Discovery of piperidin-4-yl-aminopyrimidines as HIV-1 reverse transcriptase inhibitors. N-Benzyl derivatives with broad potency against resistant mutant viruses

Article abstract of DOI:10.1016/j.bmcl.2010.05.040

An analysis of the binding motifs of known HIV-1 non-nucleoside reverse transcriptase inhibitors has led to discovery of novel piperidine-linked aminopyrimidine derivatives with broad activity against wild-type as well as drug-resistant mutant viruses. Notably, the series retains potency against the K103 N/Y181C and Y188L mutants, among others. Thus, the N-benzyl compound 5k has a particularly attractive profile. Synthesis and SAR are presented and discussed, as well as crystal structures relating to the binding motifs.

Full text of DOI:10.1016/j.bmcl.2010.05.040

Bioorganic & Medicinal Chemistry Letters 20 (2010) 4215–4218
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
Discovery of piperidin-4-yl-aminopyrimidines as HIV-1 reverse transcriptase
inhibitors. N-Benzyl derivatives with broad potency against resistant
mutant viruses
a
b,
Denis J. Kertesz a, , Christine Brotherton-Pleiss , Minmin Yang , Zhanguo Wang b, Xianfeng Lin b,
Zongxing Qiu b, Donald R. Hirschfeld a, Shelley Gleason a, Taraneh Mirzadegan a,à, Pete W. Dunten a,§
Seth F. Harris a,–, Armando G. Villaseñor a, Julie Qi Hang a, Gabrielle M. Heilek a, Klaus Klumpp a
*
,
a Roche Palo Alto LLC, 3431 Hillview Avenue, Palo Alto, CA 94304, USA
b Roche R&D Center (China), Shanghai 201203, China
a r t i c l e i n f o
a b s t r a c t
Article history:
An analysis of the binding motifs of known HIV-1 non-nucleoside reverse transcriptase inhibitors has led
to discovery of novel piperidine-linked aminopyrimidine derivatives with broad activity against wild-
type as well as drug-resistant mutant viruses. Notably, the series retains potency against the K103 N/
Y181C and Y188L mutants, among others. Thus, the N-benzyl compound 5k has a particularly attractive
profile. Synthesis and SAR are presented and discussed, as well as crystal structures relating to the bind-
ing motifs.
Received 24 March 2010
Revised 10 May 2010
Accepted 12 May 2010
Available online 16 May 2010
Keywords:
HIV
Ó 2010 Elsevier Ltd. All rights reserved.
Antiviral
NNRTI
Reverse transcriptase
Inhibitor
Aminopyrimidine
Viral mutants
Treatment of HIV has become more efficacious in recent years
in large part due to use of combination therapies employing sets
of drugs with different mechanisms of action.1 Non-nucleoside re-
verse transcriptase inhibitors (NNRTIs) are valuable components of
these treatment regimens. These drugs bind in an allosteric site
near the catalytic region of the HIV-1 RT enzyme, thereby inducing
changes in protein geometry which effectively disable the enzyme
from carrying out its function.2 However, mutation of the RT en-
zyme is facile and has resulted in viable viral strains resistant to
first generation NNRTIs.3 Thus, drugs such as Efavirenz and Nevira-
pine have been in clinical use for some time, and are quite effective
against the wild-type virus found in treatment-naive patients.
However, they have a low barrier to resistance and are less effec-
tive in treatment-experienced patients who carry viruses with
NNRTI resistant mutations.4
Given the effectiveness of the initial NNRTI drugs, considerable
effort has been applied towards finding new molecules which retain
activity against the important mutant viruses.5 Among the strains
most often associated with resistance in the clinic are those with
mutations at K103N and Y181C, found singly or in combination with
other variants in the NNRTI binding site.6 For example, K103N/
Y181C (KY) is a frequently encountered double mutant.7 We report
here a new effort towards finding NNRTIs which are broadly active
against resistant viruses as well as the wild-type virus.
Etravirine (TMC125) is a second generation NNRTI which has
recently been approved for clinical use (Fig. 1).8 It has good potency
against wild-type (WT) and many resistant viruses, but its devel-
opment was challenged by poor solubility.9 The benzophenone
compounds GW-678248 and GW-69564 have also been described
as having broad potency against WT and the important mutants.10
In addition, oral bioavailability of these compounds could be en-
hanced by conversion to N-acyl prodrugs, and GW-678248 was
developed as such.11 We considered combining the perceived
advantages of Etravirine and the benzophenone series into a single
molecule.
* Corresponding author. Tel.: +1 650 961 4224.
E-mail address: deniskertesz@comcast.net (D.J. Kertesz).
Present address: Pharmablock R&D Co. Ltd, 10 Xuefu Rd., Pukou, Nanjing 210061,
PR China.
à
Present address: Johnson and Johnson R & D, LLC, San Diego, CA 92121, USA.
Present address: Stanford Synchrotron Radiation Lightsource, Stanford University,
§
Menlo Park, CA 94025, USA.
–
Present address: Genentech, 1 DNA Way, South San Francisco, CA 94080, USA.
0960-894X/$ - see front matter Ó 2010 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2010.05.040

4216
D. J. Kertesz et al. / Bioorg. Med. Chem. Lett. 20 (2010) 4215–4218
H
ising for attaching a phenylsulfonamide head group to an amino-
Cl
O
H2N
Br
N
O
N
pyrimidine core/tail unit via an appropriate linker unit. Thus, we
set out to make the structures necessary to test our hypothesis.
We began with the preparation of alkyl-chain linked amides
such as 4 (Scheme 1). Syntheses proceeded via tandem nucleo-
philic displacements on 2,4-dichloropyrimidines, the first being
with 2,6-dimethyl-4-cyanophenol to give reagents 1 and 2. Further
treatment with amines such as 3 (prepared as shown from pro-
tected b-alanine) gave products as 4.
O
O
SO2NH2
N
H
N
CN
GW-678248: X = Cl, Y = CN
GW-69564: X = F, Y = CF3
X
CN
Etravirine
(TMC125)
Y
Figure 1. Structures of Etravirine and benzophenones.
Inhibitory activities of the products were measured against
recombinant wild-type and mutant enzymes.13 Bioassay results
with 4 were disappointing (Table 1), but the concept remained of
sufficient interest that a parallel synthesis effort was initiated to
study other types of linkers. Intermediates 1 and 2 were then trea-
ted with a variety of commercially available amines, and the activ-
ities of the most interesting products (e.g., 5–7) are also shown in
Table 1. The N-benzyl piperidine 5b was found to have the best
activity profile, and a detailed SAR exploration of that scaffold
was initiated. The N-phenyl piperidine 6 was less active but re-
mained of interest for future work.
A co-crystal structure of Etravirine bound in the NNRTI pocket
was generated in our laboratories,12 and was overlaid with the
published structure of GW-69564, as seen in Figure 2. One evident
difference in the way these compounds occupy the site is the
‘horseshoe’ conformation of Etravirine as compared to the largely
‘extended’ shape of GW-69564. Further differences in binding
modes include the polar interactions with the protein backbone.
Thus, Etravirine makes a hydrogen bond from the aminopyrimi-
dine NH to the backbone carbonyl of lysine K101, while GW-
69564 has a similar interaction between its amide carbonyl and
the NH of the K103 amide. Also, the core ring of each drug sits dif-
ferently in the central space so that the phenyl ‘tail’ rings enter the
back pocket at different angles. Each tail ring then effects the
important edge-to-face interaction with conserved tryptophan
W229 from a different angle, and the protein back-pocket regions
also show shifts for the residues of Y188, W229 and Y181. Lastly,
GW-69564 interacts in a front pocket region not accessible to Etra-
virine. There the anilide ring can stack with P236 and the side-
chain of V106. The 4-sulfonamide itself is positioned for possible
hydrogen bonding with P236, and also appears to possess a vector
to solvent.
X
X
OH
CN
O
Cl
a
1 X = H
N
2 X = Br
+
N
O
N
N
NC
Cl
Cl
O
Cl
H
N
b,c,d
H2N
N
CO2H
3
SO2NH2
O
H
N
R2
R3
N
R1(linker)
e
1,2
Our plan was to prepare hybrid molecules using a back pocket
substructure similar to that of Etravirine, combined with features
from the benzophenones to give the additional polar and front
pocket interactions. We conjectured this might give a drug with
similar or superior potency and mutant profile to the parents, as
well as possibly improved pharmacokinetic properties. The struc-
ture overlay in Figure 2 implies that blending the two molecules
should be synthetically feasible. That is, the geometry looks prom-
N
X
4-7
O
CN
Scheme 1. Reagents and conditions: (a) K2CO3, DMF; (b) (COCl)2, DMF; (c) TEA,
CH2Cl2, 4-aminosulfonyl-2-chloro-aniline; (d) H2NNH2, EtOH, 90 °C; (e) 3 or other
amines, DIPEA, NMP, 100 °C.
Table 1
Enzyme inhibitory potency of derivatives 4–7
X
R2
N
R1
O
N
NC
N
H
R3
KYa
#
R1
X
R2
Cl
R3
WTa
900
Y188La
>10K
H
N
4
H
SO2NH2
H
>1K
216
O
5a
H
H
87
372
N
5b
6
Br
Br
H
H
H
H
19
15
70
N
N
192
112
334
N
7
Br
H
H
67
>1K
718
Figure 2. Overlay of crystal structures of Etravirine (green) and GW-69564 (tan)
bound to wild type HIV-RT enzyme.
a
IC50 in nM, heteropolymeric enzyme inhibition assay (see Ref. 13).

D. J. Kertesz et al. / Bioorg. Med. Chem. Lett. 20 (2010) 4215–4218
4217
Further examples of N-benzyl piperidines were made in high
yields via reactions of 1 and 2 with 4-amino-N-Boc-piperidine
nitrile 5n, carboxamide 5o and carboxylic acid 5p were less potent,
although acid 5p was of note for its potential solubility advantage.
The best N-benzyl head-ring substitution patterns were similar
to that of GW-69564, which we took as early evidence that the
compounds were binding in accord with the design. However,
the presence of the basic nitrogen in the piperidine linker group
was difficult to rationalize since there was no salt bridging partner
proximal to that amine in our docked models. The question was
resolved when we obtained a crystal structure for 5h bound to
the wild type enzyme, and that structure showed a water molecule
linking the piperidine nitrogen with the backbone NH of K103
( Fig. 3).12 From this structure we surmised that the piperidine
nitrogen is not formally protonated, as it would be difficult to
understand the stabilization of a charged species in the lipophilic
binding site environment. It is then possible that some of the en-
hanced potency seen with electron withdrawing R3 groups in Table
2 results from the reduced basicity of the piperidine nitrogen, and
thence a lower cost of binding the neutral form.
Further aspects of the modeling hypothesis were confirmed
when the 5h crystal structure was overlaid with the protein bound
structures of Etravirine and GW-69564 as shown in Figure 4. Thus,
in 5h the edge-to-face interaction of the tail ring with W229 and
the hydrogen bond between the aminopyrimidine NH and the
K101 backbone carbonyl both emulate Etravirine, and these factors
may contribute to the similarity in mutant profile for the two com-
pounds. Likewise, the water-mediated H-bond of 5h with K103
effectively parallels the interaction of the GW-69564 carbonyl with
the same K103 NH, and occupancy of the lipophilic P236/V106
front pocket is similar as well. Finally, an interaction between
the 4-sulfone oxygen of 5h and V106 can be seen near the pro-
tein/solvent interface (Fig. 3). Therefore, the hybrid molecule 5h
appears to have achieved the main interactions hoped for in the
design.
followed by deprotection, affording intermediates
8 and 9
(Scheme 2). Treatment of 8 and 9 with various benzyl bromides
then gave products 5c–5p. Alternatively, reactions of 4-(N-boc-
amino)-piperidine with the benzyl bromides and deprotection
gave amine reagents 10, and further treatments with 1 and 2 then
gave the same final products.14
Antiviral efficacy data15 as well as enzyme data are presented in
Table 2 for a selection of these compounds. Enzyme data for the
chloride and nitrile-substituted N-benzyl analogs 5c–5g implied
a preference for a polar group in the 4-position. Such substituents
turned out to be vital for antiviral potency, particularly against the
KY mutant, as can be seen by comparing analogs 5h–5p with
unsubstituted 5a and 5b. Sulfonyl substituents were best in the
4-position, as among analogs 5h–5m, and addition of a 2-chloride
in combination with a 50-bromide on the core ring led to the excel-
lent antiviral profile noted for the 4-methylsulfone 5k. Corre-
sponding 4-sulfonamide 5m had a similar profile, albeit with
lower KY activity. Other combinations of substituents such as the
H
N
O
c
N
H2N
a,b
N
NH
X
N
BOC
8 X = H
9 X = Br
R2
R1
H
CN
N
N
R2
N
N
X
H2N
a
O
5c-5p
N
R1
10
CN
The best compounds from this study were the sulfone 5k and
the sulfonamide 5m. It may be noted that antiviral activities for
these compounds were more favorable than might have been pre-
dicted from their enzyme inhibitory potencies. These differences
relate to template-specific variations between HIV-RT-RNA bind-
ing affinity and polymerase processivity, and have been observed
in most NNRTI series.16 An expanded set of antiviral data for 5k
Scheme 2. Reagents and conditions: (a) 1 or 2, DIPEA, NMP, 100 °C; (b) TFA, CH2Cl2;
(c) benzyl bromides, K2CO3, DMF.
Table 2
Enzyme and antiviral potency of N-benzyl piperidines 5a–5p
H
R3
N
N
N
N
R1
R2
O
CN
Enzyme IC50 (nM)
Antiviral IC50 (nM)
Y188Lb
>100 43.6
>100 9.1
#
R1 R2
R3
WTa KYa Y188La WTb KYb
5a
5b
5c
5d
5e
5f
5g
5h
5i
H
H
H
H
H
H
H
H
H
H
H
H
87
216 372
3.3
6.5
—
Br
Br
Br
Br
Br
Br
H
19
15
229
129 793
70
—
3-Cl
4-Cl
2-CN
3-CN
4-CN
241
152
29
23
21
—
—
—
—
—
62
27
18
50
14
30
28
25
21
159
69
28
43
37
41
72
—
—
—
—
—
—
—
—
—
4-SO2CH3 34
4-SO2CH3 28
2-Cl 4-SO2CH3 22
0.3
5.2
24.6 67.2
6.1
4.4
9.9
17.6 67.4
41.5 52.7
20.2
35.7
16.9
<1.2
46.5
2.8
1.4
3.2
11.6
30.2
—
Br
H
28
5j
5k
5l
263
216
105
208
—
Br 2-Cl 4-SO2CH3 45
2-Cl 4-SO2NH2 84
5m Br 2-Cl 4-SO2NH2 90
7.3
21.4
18.9
H
5n
5o
5p
Br 2-Cl 4-CN
Br 2-Cl 4-CONH2
Br 2-Cl 4-CO2H
146
131
32
519
86
3.9
74.2
a
Enzyme inhibition, heteropolymeric assay.
Cell based antiviral assay in the presence of 10% FBS (see Ref. 15).
Figure 3. Crystal structure of N-benzyl piperidine 5h showing water bridged
H-bond interaction with K103 NH.
b

4218
D. J. Kertesz et al. / Bioorg. Med. Chem. Lett. 20 (2010) 4215–4218
cule between the piperidine nitrogen and the K103 backbone. Fur-
ther work is in progress towards improving overall antiviral profile
and protein shifted potency for derivatives using the novel piperid-
inyl aminopyrimidine substructure. Thus, exploration starting
from the secondary N-phenyl lead 6 has already yielded com-
pounds with superior antiviral properties, and that work will be
the subject of another publication.
Acknowledgements
The authors thank Eric Sjogren, Robert Wilhelm, Hans Maag and
Guozhi Tang for many helpful discussions. We also thank Yanzhou
Liu, Kate Comstock, Yongguo Li, Tao Sui and Wenzhi Chen for their
analytical expertise, and Yu Li, Eugene Chow and Jun Mei Yan for
their bioassay diligence.
References and notes
1. Flexner, C. Nat. Rev. Drug Discovery 2007, 6, 959.
2. Balzarini, J. Curr. Top. Med. Chem. 2004, 4, 921.
3. Pauwels, R. Curr. Opin. Pharm. 2004, 4, 437.
4. Adkins, J. C.; Noble, S. Drugs 1998, 56, 1055.
5. Sweeney, Z. K.; Klumpp, K. Curr. Opin. Drug Discovery Dev. 2008, 11, 458.
6. D’Aquila, R. T.; Schapiro, J. M.; Brun-Vezinet, F.; Clotet, B.; Conway, B.; Demeter,
L. M.; Grant, R. M.; Johnson, V. A.; Kuritzkes, D. R.; Loveday, C.; Shafer, R. W.;
Richman, D. D. Top. HIV Med. 2003, 11, 92.
Figure 4. Overlay of crystal structures of 5h (yellow), Etravirine (green) and
GW-69564 (tan) in wild type HIV-RT enzyme.
7. Antinori, A.; Zaccarelli, M.; Cingolani, A.; Forbici, F.; Rizzo, M. G.; Trotta, M. P.;
Di Giambenedetto, S.; Narciso, P.; Ammassari, A.; Girardi, E.; De Luca, A.; Perno,
C. F. Aids Res. Hum. Ret. 2002, 18, 835.
8. Madruga, J. V.; Cahn, P.; Grinsztejn, B.; Haubrich, R.; Lalezari, J.; Mills, A.;
Pialoux, G.; Wilkin, T.; Peeters, M.; Vingerhoets, J.; de Smedt, G.; Leopold, L.;
Trefiglio, R.; Woodfall, B. Lancet 2007, 370, 29.
9. Schöller-Gyüre, M.; Kakuda, T. N.; De Smedt, G.; Vanaken, H.; Bouche, M.-P.;
Peeters, M.; Woodfall, B.; Hoetelmans, R. M. W. Br. J. Clin. Pharmacol. 2008, 66,
508.
Table 3
Antiviral potencies of 5k and 5m compared to reference NNRTIsa
WT,
10%
FBS
WT,
40%
K103N/ K103N/ Y188L G190A V106A
Y181C
L100I
HuSb
10. Ren, J.; Chamberlain, P. P.; Stamp, A.; Short, S. A.; Weaver, K. L.; Romines, K. R.;
Hazen, R.; Freeman, A.; Ferris, R. G.; Andrews, C. W.; Boone, L.; Chan, J. H.;
Stammers, D. K. J. Med. Chem. 2008, 51, 5000.
11. Sahlberg, C.; Zhou, X. Anti-Infect. Agents Med. Chem. 2008, 7, 101.
12. Atomic coordinates for the structures of Etravirine and 5h bound to HIV/RT
have been deposited with the RCSB Protein Data Bank (PDB) under the access
codes 3M8P, and 3M8Q, respectively.
Efavirenz
5k
5m
1.8
6.1
9.9
18.1
48.7
61.5
25.9
82.8
7.3
18.9
9.2
>500
5.9
9.1
>1K
2.8
3.2
3.1
6.0
1.8
6.0
1.1
5.7
2.0
2.3
2.0
Etravirine 2.1
9.5
a
Antiviral IC50’s are given in nM.
b
WT activity in the presence of 40% HuS is an indicator of protein shifted
13. Enzymatic assay: IC50 values were obtained from a SPA heteropolymeric assay.
RNA-dependent DNA polymerase activity was measured using a biotinylated
primer oligonucleotide and tritiated dNTP substrate. Newly synthesized DNA
was quantified by capturing the biotinylated primer molecules on streptavidin
coated Scintillation Proximity Assay (SPA) beads (Amersham).
14. For a detailed description of the syntheses of aminopyrimidine inhibitors see:
Kertesz, D. J.; Brotherton-Pleiss, C.; Yang, M. US 2008/0146595.
15. Replication assay: EC50 values were obtained following incubation of compound
with infected MT4 cells in the presence of 10% fetal bovine serum (FBS). The
assay assesses the reduction in the cytopathic effect of HIV-1 on MT4 cells in
the presence of HIV-RT inhibitors. Each value represents the average of at least
two independent assays. For a detailed description of the anitiviral assay see:
Pauwels, R.; De, C. E.; Desmyter, J.; Balzarini, J.; Goubau, P.; Herdewijn, P.;
Vanderhaeghe, H.; Vandeputte, M. J. Virol. Methods 1987, 16, 171.
16. Carroll, S. S.; Olsen, D. B.; Bennett, C. D.; Gotlib, L.; Graham, D. J.; Condra, J. H.;
Stern, A. M.; Shafer, J. A; Kuo, L. C. J. Biol. Chem. 1993, 268, 276.
17. Determination of potency versus wild type virus was repeated in the presence
of 40% human serum (HS) for these compounds. The resulting ‘protein shifted
potency’ as compared to the result with 10% FBS (see Ref. 15) is considered an
indicator of potential for in vivo activity.
potency.17
and 5m is given in Table 3 and includes values for protein shifted
WT potency,17 antiviral activities versus further mutants of inter-
est, and comparative data for selected reference NNRTIs. Analog
5k was quite potent against all mutants tested, and as such dis-
played a highly promising profile for a next-generation NNRTI.
5m was less active but the sulfonamide analogs remain of interest
for their potential as prodrug substrates.
In summary, we have discovered a novel series of N-benzyl
piperidine compounds which combine features of both K101 and
K103 main-chain binding types of NNRTIs. The best compound
5k has good potency against wild type RT and several important
mutant strains in both enzyme and antiviral assays. The series
has a unique binding mode which utilizes a bridging water mole-