Optimization of pyrimidinyl- and triazinyl-amines as non-nucleoside inhibitors of HIV-1 reverse transcriptase

Article abstract of DOI:10.1016/j.bmcl.2006.08.037

Non-nucleoside inhibitors of HIV-1 reverse transcriptase are being pursued through synthesis and assaying for anti-viral activity. Following computational analyses, the focus has been on the motif Het-NH-Ph-U, where Het is an aromatic heterocycle and U is an unsaturated, hydrophobic group. Previous investigations with Het = 2-thiazoyl and 2-pyrimidinyl are extended here to triazinyl derivatives. The result is several NNRTIs in the 2-20 nM range with negligible cytotoxicity and auspicious predicted pharmacological properties.

Full text of DOI:10.1016/j.bmcl.2006.08.037

Bioorganic & Medicinal Chemistry Letters 16 (2006) 5664–5667
Optimization of pyrimidinyl- and triazinyl-amines as non-nucleoside
inhibitors of HIV-1 reverse transcriptase
Vinay V. Thakur,a Joseph T. Kim,a Andrew D. Hamilton,a Christopher M. Bailey,b
Robert A. Domaoal,b Ligong Wang,a,b Karen S. Andersonb,* 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 11 July 2006; accepted 1 August 2006
Available online 22 August 2006
Abstract—Non-nucleoside inhibitors of HIV-1 reverse transcriptase are being pursued through synthesis and assaying for anti-viral
activity. Following computational analyses, the focus has been on the motif Het–NH–Ph–U, where Het is an aromatic heterocycle
and U is an unsaturated, hydrophobic group. Previous investigations with Het = 2-thiazoyl and 2-pyrimidinyl are extended here to
triazinyl derivatives. The result is several NNRTIs in the 2–20 nM range with negligible cytotoxicity and auspicious predicted phar-
macological properties.
Ó 2006 Elsevier Ltd. All rights reserved.
The AIDS crisis and the need for more effective thera-
peutic agents to combat HIV infection continue.1 In re-
O
O
O
cent papers, results were presented on our initial efforts
to design non-nucleoside inhibitors of HIV reverse
transcriptase (NNRTIs).2,3 In particular, diarylamines
have been pursued in a motif, Het–NH–Ph–U, where
Het is a heterocycle and U is an unsaturated, hydropho-
bic group. Computational analyses featuring free energy
perturbation (FEP) calculations directed us toward
Het = 2-thiazolyl and 2-pyrimidinyl derivatives. The
parent compounds 1 and 2 (X = H, R = H) turned out
to be 10–30 lM inhibitors in an MT-2 cell protection as-
say; modest lead optimization provided the chloro,
methoxy derivative of 2 (X = Cl, R = OMe) as a 10-
nM NNRTI.2,3 Subsequent FEP computations indicat-
ed that triazines 3 should also have promising activity,
and further motivation for their pursuit was provided
by possible avoidance of the cytotoxicity that plagued
some of the more active pyrimidines, particularly with
X = CN.3 Low solubility is also a common issue with
NNRTIs, and the triazines could be expected to show
benefits in that regard.4 Key findings from the resultant
investigations are summarized here.
X
X
X
S
N
N
N
N
R
N
H
N
N
H
R
N
N
H
1
2
3
Chemistry and biology. Syntheses of the 2-thiazole
and 2-pyrimidine derivatives were described previous-
ly and followed the general route in Scheme 1.3 For
the triazene analogs, dichloro and trichlorotriazine
were used as starting materials. They were typically
added to the preformed dimethylallyl (DMA)-deriva-
tized aminophenol, which was prepared by SnCl2
reduction of the corresponding nitrophenol with sub-
sequent substitution of chlorines on the triazine. The
routes for a di- and a tri-substituted triazine are
shown in Scheme 2. The parent triazine 3 (R = H)
was not easily available owing to instability of chlo-
rotriazine. Monochloro triazines such as 5 (R = Cl)
could also be converted to the amino derivatives in
near quantitative yield by treatment with the amine
in methanol. The thiomethoxy analog of 5 (R = H)
was prepared in 77% yield using NaSMe in THF
rather than NaOMe.
Keywords: NNRTI; Anti-HIV; Structure-based drug design; Triazi-
nylamine; 2-Pyrimidinylamine; Free energy perturbation.
A variety of alternatives for the U group was also tested
in the 2-thiazole series, and ODMA emerged as pre-
ferred for potency.3 Further investigations were made
*
Corresponding authors. Tel.: +1 203 432 6278; fax: +1 203 432 6299
(W.L.J.); e-mail: william.jorgensen@yale.edu
0960-894X/$ - see front matter Ó 2006 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2006.08.037

V. V. Thakur et al. / Bioorg. Med. Chem. Lett. 16 (2006) 5664–5667
5665
OH
Activities against the IIIB strain of HIV-1 were deter-
mined using MT-2 human T-cells; the EC50 values are
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% was ob-
tained simultaneously.6,7
OH
X
X
TsOH
N
N
+
dioxane
N
H
R
N
Cl
R
N
H2N
ODMA
X
Activities and cytotoxicities. Results are listed in Table 1
for the triazene and pyrimidine derivatives 7–10. Some
general patterns for equivalently substituted systems
are: (1) the triazenes are less active than the pyrimidines
by a factor of 3–5; (2) the cyano analogs (X = CN) are
more active than the chloro analogs (X = Cl) by a factor
N
Br
N
H
Cs2CO3, acetone
R
N
Scheme 1.
ODMA
R
N
R
ODMA
Table 1. Anti-HIV-1 activity (EC50) and cytotoxicity (CC50), lM, for
triazene and pyrimidine derivatives
CN
CN
NaHCO3
N
N
N
N
+
90%, R=H
R
Y
R
Y
N
H
Cl
Cl
N
Cl
H2N
4
X
X
N
N
N
ODMA
R
CN
NaOMe
R'
N
N
H
R'
N
N
H
N
N
82%, R=H
9 (X=Cl, Y=ODMA)
10 (X=CN, Y=ODMA)
11 (X=Cl, R'=H)
N
H
MeO
N
7 (X=Cl, Y=ODMA)
8 (X=CN, Y=ODMA)
5
Me
N
Cl
a
b
Compound
X
R
R0
EC50
CC50
MeMgBr
73%
i. 4, NaHCO3, 98%
ii. NaOMe, 95%
Cl
N
N
N
N
7a
7b
Cl
Cl
Cl
OMe
NH2
H
NA
45.0
18.0
48.0
3.1
5 (R = Me)
H
0.031
0.390
0.031
12.0
Cl
Cl
N
Cl
7c
7d
Cl
Cl
H
NHMe
Me
H
Cl
Scheme 2.
7e
Cl
Cl
40.0
26.0
12.0
20.0
>100
42.0
7f
7g
OMe
OMe
OMe
Cl
Cl
NA
Cl
Cl
Me
OMe
H
0.100
0.290
0.310
0.011
0.015
0.022
0.005
0.009
0.010
0.075
0.006
0.250
0.018
3.0
7h
8a
for the 2-pyrimidines. Attachment was normally effected
by base-catalyzed substitution (Scheme 1) or by alcohol
coupling under Mitsunobu conditions (Ph3P, DEAD).3
However, phenyl and tolyl ether analogs, Het–NH–
PhX–OPhR0, were prepared by coupling (R0Ph)2I+ salts
and the phenol derivative, while preparation of a thioe-
ther analog utilized PhSCu to yield selectively the de-
sired meta relationship with the amino group in 6
(Scheme 3).5
CN
CN
CN
CN
CN
CN
Cl
8b
8c
OMe
NH2
OMe
SMe
OMe
OMe
NH2
NHMe
OMe
SMe
OMe
OEt
H
H
0.20
>100
8.4
8h
8i
OMe
H
8j
9b
NH2
H
0.11
9.0
9c
9d
Cl
Cl
H
0.50
0.69
28.0
2.8
H
9h
9i
Cl
Cl
OMe
H
9j
9k
Cl
Cl
NH2
H
19.0
19.0
>25
0.22
2.5
PhR'
0.041
NA
NA
O
OH
9l
9m
Cl
Cl
CH2OMe
CH2OH
H
H
(PhR')2I+BF4
-
Cl
Cl
H
N
N
9n
9o
Cl
Cl
H
0.200
0.039
0.016
0.002
NA
Cu, Et3N, 67-85%
(R = H, OMe)
Me
Et
H
0.15
0.33
0.23
0.041
0.022
0.810
0.036
>10
>0.10
>1
N
H
N
H
R
N
R
N
9p
10b
Cl
H
CN
CN
CN
CN
CN
OMe
NH2
NHMe
OMe
H
H
Br
SPh
10c
10d
H
i. PhSCu, py., quin.
220 oC, 81%
Cl
Cl
H
0.005
0.160
0.017
0.110
0.002
0.002
10h
10n
OMe
H
ii. SnCl2, 95%
O2N
H2N
Ph
Nevirapine
Efavirenz
TMC125
6
S
Cl
6, TsOH, 100 o
C
N
N
a For 50% protection in MT-2 cells; antiviral curves used triplicate
60%
N
H
MeO
N
MeO
N
Cl
samples at each concentration. NA for EC50 > CC50
.
b For 50% inhibition of MT-2 cell growth; toxicity curves also used
triplicate samples.
Scheme 3.

5666
V. V. Thakur et al. / Bioorg. Med. Chem. Lett. 16 (2006) 5664–5667
of 3–10; (3) compounds with a single OMe, SMe, or
NHMe substituent on the heterocycle are particularly
potent, that is, 7–10b, 10d, 10i; (4) the triazenes show lit-
tle cytotoxicity with the exception of the amino, cyano
compounds 8c and 8j; (5) the cyano pyrimidines are
both potent and cytotoxic, though 10b has a safety mar-
gin (CC50/EC50) > 100; and (6) many of the compounds
are highly potent with EC50 values below 20 nM, and
three of the potent triazines (8b, 8h, 8i) also have safety
margins >1000. Understanding of the origins and varia-
tions of the cytotoxicity is lacking. Since it was more
pronounced for the 2-pyrimidines than the 2-thiazoles,
a recognition element associated with the heterocycle
in the present NNRTI series appeared to be operative.
Indeed, this notion is further supported by the favorable
results for the triazenes.
Figure 1. Typical snapshot of 7b bound to HIV-RT from an MC
simulation. Carbon atoms of 7b are gold; from the left, Tyr181,
Tyr188, Phe227, Leu100, Lys101; Trp229 at the top, Val106 at the
bottom. H-bond with Lys101 O on right. Some residues in front
including Glu138 have been removed for clarity. The water on N5 is
also H-bonded to a carboxylate O of Glu138.
Results are also included in Table 1 for three reference
NNRTIs, nevirapine (ViramuneÒ), efavirenz (SustivaÒ),
and TMC125 (etravirine). The present compounds are
considerably more effective against WT HIV-1 than
nevirapine, and the most active ones are in the low
nM-range like efavirenz and TMC125. Of course, phar-
macologically important properties of the NNRTIs are
also relevant,2,3 as discussed further below.
in the FEP calculations, so the positioning of a substitu-
ent, for example, whether it would correspond to R or
R0 in 9 in a complex with RT, was not addressed in de-
tail. In fact, by visualization, energy minimizations, and
even Monte Carlo (MC) simulations in explicit water it
was ambiguous which position would be preferred for
the methoxy group in 9b. MC/FEP calculations were
subsequently performed at 25 °C to address this issue;
they followed standard protocols2,8 including the use
of 178 residues of RT, 1250 (complex) and 2000
(unbound) TIP4P water molecules, the OPLS/CM1A
force field,9 and 14 free-energy windows for each
perturbation.
The results for the alternative choices for the U group in
the pyrimidine series are listed in Table 2. As with the
2-thiazoles for which various heteroarylmethoxy options
for U were tried,3 no choice has emerged superior to
ODMA. Removal of the Z-methyl group of ODMA
to yield the E-buten-2-yl ethers, 11a and 11c, reduces
potency ca. 10-fold. However, the structural analyses
below, including Figure 1, suggest that this change could
be advantageous for increased resilience to the Y181C
variant of HIV-RT. The phenyl ethers 11b and 11d are
ca. 100-fold less potent than the DMA ethers, and only
the m-tolyl analog 11f showed somewhat improved per-
formance. Among the aryl ethers, the thioether 11h was
the most potent, though its EC50 is still 32-fold higher
than that for the DMA ether 9b.
The preferred orientation of a methyl group was ad-
dressed first. To assist in visualization, Figure 1 can be
consulted. The bound inhibitors adopt a right-handed
helical form. The substituent on the heterocycle can
either be directed ‘in’ toward Tyr181 and Tyr188, as
illustrated for 7b, or ‘out’ toward Lys101. During the
simulations, interconversion between ‘in’ and ‘out’ never
occurred. MC/FEP calculations were performed for
Molecular modeling. In the previous study,2 structures
were built for complexes of the 2-thiazoyl and 2-pyri-
midinyl NNRTIs, and they were validated by the good
accord between FEP-computed and observed relative
activities for variation of Het and Y in Het–NH–PhY–
ODMA inhibitors. The heterocycles were unsubstituted
the closed cycle: Me2-9 ! Mein-9 ! H2-9 ! Meout
-
9 ! Me2-9. The hysteresis for the free energies of
binding using this cycle was 0.40 0.25 kcal/mol. Then,
additional MC/FEP calculations were performed to
convert OMe, SMe, and other small substituents to
the methyl analogs, which yielded the relative free ener-
gies of binding in Table 3.
Table 2. Anti-HIV-1 activity (EC50) and cytotoxicity (CC50), lM, for
derivatives of pyrimidine 11a
Compound
R
Y
EC50
CC50
9n
H
H
H
ODMA
(E)-OCH2CH@CHCH3
OPh
0.200
2.3
2.5
31.0
30.0
9.0
There is a very strong preference, 6–10 kcal/mol, for the
Me, OMe, SMe, and OEt groups to be ‘in.’ They project
into the pocket near Ca and Cb of Tyr181 and Tyr188
(Fig. 1). The pocket cannot accommodate substituents
larger than OEt, and there is some reduction in activity
in going from OMe to OEt. Water is not observed in this
region during the MC simulations. Thus, N3 of the
polyazines does not participate in hydrogen bonding in
the complexes. Also, the pocket is formed for the unsub-
stituted cases, so there is no penalty for further cavity
11a
11b
9b
13.0
OMe ODMA
OMe (E)-OCH2CH@CHCH3
OMe OPh
0.010
11c
11d
11e
11f
11g
11h
0.089 17.0
2.5
NA
38.0
13.0
OMe O–o-MePh
OMe O–m-MePh
OMe O–p-MePh
OMe SPh
0.540 18.0
10.0 >100
0.320 21.0
a NA for EC50 > CC50
.

V. V. Thakur et al. / Bioorg. Med. Chem. Lett. 16 (2006) 5664–5667
5667
Table 3. MC/FEP results for relative free energies of binding for
analogs of 9 with HIV-RT
Table 4. Predicted properties for selected NNRTIs
MWa
QPlogPb
QPlogSc
QPPCaco
d
Compound
Compound
Rin
R0out
DDGa
EC50 (lM)
Nevirapine
Eefavirenz
266.3
315.7
456.6
335.8
366.4
435.3
320.8
319.8
311.3
341.4
327.4
319.8
333.8
310.4
2.5
3.5
2.6
5.1
3.3
2.7
4.1
3.7
2.9
3.4
3.4
4.6
5.1
3.5
ꢀ3.2
ꢀ5.0
ꢀ5.7
ꢀ5.7
ꢀ6.5
ꢀ6.7
ꢀ4.8
ꢀ4.9
ꢀ5.1
ꢀ5.5
ꢀ5.7
ꢀ5.1
ꢀ5.2
ꢀ5.4
2090
1585
218
9n
H
H
0.0
ꢀ0.78 0.18
3.00 0.14
0.200
Me
H
Me
Me
Delavirdine
UC781
6717
150
Rilpivirine
TMC125
9o
9p
9h
Me
Et
H
H
ꢀ3.23 0.41
ꢀ4.18 0.15
0.50 0.40
0.06 0.31
0.039
0.016
0.250
75
7b
7d
8b
8h
8i
2911
1745
739
OMe
OMe
OMe
Me
1153
797
9b
9k
OMe
H
H
OMe
ꢀ2.41 0.26
0.010
0.041
9b
9k
10b
4512
4209
1062
4.99 0.28
OEt
H
H
OEt
ꢀ1.39 0.31
7.49 0.41
a Molecular weight.
b Predicted octanol/water logP from QikProp, v 3.0.
9m
9i
H
CH2OH
0.05 0.32
>0.220
0.018
c Predicted aqueous solubility from QikProp, v 3.0; S in mol/L.
SMe
H
H
ꢀ2.68 0.29
7.21 0.31
ꢀ2.51 0.28
5.63 0.26
ꢀ1.89 0.26
d Predicted Caco-2 cell permeability in nm/s from QikProp, v 3.0.
SMe
NH2
Me
Me
NH2
Cl
NH2
a DDG in kcal/mol, corrected by RTln2 (+0.41 kcal/mol) for conformer
loss on binding when R 5 R0.
In conclusion, lead optimization in the Het–NH–Ph–U
motif with Het = 2-pyrimidinyl and triazinyl has led to
the discovery of highly potent NNRTIs with low cyto-
toxicity and auspicious predicted properties. Develop-
ment continues including crystallography and assaying
against mutant HIV strains.
growth to accommodate the small substituents. There is
no serious steric problem for the ‘out’ orientation; its
disfavoring appears more associated with interference
with hydration of the side chains of Glu138, Lys101,
and Lys103. The H-bond donating groups NH2 and
CH2OH do favor being ‘out’ since their H-bonding
needs are not satisfied if they are ‘in;’ in both cases they
donate an H-bond to the COOꢀ of Glu138.
Acknowledgment
Gratitude is expressed to the National Institutes of
GM32136,
Health (AI44616,
GM49551) for support.
GM35208,
and
The preferences are expected to carry over to the triazines.
Furthermore, as shown in Figure 1, a water molecule is
found to donate an H-bond to N5. This water is also
donating an H-bond to a carboxylate O of Glu138 and
accepting H-bonds from two water molecules; the one
on the left is also donating an H-bond to the O@C of
Tyr181. Thus, N5 is well accommodated in the complexes,
but it is also well accommodated in the unbound state, so
there is no binding boost over the pyrimidines. Also, the
‘in’ alkoxy oxygen is rarely hydrogen-bonded to a water
molecule in any of the MC runs for bound 7b, 9b, 9h,
and 9k. Finally, it is noted that the trends in the free ener-
gy results correlate well with the observed activities (Table
3). The methyl and ethyl analogs are just predicted to be
somewhat better bound than their activities reflect.
References and notes
1. 2006 Report on the global AIDS Epidemic; UNAIDS:
Geneva, 2006.
2. Jorgensen, W. L.; Ruiz-Caro, J.; Tirado-Rives, J.; Basa-
vapathruni, A.; Anderson, K. S.; Hamilton, A. D. Bioorg.
Med. Chem. Lett. 2006, 16, 663.
3. Ruiz-Caro, J.; Basavapathruni, A.; Kim, J. T.; Bailey, C.
M.; Wang, L.; Anderson, K. S.; Hamilton, A. D.;
Jorgensen, W. L. Bioorg. Med. Chem. Lett. 2006, 16,
668.
4. Jorgensen, W. L.; Duffy, E. M. Adv. Drug Deliv. Rev.
2002, 54, 355.
5. Gong, D.; Li, J.; Yuan, C. Synth. Commun. 2005, 35,
55.
Predicted properties. Some predicted properties from
QikProp for a selection of the current compounds and
other NNRTIs are summarized in Table 4; rms errors
for QikProp predictions are 0.5–0.6 log unit.10 When
QikProp is run on 1700 known oral drugs,11 90%
have MW < 470, QPlogP < 5.0, QPlogS > ꢀ5.7, and
QPPCaco > 22 nm/s. Many of the potent compounds
reported here compare favorably with all these limits,
so optimism can be expressed for acceptable oral bio-
availability. Predicted solubilities, in particular, show
significant improvement over rilpivirine and TMC125,
which are in clinical trials.
6. Lin, T. S.; Luo, M. Z.; Liu, M. C.; Pai, S. B.; Dutschman,
G. E.; Cheng, Y. C. Biochem. Pharmacol. 1994, 47, 171.
7. Ray, A. S.; Yang, Z.; Chu, C. K.; Anderson, K. S.
Antimicrob. Agents Chemother. 2002, 46, 887.
8. Jorgensen, W. L.; Tirado-Rives, J. J. Comput. Chem. 2005,
26, 1689.
9. Jorgensen, W. L.; Tirado-Rives, J. Proc. Natl. Acad. Sci.
2005, 102, 6665.
10. QikProp,
2006.
v
3.0; Schro¨dinger LLC: New York,
11. Proudfoot, J. R. Bioorg. Med. Chem. Lett. 2005, 15,
1087.