Optimization of diarylamines as non-nucleoside inhibitors of HIV-1 reverse transcriptase

Article abstract of DOI:10.1016/j.bmcl.2005.10.037

Following computational analyses, potential non-nucleoside inhibitors of HIV-1 reverse transcriptase have been pursued through synthesis and assaying for anti-viral activity. The general class Het-NH-Ph-U has been considered, where Het is an aromatic heterocycle and U is an unsaturated, hydrophobic group. Results for compounds with Het = 2-thiazoyl and 2-pyrimidinyl are the focus of this report.

Full text of DOI:10.1016/j.bmcl.2005.10.037

Bioorganic & Medicinal Chemistry Letters 16 (2006) 668–671
Optimization of diarylamines as non-nucleoside inhibitors of
HIV-1 reverse transcriptase
Juliana Ruiz-Caro,a Aravind Basavapathruni,b Joseph T. Kim,a Christopher M. Bailey,b
Ligong Wang,a,b Karen S. Anderson,b,* Andrew D. Hamiltona and William L. Jorgensena,*
aDepartment of Chemistry, Yale University, New Haven, CT 06520-8107, USA
bDepartment of Pharmacology, Yale University School of Medicine, New Haven, CT 06520-8066, USA
Received 23 September 2005; accepted 11 October 2005
Available online 17 November 2005
Abstract—Following computational analyses, potential non-nucleoside inhibitors of HIV-1 reverse transcriptase have been pursued
through synthesis and assaying for anti-viral activity. The general class Het–NH–Ph–U has been considered, where Het is an
aromatic heterocycle and U is an unsaturated, hydrophobic group. Results for compounds with Het = 2-thiazoyl and 2-pyrimidinyl
are the focus of this report.
Ó 2005 Elsevier Ltd. All rights reserved.
Delineation of general design principles and computa-
tional modeling has led to pursuit of diarylamines as
non-nucleoside inhibitors of HIV reverse transcriptase
(NNRTIs).1 The Het–NH–Ph–U class has been consid-
ered in depth, where Het is an aromatic heterocycle and
U is an unsaturated, hydrophobic group. Lead optimi-
zation has focused to date on Het = 2-thiazoyl and
2-pyrimidinyl derivatives, as described here.
obtained commercially or they were prepared in two
or three steps as in Scheme 2. Activities against the IIIB
strain of HIV-1 were determined using MT-2 human T-
cells; the EC50 values are the dose required to achieve
50% cytopathic protection of the HIV-infected cells
using the MTT colorimetric method.2,3 As listed in
Table 1, the parent compound 4a turned out to be a
10 lM inhibitor. Addition of a methyl, methoxy, or eth-
yl group provides a modest boost, while propyl and iso-
propyl are deactivating. However, substitution with
Thiazoles. Variations of the R-group in the 4-position of
the phenyl ring were explored for the thiazole series 4
with U as O-3,3-dimethylallyl (ODMA) via the synthetic
route in Scheme 1. The requisite aminophenols were
OH
OH
OH
NH2
Cl
Cl
1) NaNO2
2) CuCl, HCl
76%
SnCl2, EtOH
94%
O2N
O2N
H2N
OH
OH
5
6
2d
R
R
N
N
37% HCl
NH2
NH2
+
Br
10% aq. EtOH
80-90ºC
N
H
S
S
R
R
KNO3
1) NaNO2
H2N
H2SO4
2) H2O, 80ºC
1
2
3
O2N
7
94-99%
8
48-60%
O
R
N
OH
R
OH
R
Br
H2 (1 atm)
PtO2
N
H
S
Cs2CO3, acetone
O2N
H2N
4a-h
9
94-99%
2e-g
Scheme 1.
OH
O
CN
CN
1) LiCl, 160ºC, quant.
2) SnCl2.2H2O, 6N HCl, 70%
Keywords: NNRTI; Computer-aided drug design; Diarylamine; Anti-
HIV drug; Thiazoylamine; Pyrimidinylamine.
H2N
O2N
10
2h
*
Corresponding authors. Tel.: +1 203 432 6278; fax: +1 203 432
6299; e-mail: william.jorgensen@yale.edu
Scheme 2.
0960-894X/$ - see front matter Ó 2005 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2005.10.037

J. Ruiz-Caro et al. / Bioorg. Med. Chem. Lett. 16 (2006) 668–671
669
Table 1. Anti-HIV-1 activity (EC50) and cytotoxicity (CC50), lM, for
thiazoles 4 and reference compoundsa
Table 2. Anti-HIV-1 activity (EC50) and cytotoxicity (CC50), lM, for
thiazoles 12–21a
Compound
R
EC50
CC50
Compound R R0O
EC50 CC50
4a
4b
H
CH3
10
23
17
42
26
31
23
31
12d
12h
13d
13h
14d
14h
15d
15h
16d
16h
17d
18d
19d
19h
20d
21d
Cl 2-Furanylmethoxy
4.0 18
1.2 18
7.0 26
3.0
3.8
0.30
5.0
23
CN 2-Furanylmethoxy
Cl 3-Furanylmethoxy
CN 3-Furanylmethoxy
Cl 2-Thienylmethoxy
CN 2-Thienylmethoxy
Cl 3-Thienylmethoxy
CN 3-Thienylmethoxy
Cl 2-Thiazoylmethoxy
CN 2-Thiazoylmethoxy
Cl Benzyloxy
4c
4d
OCH3
Cl
NA
10
4.0
38
4.1
4e
CH2CH3
CH2CH2CH3
CH(CH3)2
CN
4f
4g
NA
NA
0.21
NA
5.5 38
NA
4h
0.75
1.1
4i
d4T
COOH
17
>100
>10
NA 20
4.5 20
6.0 17
1.0 18
5.0 15
3
0.12
1
Nevirapine
UC781
TMC125
0.002
<0.002
>100
>1
Cl (E/Z)-3-Ethyl,3-methylallyloxy
Cl Cyclopent-1-enylmethoxy
CN Cyclopent-1-enylmethoxy
NA
1.7
a NA for EC50 > CC50
.
Cl 2-Methyl-cyclopent-1-enylmethoxy NA 11
NA 15
Cl Cyclohexylmethoxy
CN
a NA for EC50 > CC50
.
O
R
CN
O
Cl
O
S
R'
O
OH
N
N
Br
N
O
R
R
N
H
N
N
H
N
H
N
R'Br, Cs2CO3
N
H2N
O
N
H
N
H
S
S
11
UC781
TMC125
or R'OH, Ph3P, DEAD
3
12-21
chlorine or a cyano group enhances activity to the 0.3
and 0.2 lM levels. These qualitative trends agree well
with the MC/FEP predictions,1 especially the preference
for R = Cl and CN in 4.
Scheme 3.
O
O
R'''
R
R
The position of the chlorine in the computed structure
for the complex of RT with 4d is similar to those found
in computed and X-ray structures for RT-complexes
with other NNRTIs including efavirenz, PNU142721,
UC781, and 8-chloro-TIBO.1 In addition, the cyano
alternative is precedented in TMC120, TMC125, and
rilpivirine. Computation of the electric field of the pro-
tein at the position of the R group in 4 in the complexes
supports the favorability of the interaction of the pro-
tein with the C–Cl and C–CN dipoles. Prior SAR results
on this substituent effect are notable. For analogues of
UC781, comparative data were only reported for
R = Cl, CN, and OCH3.4 These activities show negligi-
ble variation, while the methoxy analogue is 10- to 20-
fold less active here. More data are available for pyraz-
inone 11 and 8-R-TIBO from MT-4 cell-based assays.5,6
The results are again quite different with activities of
0.25, >100, 0.50, 0.025, and <0.001 lM for 11 with
R = H, Me, CF3, Cl, and CN, that is, the relative gain
is much greater for R = CN and the inactivity of the
methyl compound is striking. For the TIBOs, the
activities are 0.044, 0.014, 0.004, 0.034, and 0.056 lM
for 8-R = H, Me, Cl, OMe, and CN; here, CN provides
negligible boost over the parent. Thus, SAR patterns at
analogous sites in NNRTIs are erratic; the rest of the
structure matters.
N
R''Ar
N
H
R''
N
N
H
22
23
groups.1 So far, only other ether derivatives have been
considered (Table 2); synthesis proceeded by base-cat-
alyzed O-alkylation, as for 4, or via a Mitsunobu
reaction (Scheme 3).
In the computational analysis,1 the furanylmethoxy,
thienylmethoxy, and cyclopent-1-enylmethoxy alterna-
tives ranked similarly to ODMA for the U group.
Indeed, activity is found in these cases, for example,
12d, 13d, 14d, 15d, and 19d, though it was surprising
that none surpassed the activity of the ODMA analogue
4d. In all cases, the computations address binding
differences, which are related to, but are not the same
as, the biological readout from the cell-based assays.
Additional factors come into play in the assays includ-
ing permeability, intracellular metabolism, and binding
to alternative cellular components. In vitro binding
assays with recombinant protein are currently underway
to gain further insights in this regard.
Pyrimidines and other heterocycles. The MC/FEP calcu-
lations favored 2-thiazole, 2-oxazole, and 2-pyrimidine
over furans, pyridines, and pyrazine for the heterocycle
in the Het–NH–Ph–U motif.1 The predictions have been
tested so far through synthesis and assaying for the
derivatives of 22 and 23 that are listed in Table 3.
Attention was next turned to alternatives to ODMA
for the unsaturated group. The analyses with the
BOMB program had suggested several promising
possibilities including various heteroarylmethoxyl

670
J. Ruiz-Caro et al. / Bioorg. Med. Chem. Lett. 16 (2006) 668–671
Table 3. Anti-HIV-1 activity (EC50) and cytotoxicity (CC50), lMa
Glu138 of HIV-RT.1,9 Though there is an associated,
greater desolvation penalty in these cases, additional
cellular factors, as mentioned above, may be relevant.
The differences between 23e and f are also curious and
warrant further exploration with recombinant protein.
In any event, several compounds have been identified,
which are significantly more active than nevirapine
(Table 1) or NRTIs,2 so their future depends on their
pharmacological properties and resistance profiles.
Compound R
Ar
R00
R000
EC50 CC50
4a
4d
H
Cl
2-Thiazoyl
2-Thiazoyl
—
—
—
—
10
23
26
80
35
15
0.30
9
22a
22b
22c
22d
22e
22f
23a
23b
23c
23d
23e
23f
23g
23h
23i
OMe 2-Thiazoyl
Me 2-Pyridinyl
4-CH2OH —
H
H
H
H
H
—
—
—
—
—
H
H
H
H
H
H
H
H
H
H
H
NA
3.2
Cl 2-Pyridinyl
CN 2-Pyridinyl
NA
NA
NA
30
0.29
H
Cl
H
Pyrazinyl
Pyrazinyl
23
15
2-Pyrimidinyl H
>100
78
It should be noted that the cytotoxicity (CC50) of all
nitrile derivatives, for example, 4h, 22d, 23c, and of the
amino- and methyl-pyrimidines, for example, 23e,j,n,
and o, has been high, so their further investigation as
NNRTIs is less promising. The origin of cytotoxicity
is unknown, though similar substructures are not
problematic in TMC125 (Table 1), rilpivirine, or
PNU142721. The cytotoxicity of 23n and o was disap-
pointing as these disubstituted pyrimidine derivatives
score particularly well in the binding calculations. Addi-
tional analogues are being pursued along with testing of
the more potent compounds against common mutant
forms of HIV.
Me 2-Pyrimidinyl H
CN 2-Pyrimidinyl H
2.8
NA
0.20
NA
0.065
0.018
0.010
0.32
0.075
NA
NA
<0.2
Cl
Cl
Cl
Cl
Cl
2-Pyrimidinyl H
2.5
<0.2
2.5
2-Pyrimidinyl Me
2-Pyrimidinyl CF3
2-Pyrimidinyl SMe
2-Pyrimidinyl OMe
2.8
9.0
Me 2-Pyrimidinyl OMe
41
23j
Cl
Cl
Cl
Cl
Cl
Cl
Cl
Cl
2-Pyrimidinyl NH2
2-Pyrimidinyl CH2OH
0.5
23k
23l
0.22
>25
28
2-Pyrimidinyl CH2OMe H
23m
23n
23o
23p
23q
2-Pyrimidinyl OMe
2-Pyrimidinyl NH2
2-Pyrimidinyl NH2
2-Pyrimidinyl NH2
2-Pyrimidinyl NH2
6-OMe 0.25
6-Me NA
<0.004
0.014
19
6-Cl
6-OMe 3.0
NA
Predicted properties. Some predicted properties from
QikProp for a selection of the current compounds and
known NNRTIs are summarized in Table 4; the overall
rms errors for QikProp predictions of logP, logS, and
logPCaco are 0.5–0.6 log unit.10,11 For further refer-
ence, we have run QikProp for 119 oral drugs with
reported experimental values for percentage human
absorption (%F). For the 112 compounds with
%F > 10%, only four have MW > 500, three have
QPlogP > 5, and three have QPlogS < ꢀ5.5. For a
larger set of 575 drugs, 90% have MW < 470,
QPlogP < 4.7, QPlogS > ꢀ5.8, and QPPCaco > 12.
The MW and logP limits are similar to those in Lipin-
skiÕs rule-of-five, that is, MW < 500 and logP < 5.13
5-Br
0.70
12
a NA for EC50 > CC50
.
CH2OH
N
OH
CH2OH
N
O
1. TBSCl, imidazole
2. LiOH, DMF
Cl
Cl
3. DMABr, Cs2CO3
4. TBAF
N
N
H
N
N
H
24
23k
Scheme 4.
Synthesis of the heteroaryl compounds was generally
achieved as for 4 with replacement of the bromothiazole
1 with an appropriate heteroaryl halide, for example,
1-bromopyridine or 2-chloro,4-R00-pyrimidine. The
coupling for the 2-chloropyrimidines occurred in ca.
70% yield using TsOH in dioxane.7 For 23k, the pres-
ence of the two hydroxyl groups in the intermediate 24
required inclusion of a selective deprotection step8 as
part of the sequence in Scheme 4, which proceeded in
91% overall yield. Compound 23k was then methylated
(MeI and Cs2CO3) to provide 23l in 72% yield.
Table 4. Predicted properties for selected NNRTIs
Compound
MWa
QPlogPb
QPlogSc
QP PCacod
Nevirapine
Efavirenz
Delavirdine
UC781
TMC120
TMC125
Rilpivirine
4a
266.3
315.7
456.6
335.8
329.4
435.3
366.4
260.4
294.8
255.3
289.8
303.8
357.8
335.9
319.8
304.8
319.8
2.5
3.5
2.8
5.1
3.9
2.5
3.4
3.5
4.1
3.7
4.1
4.5
5.0
4.8
4.3
3.0
3.1
ꢀ3.6e
ꢀ5.0
ꢀ5.6
ꢀ5.8
ꢀ6.3
ꢀ6.5
ꢀ6.4
ꢀ3.8
ꢀ4.6
ꢀ4.1
ꢀ4.6
ꢀ4.9
ꢀ5.8
ꢀ5.2
ꢀ4.7
ꢀ3.9
ꢀ4.0
2090
1586
285
7570
779
62
150
As predicted,1 the thiazoles and pyrimidines, for exam-
ple, 4d and 23d, were much more active than the corre-
sponding 2-pyridines and pyrazines, for example, 22c
and f. This encouraged further optimization for R00
and R000 in the pyrimidine series 23. Models were con-
structed and scored with BOMB, which supported the
addition of small groups including OCH3, NH2, and
CH2OH. Subsequent synthesis readily provided several
potent compounds, for example, 23f–h and j. Addition
of a 5-Br substituent, as in TMC125, to 23j to yield
23q was not productive. Furthermore, the lack of activ-
ity for 23k was surprising. As for the amino group in
TMC125 and 23j, the hydroxymethyl groups in 23k
and 22a are predicted to form a hydrogen bond with
3775
3750
3733
3942
4303
4006
6542
5668
1211
1721
4d
23a
23d
23e
23f
23g
23h
23j
23k
a Molecular weight.
b Predicted octanol/water logP from QikProp, v 2.3.
c Predicted aqueous solubility from QikProp, v 2.3; S in mol/l.
d Predicted Caco-2 cell permeability in nm/s from QikProp, v 2.3.
e An experimental value of ꢀ3.2 has been reported in Ref. 12.

J. Ruiz-Caro et al. / Bioorg. Med. Chem. Lett. 16 (2006) 668–671
671
CN
Acknowledgment
NHSO2CH3
N
CN
Gratitude is expressed to the National Institutes of Health
(AI44616, GM35208, and GM49551) for support.
N
HN
NH
N
N
N
N
H
NH
O
References and notes
delavirdine
rilpivirine
1. Jorgensen, W.L.; Ruiz-Caro, J.; Tirado-Rives, J.; Basava-
pathruni,A.;Anderson,K.S.;Hamilton,A.D.Bioorg.Med.
Chem. Lett., in press, doi:10.1016/j.bmcl.2005.10.038.
2. Lin, T. S.; Luo, M. Z.; Liu, M. C.; Pai, S. B.; Dutschman,
G. E.; Cheng, Y. C. Biochem. Pharmacol. 1994, 47, 171.
3. Ray, A. S.; Yang, Z.; Chu, C. K.; Anderson, K. S.
Antimicrob. Agents Chemother. 2002, 46, 887.
4. Balzarini, J.; Brouwer, W. G.; Dao, D. C.; Osika, E. M.; De
Clercq, E. Antimicrob. Agents Chemother. 1996, 40, 1454.
5. Heeres, J.; de Jonge, M. R.; Koymans, L. M. H.;
Daeyaert, F. F. D.; Vinkers, M.; Van Aken, K. J. A.;
Arnold, E.; Das, K.; Kilonda, A.; Hoornaert, G. J.;
Compernoolle, F.; Cegla, M.; Azzam, R. A.; Andries, K.;
The most potent compounds reported here, 23g and h,
compare favorably with all of these limits, so optimism
can be expressed for acceptable oral bioavailability.
In general, NNRTIs are relatively hydrophobic, so solu-
bility is expected to be more problematic than cell perme-
ability.1,14 This is reflected in Table 4, and the compounds
whose oral bioavailabilities are known to be low, delavir-
dine, UC781, and TMC125, have QPlogS values of ꢀ5.5
or less. Many factors can limit bioavailability, but poor
aqueous solubility is not a desirable starting point.14
Rilpivirine is also predicted and reported15 to have very
low aqueous solubility and high serum–protein binding,
though the oral bioavailability in rats and dogs is 31–
32%.15 It has been proposed that rilpivirine gains bio-
availability by formation of aggregates in a specific size
range, which are absorbed and transported through the
lymphatic system.15,16 Rational design for such a feature
is currently elusive, so the maxim remains that oral for-
mulation of compounds with logS below ca. ꢀ5.5 is likely
to be difficult.11
´
de Bethune, M.-P.; Azijn, H.; Pauwels, R.; Lewi, P. J.;
Janssen, P. A. J. J. Med. Chem. 2004, 47, 2550.
6. Ho, W.; Kukla, M. J.; Breslin, H. J.; Ludovici, D. W.;
Grous, P. P.; Diamond, C. J.; Miranda, M.; Rodgers, J. D.;
Ho, C. Y.; De Clercq, E.; Pauwels, R.; Andries, K.; Janssen,
M. A. C.; Janssen, P. A. J. Med. Chem. 1995, 38, 794.
7. Bursavich, M. G.; Lombardi, S.; Gilbert, A. M. Org. Lett.
2005, 7, 4113.
8. Ankala, S. V.; Fenteany, G. Tetrahedron Lett. 2002, 43, 4729.
´
9. Udier-Blagovic, M.; Tirado-Rives, J.; Jorgensen, W. L. J.
Am. Chem. Soc. 2003, 125, 6016.
10. Jorgensen, W. L. QikProp, v 2.3; Schro¨dinger LLC:
New York, 2005.
11. Jorgensen, W. L.; Duffy, E. M. Adv. Drug Deliv. Rev.
2002, 54, 355.
12. Morelock, M. M.; Choi, L. L.; Bell, G. L.; Wright, J. L. J.
Pharm. Sci. 1994, 83, 948.
13. Lipinski, C. A.; Lombardo, F.; Domini, B. W.; Feeney, P.
J. Adv. Drug. Deliv. Rev. 1997, 23, 3.
14. Jorgensen, W. L. Science 2004, 303, 1813.
15. Janssen, P. A. J.; Lewi, P. J.; Arnold, E.; Daeyaert, F.;
de Jonge, M.; Heeres, J.; Koymans, L.; Vinkers, M.;
Guillemont, J.; Pasquier, E.; Kukla, M.; Ludovici, D.;
In conclusion, a joint computational and experimental
study has been carried out to develop new NNRTIs
which combine good efficacy, desirable pharmacologi-
cal properties, and ease of synthesis. This general goal
for drug discovery has been pursued here with the aid
of a computational approach featuring lead genera-
tion with the ligand-growing program BOMB, lead-
optimization with free-energy perturbation calcula-
tions, and prediction of pharmacological properties
with QikProp. In concert with synthetic forethought
in proposing lead templates, the present approach
facilitated identification of 10–30 lM lead compounds,
which could rapidly be refined to a 10 nM NNRTI
with predicted properties that are auspicious for oral
formulation.
´
Andries, K.; de Bethune, M.-P.; Pauwels, R.; Das, K.;
Clark, A. D., Jr.; Frenkel, Y. V.; Hughes, S. H.; Medaer,
B.; de Knaep, F.; Bohets, H.; De Clerck, F.; Lampo, A.;
Williams, P.; Stoffels, P. J. Med. Chem. 2005, 48, 1901.
16. Frenkel, Y. V.; Clark, A. D., Jr.; Das, K.; Wang, Y.-H.;
Lewi, P. J.; Janssen, P. A. J.; Arnold, E. J. Med. Chem.
2005, 48, 1974.