental screening in which the activity window
for an assay is missed owing to a seemingly minor structural
omission or addition. Even given a viable core, the chance of
an active substituent pattern for a target being included in a
1
4
molecular library is extremely low. Of course, the nature of
the assay can also be an issue. It is possible that some inactive
compounds in a cell-based assay might be active in an in vitro
assay for enzyme inhibition. However, in the present case, good
correlations have been observed consistently for NNRTIs in

5
326 Journal of Medicinal Chemistry, 2007, Vol. 50, No. 22
Barreiro et al.
Figure 2. Training results for scoring using BOMB for 339 complexes
with HIV-RT (magenta: uracil derivatives; green: analogs of 2),
COX-2 (blue), FKBP (red), and p38 kinase (black). The correlation
2
coefficient for the fit, r , is 0.58, and the root-mean-square error is
0
.82 log unit. Indicator variables are used to provide constant offsets
Figure 3. Structure for the complex of 1 with HIV-RT as computed
by BOMB. The 176 residues nearest the ligand are included in the
calculations.
for three datasets.
assays for the inhibition of reverse transcription and in cell-
based anti-HIV analyses.15 Furthermore, the value of a lead that
was not active in the cell-based assay would be comparatively
low.
a docking-like scoring function. The current scoring function
has been trained to reproduce experimental activity data for 339
complexes of HIV-RT, COX-2, FK506 binding protein (FKBP),
and p38 kinase (Figure 2). The scoring function only contains
five descriptors that were obtained by linear regression: the
octanol/water partition coefficient for the analog as computed
From False-Positive to Lead
The possibility that the virtual screening provided a near-
miss was pursued to ascertain whether a false-positive could
be evolved into a lead and to determine how close the miss
was in structure to a true active. The oxadiazole motif was
chosen since 1 ranked so highly in the virtual screening. The
predicted structure of the complex (Figure 1) seemed reasonable
and reminiscent of the structures of complexes of HIV-RT with
analogs of 4.1 Three analogs of 1 differing in the substitution
of the anilinyl ring were also in the top-50 ranked compounds:
2
1
by QikProp (QPlogPo/w), the amount of hydrophobic surface
area for the protein that is buried upon complex formation
(∆FOSAP), the number of potential hydrogen-bond donating
hydrogens in the analog (HBDNL), the number of non-
conjugated amides in the analog, and the number of “bad”
protein-analog contacts in the complex (NBAD). The latter
represent structural mismatches between two atoms within 4
Å, typically between a potential hydrogen-bonding oxygen or
nitrogen and a saturated carbon atom or between a potential
hydrogen-bond accepting O or N and another such atom or an
aryl carbon atom.
6,17
3
-fluoro (S10076), 3-chloro, 4-methyl (S10085), and 2,4-
dichloro (S10089). All four compounds share the 3,4-dimethox-
ybenzyl group. Suspicion centered on this subunit and on the
oxadiazole ring itself. 1 and 4 can be viewed as members of a
general structural class, U-Het-NH-PhX,3a where U is an
Interestingly, (a) the most significant descriptor is QPlogPo/w,
2
which alone yields a fit with an r of 0.47, and (b) inclusion of
unsaturated group such as a substituted benzyl, anilinyl, or
phenoxy group, Het is a heterocycle, and PhX is a substituted
phenyl group. NNRTIs have been reported by Janssen and
co-workers where Het is a pyrimidine, triazine, or pyrazi-
none.
3
energetic results from full conjugate-gradient optimizations of
the complexes created by BOMB was found to have little impact
on the accuracy of the scoring. Although the BOMB scoring is
still under development, the current version provides a useful
evaluation of potential activity (Figure 2).
1
8
1
9,20
However, there have not been reports where U is a
,4-disubstituted benzyl group or Het is a 5-membered-ring
heterocycle. Finally, the para-methyl substituent in 1 was not
viewed as optimal; chloro and cyano are common in the Janssen
NNRTIs, and, in our series, 4-chloro and cyano are preferred
over methyl by a factor of ca. 15 in EC50 values. Thus, surgery
was indicated for both phenyl rings of 1, though it was possible
that no oxadiazole derivative would be viable.
A standard protocol for a substituent scan with BOMB is to
replace each hydrogen of a core with 10 small groups that have
been selected for difference in size, electronic character, and
hydrogen-bonding patterns: Cl, CH3, NH2, OH, CH2NH2, CH2-
OH, CHO, CN, NHCH3, and OCH3. To begin, the structure of
1 bound to RT was built with BOMB using the anilinylbenzy-
loxadiazole core and the structure of RT from the 1s9e PDB
file. The switch from the 1rt4 structure arose because the 1s9e
structure has higher resolution (2.6 vs 2.9 Å), and the 1s9e ligand
is more similar to the U-Het-NH-PhX motif of the oxadiazole;
it is a U-Het1-L-Het2 member with Het1 as a pyridinone, U as
3,5-dimethylphenoxy, and L, a linking chain, as CH2SCH2. The
structure of the 1 complex built using BOMB is illustrated in
Figure 3. It is similar to the structure in Figure 1 except for the
conformation of the two methoxy groups. The conformation in
Figure 1 generated by Glide with both methoxy groups in-plane
and the one in Figure 3 from BOMB with one almost
perpendicular to the benzene ring differ in energy by only 0.16
kcal/mol for o-dimethoxybenzene at the MP2/6-311G** level.23
3
BOMB Analyses. At this point, a full substituent scan was
indicated for replacement of each phenyl hydrogen in the
22
2
-anilinyl-5-benzyloxadiazole core. This was carried out with
the BOMB program, which creates analogs by adding substit-
3
a
uents to a core that has been placed in a binding site.
A
thorough conformational search is performed for each analog,
and the position, orientation, and dihedral angles for the analog
are optimized using the OPLS-AA force field for the protein
8
and OPLS/CM1A for the analog. The protein is rigid except
for optimization of the terminal dihedral angles for side chains
with hydrogen-bonding groups (e.g., the OH of tyrosine and
the carboxylate group of aspartate). The resultant conformer
for each analog with the lowest energy is then evaluated with

EVolVing False-PositiVes to ActiVe Anti-HIV Agents
Scheme 1. Recent NNRTI Hits from HTS (Ref 12)
Journal of Medicinal Chemistry, 2007, Vol. 50, No. 22 5327
Table 1. Anti-HIV-1 Activity (EC50) and Cytotoxicity (CC50) in µM, BOMB Score, and Protein-Ligand Interaction Energy
BOMB
score
compd
R1
CH3
Cl
Cl
Cl
Cl
Cl
H
R2
R3
R4
EC50a
CC50b
EXXc
1
1
1
1
1
1
1
1
1
1
2
H
H
H
H
H
H
H
CH3
OCH3
Cl
3-OCH3
4-Cl
3-Cl
3-CH3
2-Cl
H
4-OCH3
H
NA
NA
>10
NA
0.82
NA
NA
1.3
NA
0.31
NA
1.6
61
>100
32
-4.1
-5.5
-5.7
-5.7
-5.9
-5.3
-4.5
-5.8
-5.4
-5.7
-6.2
-59.3
-56.1
-53.7
-57.4
-61.3
-55.6
-49.8
-64.6
-68.4
-65.2
-63.8
1
2
3
4
5
6
7
8
9
0
H
5-CH3
6-Cl
H
>100
20
31
H
H
>100
>100
9
Cl
2-Cl
2-Cl
2-Cl
2-Cl
6-Cl
6-Cl
6-Cl
6-Cl
2-Cl,4-Cl
Cl
Cl
>100
CF3
9
d4T
nevirapine
>100
>10
0.11
a
For 50% protection in MT-2 cells; antiviral curves used triplicate samples at each concentration. NA for EC50 > CC50. b For 50% inhibition of MT-2
c
cell growth; toxicity curves also used triplicate samples. From conjugate gradient optimization of the complex using a dielectric constant of 2 with MCPRO;
units are kcal/mol.
BOMB was then used to build the 100 complexes corre-
sponding to the replacement of each phenyl hydrogen in the
core with the 10 small groups. This revealed that the top-25
scoring analogs are dominated by substitution of either chlorine
or a methyl group at the 3-, 4-, and 2-position in the benzyl
ring and at the 4-, 2-, and 3-position of the anilinyl ring in
roughly that order. The top three analogs have chlorine at the
Synthesis and Assay Results. The oxadiazole derivatives
were prepared via cyclization of substituted phenylacetic
25
26
hydrazides and substituted phenyl isocyanide dichlorides, as
illustrated in Scheme 2 for 14. The procedure finds precedent
in the synthesis of N,N-diphenyloxadiazol-2,5-diamines from
27
semicarbazides. Full synthetic and spectral details are provided
in the Supporting Information. The assay results for the first
three compounds that were prepared (11-13) were hardly
encouraging. Compounds 11 and 13 are benign (inactive and
non-cytotoxic), and for 12 there was, at best, a small hint of
activity above 10 µM that was masked by the onset of
cytotoxicity with a CC50 of 32 µM. It was also somewhat
surprising that, if 12 with a 3-chloro substituent in the benzyl
ring were active, the 3,5-dimethyl analog (13) showed no activity
up to 100 µM. A 3,5-dimethylphenoxy group is present as the
U group in the pyridinone ligand (a 2-nM NNRTI) in the 1s9e
3-, 4-, and 5-position in the benzyl ring, where the methoxy
groups in Figure 3 would be said to be at the 3- and 4-positions.
Monte Carlo/free-energy perturbation calculations in explicit
3
water were also performed for mono-chlorination and favored
substitution at the 2- and 6-positions of the benzyl ring and the
4
-position of the anilinyl ring. The 45 possible dichloro-
substituted compounds were built next with BOMB. The 10
best scores are within 0.2 log unit, but revealed a preference
for chlorine at the 4-position in the anilinyl ring and at the 5-
and 6-positions in the benzyl ring.
2
2
crystal structure, but it is clearly ineffective in the present
oxadiazole series.
Given this information, synthesis of compounds 11-20 in
Table 1 was initiated. The table includes the BOMB scores for
Nevertheless, 14 was then prepared to probe the 2,6-dichloro
pattern that was supported by the computational analyses and
has precedent in the 2,6-difluoro and 2,6-dimethyl substitution
for the U groups in 2 and 4. Compound 15 was also prepared
as a reference compound, although there was no hope for it as
an active in view of the results for the dichloro-analogs 11 and
12. Fortunately, 14 was found to be a definite active at the 820-
nM level and a respectable low-molecular-weight (355) lead
for further optimization. Thus, 1 can be classified as a near-
miss with a need, at least, for an alternative substitution pattern
in the benzyl ring. For reference, it is noted that the nucleoside
inhibitor d4T and the NNRTI nevirapine are active at the 1.6-
µM and 110-nM levels in the MT-2 assay (Table 1).
1
and the prepared compounds as well as the protein-ligand
interaction energy for the complexes after conjugate-gradient
24
optimization of the structures from BOMB using MCPRO with
the OPLS-AA (protein) and OPLS/CM1A (ligand) force fields
and a dielectric constant of 2. At this point, it is noted that 1
scored poorly in comparison to the polychloro analogs; the
difference was dominated by 1 being predicted to be less
lipophilic by 1-2 log units in QPlogPo/w (2.6 vs 3.6-4.2) and
for having 28 “bad” contacts versus ca. 18 for the dichloro
compounds. The additional undesirable contacts stem from the
proximity of the methoxy oxygens and the aromatic and Câ
carbons of Tyr181 and Tyr188. Furthermore, a global concern
for the series was the proximity of the carboxylate oxygens of
Glu138 to N4 of the oxadiazole and to the benzylic CH2 group.
LT00141083 Interlude. At this point, a paper appeared
reporting the results of another screening effort seeking NNR-

5
328 Journal of Medicinal Chemistry, 2007, Vol. 50, No. 22
Barreiro et al.
Scheme 2. Synthesis of 2,5-Oxadiazole Derivatives
TIs.28 In that case, a 3D-pharmacophore model was used for
initial filtering of the Derwent and CAP databases, which was
followed by docking of ca. 10 000 compounds with Glide. Six
compounds were purchased to test their “in vitro RNA-
dependent DNA polymerase activity of recombinant RT.” Five
of the six compounds were reported to be active. This includes
a compound 5 in ref 28 that was drawn as being identical to
the core in the present work (16 in Table 1). No spectral or
other characterization data were provided for the assayed
compounds.28 Compound 5 in ref 28 yielded an IC50 of 4.2 µM
in the enzyme assay; nevirapine was also reported as 0.18 µM
for reference, which is a little less potent than that in our assay
virtual screening can be evolved into a true active. Finally, in
comparing the experimental activities, BOMB scores, and
protein-ligand interaction energies (EXX) in Table 1, correla-
tions are not perfect. However, the compounds with BOMB
scores above -5.7 are all inactive. EXX is not a particularly
good index of potential binding because it tends to increase with
increasing size of the ligand; dividing it by the number of atoms
or molecular weight is flawed without correction for parts of
the ligand that make no contact with the protein. Nevertheless,
compounds with both very favorable BOMB scores and EXX
values are likely to be active. Compound 20 is the exception,
although it suffers from a symmetry loss of at least a factor of
2 in activity.
(Table 1). There was one available commercial source for 16,
LaboTest AB, and the compound was designated LT00141083.
We purchased it. The authors of the other study28 verified that
they had also purchased their compound 5 from LaboTest. A
Conclusions
Virtual screening of the Maybridge library of ca. 70 000
compounds was performed using a similarity filter, docking,
and MM-GB/SA postprocessing to seek potential NNRTIs. The
protocol did well in identifying known NNRTIs. Six representa-
tive, top-scoring compounds were purchased from Maybridge.
Two were impure and gave incorrect mass spectra. After
purification, the other four were submitted to an anti-HIV assay
using infected human T-cells and found to show no ability to
inhibit HIV replication. However, the highest-ranked library
compound, the oxadiazole derivative 1, was subjected to further
computational analysis to seek modifications that might yield
an active anti-HIV agent. A small substituent scan was
performed for the anilinylbenzyloxadiazole core with the BOMB
program, which specifically suggested removal of the methoxy
groups in the benzyl ring and the addition of two or three
chlorines or methyl groups. Synthesis and assaying of several
such compounds did yield active anti-HIV agents, as sum-
marized in Table 1. The study demonstrates that it is possible
to learn from a formally unsuccessful virtual-screening exercise
and, with the aid of computational analyses, to efficiently evolve
a false positive into a true active. The present case may be
viewed as favorable in view of the availability of significant
structure-activity data for NNRTIs; however, it provides a
prototype for future efforts and, in fact, success was far from
guaranteed owing to the sensitivity of activity to structure, the
plethora of incorrect choices that could be made, and the lack
of precedent for five-membered-ring heterocycles as the central
ring in NNRTIs.
1
H NMR spectrum was taken of the purchased sample, which
immediately revealed that it could not be 16 in view of having
the incorrect number of protons and, for example, no benzylic
protons. Thus, 16 was then synthesized and submitted to the
MT-2 anti-HIV assay. In view of the assay results for 11-15
and the BOMB scores, there was no possibility that 16 could
be active. This was confirmed (Table 1). Under the circum-
stances, it seems highly unlikely that 16 was assayed in the
2
8
1
other study. The H NMR spectra of the purchased material
and 16 are compared in the Supporting Information.
A Few More Compounds. From the BOMB analyses and
visualization of the resultant structures, replacement of hydrogen
by chlorine or a methyl at C3 in the anilinyl ring of 14 seemed
worth investigation. However, it is noted that such additions
break the symmetry of the para-substituted phenyl ring. This
can have negative impact on binding if the added substituent
in the bound state can only be at either effectively the 3- or
5
-position, while it can sample both in the unbound state. This
could lead to a factor of 2 reduction in the binding affinity or
perhaps even a factor of 10 in an extreme case in which the
rotation of the phenyl ring is free in the unbound state, but
restricted to a dihedral range of 36° when bound. For the present
case, substitution of chlorine or a methyl group at C3 or C5
appeared to be allowed sterically and yielded similar BOMB
scores, while a trifluoromethyl group is only allowed on the
side facing the benzyl group in the bound state.
In the event, 17 and 19 were synthesized along with the
trifluoromethyl analog 20. An additional product, 18, which
arose in the course of failed attempts to synthesize the
Acknowledgment. Gratitude is expressed to the National
Institutes of Health (AI44616, GM32136, GM49551) for
support. Computational assistance from Drs. Julian Tirado-Rives
and Ivan Tubert-Brohman is also appreciated.
3
-methoxy analog, was also assayed. The methyl derivative 17
did not improve on the activity of 14, and 18 and 20 were
inactive, but the chloro analog 19 is a 310-nM NNRTI (Table
1
). There is clearly room for additional lead optimization in
Supporting Information Available: Synthetic details, NMR
this series, but it has been demonstrated that a near-miss from
and HRMS spectral data for the oxadiazole derivatives in Table 1,

EVolVing False-PositiVes to ActiVe Anti-HIV Agents
Journal of Medicinal Chemistry, 2007, Vol. 50, No. 22 5329
1
the H NMR spectrum of LT00141083, and a figure illustrating
inhibitors. Bioorg. Med. Chem. Lett. 2006, 16, 4174-4177. (c) De
La Rosa, M.; Kim, H. W.; Gunic, E.; Jenket, C.; Boyle, U.; Koh,
Y.-H.; Korboukh, I.; Allan, M.; Zhang, W.; Chen, H.; Xu, W.; Nilar,
S.; Yao, N.; Hamatake, R.; Lang, S. A.; Hong, Z.; Zhang, Z.; Girardet,
J.-L. Tri-substituted triazoles as potent non-nucleoside inhibitors of
the HIV-1 reverse transcriptase. Bioorg. Med. Chem. Lett. 2006, 16,
the complex of 19 and HIV-RT. These materials are available free
of charge via the Internet at http://pubs.acs.org.
References
4444-4449.
(
1) AIDS Epidemic Update: December 2006; UNAIDS/WHO: Geneva,
Switzerland, 2006.
2) (a) Kohlstaedt, L. A.; Wang, J.; Friedman, J. M.; Rice, P. A.; Steitz,
T. A. Crystal structure at 3.5 Å resolution of HIV-1 reverse
transcriptase complexed with an inhibitor. Science 1992, 256, 1783-
(
13) Tirado-Rives, J.; Jorgensen, W. L. Contribution of conformer focusing
to the uncertainty in predicting free energies for protein-ligand
binding. J. Med. Chem. 2006, 49, 5880-5884.
(
(
14) Hann, M. M.; Leach, A. R.; Harper, G. Molecular complexity and
its impact on the probability of finding leads for drug discovery. J.
Chem. Inf. Comput. Sci. 2001, 41, 856-864.
(15) Rizzo, R. C.; Udier-Blagovic, M.; Wang, D. P.; Watkins, E. K.;
Kroeger, Smith, M. B.; Smith, R. H., Jr.; Tirado-Rives, J.; Jorgensen,
W. L. Prediction of activity for non-nucleoside inhibitors with HIV-1
reverse transcriptase based on Monte Carlo simulations. J. Med.
Chem. 2002, 45, 2970-2987.
(16) Blagovi c´ , M. U.; Tirado-Rives, J.; Jorgensen, W. L. Validation of a
model for the complex of HIV-1 reverse transcriptase with non-
nucleoside inhibitor TMC125. J. Am. Chem. Soc. 2003, 125, 6016-
6017.
(17) Das, K.; Clark, A. D., Jr.; Lewi, P. J.; Heeres, J.; de Jonge, M. R.;
Koymans, L. M. H.; Vinkers, H. M.; Daeyaert, F.; Ludovici, D. W.;
Kukla, M. J.; De Corte, B.; Kavash, R. W.; Ho, C. Y.; Ye, H.;
Lichtenstein, M. A.; Andries, K.; Pauwels, R.; de B e´ thune, M.-P.;
Boyer, P. L.; Clark, P.; Hughes, S. H.; Janssen, P. A. J.; Arnold, E.
Roles of conformational and positional adaptability in structure-based
design of TMC125-R165335 (Etravirine) and related non-nucleoside
reverse transcriptase inhibitors that are highly potent and effective
against wild-type and drug-resistant HIV-1 variants. J. Med. Chem.
1790. (b) Smerdon, S. J.; J a¨ ger, J.; Wang, J.; Kohlstaedt, L. A.;
Chirino, A. J.; Friedman, J. M.; Rice, P. A.; Steitz, T. A. Structure
of the binding site for nonnucleoside inhibitors of reverse transcriptase
of human immunodeficiency virus type 1. Proc. Natl. Acad. Sci.
U.S.A. 1994, 91, 3911-3915. (c) De Clercq, E. New approaches
toward anti-HIV chemotherapy. J. Med. Chem. 2005, 48, 1297-
313.
(
3) (a) Jorgensen, W. L.; Ruiz-Caro, J.; Tirado-Rives, J.; Basavapathruni,
A.; Anderson, K. S.; Hamilton, A. D. Computer-aided design of non-
nucleoside inhibitors of HIV-1 reverse transcriptase. Bioorg. Med.
Chem. Lett. 2006, 16, 663-667. (b) Ruiz-Caro, J.; Basavapathruni,
A.; Kim, J. T.; Wang, L.; Bailey, C. M.; Anderson, K. S.; Hamilton,
A. D.; Jorgensen, W. L. Optimization of diarylamines as non-
nucleoside inhibitors of HIV-1 reverse transcriptase. Bioorg. Med.
Chem. Lett. 2006, 16, 668-671. (c) Thakur, V. V.; Kim, J. T.;
Hamilton, A. D.; Bailey, C. M.; Domaoal, R. A.; Wang, L.; Anderson,
K. S.; Jorgensen, W. L. Optimization of pyrimidinyl- and triazinyl-
amines as non-nucleoside inhibitors of HIV-1 reverse transcriptase.
Bioorg. Med. Chem. Lett. 2006, 16, 5664-5667. (d) Kim, J. T.;
Hamilton, A. D.; Bailey, C. M.; Domaoal, R. A.; Wang, L.; Anderson,
K. S.; Jorgensen, W. L. FEP-guided selection of bicyclic heterocycles
in lead optimization for non-nucleoside inhibitors of HIV-1 reverse
transcriptase. J. Am. Chem. Soc. 2006, 128, 15372-15373.
2004, 47, 2550-2560.
(
18) Pyrimidines such as 2-6 are in the class Het-NH-Ph-U in which the
U group is now delivered into the aryl box formed by Tyr181,
Tyr188, Trp229, and Phe227 from the “right” instead of from the
(
4) (a) Blake, J. F.; Laird, E. R. Recent advances in virtual ligand
screening. Annu. Rep. Med. Chem. 2003, 38, 305-314. (b) Leach,
A. R.; Shoichet, B. K.; Peishoff, C. E. Prediction of protein-ligand
interactions. Docking and scoring: successes and gaps. J. Med. Chem.
“
left” as in Figure 1 for 1.
(
19) Ludovici, D. W.; De Corte, B. L.; Kukla, M. J.; Ye, H.; Ho, C. Y.;
Lichtenstein, M. A.; Kavash, R. W.; Andries, K.; de B e´ thune, M.-
P.; Azijn, H.; Pauwels, R.; Lewi, P. J.; Heeres, J.; Koymans, L. M.
H.; de Jonge, M. R.; Van Aken, K. J. A.; Daeyaert, F. F. D.; Das,
K.; Arnold, E.; Janssen, P. A. J. Evolution of anti-HIV drug
candidates. Part 3: Diarylpyrimidine (DAPY) analogues. Bioorg.
Med. Chem. Lett. 2001, 11, 2235-2239.
(20) Heeres, J.; de Jonge, M. R.; Koymans, L. M. H.; Daeyaert, F. F. D.;
Vinkers, M.; Van Aken Koen, J. A.; Arnold, E.; Das, K.; Kilonda,
A.; Hoornaert, G. J.; Compernolle, F.; Cegla, M.; Azzam, R. A.;
Andries, K.; de Bethune, M.-P.; Azijn, H.; Pauwels, R.; Lewi, P. J.;
Janssen, P. A. J. Design, synthesis, and SAR of a novel pyrazinone
series with non-nucleoside HIV-1 reverse transcriptase inhibitory
activity. J. Med. Chem. 2005, 48, 1910-1918.
21) QikProp, version 3.0; Schr o¨ dinger LLC: New York, 2006. QikProp
was used as a subroutine with BOMB for each generated structure.
22) Himmel, D. M.; Das, K.; Clark, A. D., Jr.; Hughes, S. H.; Benjahad,
A.; Oumouch, S.; Guillemont, J.; Coupa, S.; Poncelet, A.; Csoka, I.;
Meyer, C.; Andries, K.; Nguyen, C. H.; Grierson, D. S.; Arnold, E.
Crystal structures for HIV-1 reverse transcriptase in complexes with
three pyridinone derivatives: A new class of non-nucleoside inhibitors
effective against a broad range of drug-resistant strains. J. Med. Chem.
2006, 49, 5851-5855.
(
5) Barreiro, G.; Guimar a˜ es, C. R. W.; Tubert-Brohman, I.; Lyons, T.
M.; Tirado-Rives, J.; Jorgensen, W. L. Search for non-nucleoside
inhibitors of HIV-1 reverse transcriptase using chemical similarity,
molecular docking, and MM-GB/SA scoring. J. Chem. Inf. Model.,
in press.
6) Ren, J.; Esnouf, R. M.; Hopkins, A. L.; Warren, J.; Balzarini, J.;
Stuart, D. I.; Stammers, D. K. Crystal structures of HIV-1 reverse
transcriptase in complex with carboxanilide derivatives. Biochemistry
(
1998, 37, 14394-14403.
(
7) Friesner, R. A.; Banks, J. L.; Murphy, R. B.; Halgren, T. A.; Klicic,
J. J.; Mainz, D. T.; Repasky, M. P.; Knoll, E. H.; Shelley, M.; Perry,
J. K.; Shaw, D. E.; Francis, P.; Shenkin, P. S. Glide: A new approach
for rapid, accurate docking and scoring. 1. Method and assessment
of docking accuracy. J. Med. Chem. 2004, 47, 1739-1749.
8) (a) Jorgensen, W. L.; Maxwell, D. S.; Tirado-Rives, J. Development
and testing of the OPLS all-atom force field on conformational
energetics and properties of organic liquids. J. Am. Chem. Soc. 1996,
(
(
(
118, 11225-11236. (b) Jorgensen, W. L.; Tirado-Rives, J. Chemical
theory and computation special feature: Potential energy functions
for atomic-level simulations of water and organic and biomolecular
systems. Proc. Natl. Acad. Sci U.S.A. 2005, 102, 6665-6670.
9) (a) Still, W. C.; Tempczyk, A.; Hawley, R. C.; Hendrickson, T.
Semianalytical treatment of solvation for molecular mechanics and
dynamics. J. Am. Chem. Soc. 1990, 112, 6127-6129. (b) Jorgensen,
W. L.; Ulmschneider, J. P.; Tirado-Rives, J. Free energies of
hydration from a generalized Born model and an all-atom force field.
J. Phys. Chem. B 2004, 108, 16264-16270.
2005, 48, 7582-7591.
(
(
23) Tsuzuki, S.; Houjou, H.; Nagawa, Y.; Hiratani, K. The second stable
conformation of the methoxy groups of o-dimethoxybenzene: Sta-
bilization of perpendicular conformation by CH-O interaction. J.
Chem. Soc., Perkin Trans. 2 2002, 1271-1273.
(
24) Jorgensen, W. L.; Tirado-Rives, J. Molecular modeling of organic
and biomolecular systems using BOSS and MCPRO. J. Comput.
Chem. 2005, 26, 1689-1700.
25) Alanine, A.; Anselm, L.; Steward, L.; Thomi, S.; Vifian, W.;
Groaning, M. D. Synthesis and SAR evaluation of 1,2,4-triazoles as
A2A receptor antagonists. Bioorg. Med. Chem. Lett. 2004, 14, 817-
821.
(26) Van Dort, M.; Neubig, R.; Counsell, R. E. Radioiodinated p-
iodoclonidine. A high-affinity probe for the R2-adrenergic receptor.
J. Med. Chem. 1987, 30, 1241-1244.
(27) Zinner, G.; Heitmann, M. Arbimide and carbonyl cyclizations of
semicarbazides. Arch. Pharm. 1981, 314, 193-196.
(
(
10) MacroModel, version 9.0; Schr o¨ dinger LLC: New York, 2005.
11) (a) Lin, T. S.; Luo, M. Z.; Liu, M. C.; Pai, S. B.; Dutschman, G. E.;
Cheng, Y. C. Antiviral activity of 2′,3′-dideoxy-â-L-5-fluorocytidine
(
(â-L-EddC) and 2′,3′-dideoxy-â-L-cytidine (â-L-ddC) against hepatitis
B virus and human immunodeficiency virus type 1 in vitro. Biochem.
Pharmacol. 1994, 47, 171-174. (b) Ray, A. S.; Yang, Z.; Chu, C.
K.; Anderson, K. S. Novel use of a guanosine prodrug approach to
convert 2′,3′-didehydro-2′,3′-dideoxyguanosine into a viable antiviral
agent. Antimicrob. Agents Chemother. 2002, 46, 887-891.
(
12) (a) Muraglia, E.; Kinzel, O. D.; Laufer, R.; Miller, M. D.; Moyer,
G.; Munshi, V.; Orvieto, F.; Palumbi, M. C.; Pescatore, G.; Rowley,
M.; Williams, P. D.; Summa, V. Tetrazole thioacetanilides: Potent
non-nucleoside inhibitors of WT HIV reverse transcriptase and its
K103N mutant. Bioorg. Med. Chem. Lett. 2006, 16, 2748-2752. (b)
Wang, Z.; Wu, B.; Kuhen, K. L.; Bursulaya, B.; Nguyen, T. N.;
Nguyen, D. G.; He, Y. Synthesis and biological evaluations of
sulfanyltriazoles as novel HIV-1 non-nucleoside reverse transcriptase
(28) Barreca, M. L.; De Luca, L.; Iraci, N.; Rao, A.; Ferro, S.; Maga, G.;
Chimirri, A. Structure-based pharmacophore identification of new
chemical scaffolds as non-nucleoside reverse transcriptase inhibitors.
J. Chem. Inf. Model. 2007, 47, 557-562.
JM070683U