Combining symmetry elements results in potent naphthyridinone (NTD) HIV-1integrase inhibitors

Article abstract of DOI:10.1016/j.bmcl.2011.08.082

A series of naphthyridinone HIV-1 integrase strand-transfer inhibitors have been designed based on a psdeudo-C2 symmetry element present in the two-metal chelation pharmacophore. A combination of two distinct inhibitor binding modes resulted in potent inhibition of the integrase strand-transfer reaction in the low nM range. Effects of aryl and N1 substitutions are disclosed including the impact on protein binding adjusted antiviral activity.

Full text of DOI:10.1016/j.bmcl.2011.08.082

Bioorganic & Medicinal Chemistry Letters 21 (2011) 6461–6464
Contents lists available at SciVerse ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
Combining symmetry elements results in potent naphthyridinone (NTD) HIV-1
integrase inhibitors
Brian A. Johns a, , Takashi Kawasuji b, Jason G. Weatherhead a, Eric E. Boros a, James B. Thompson a,
⇑
Edward P. Garvey a, , Scott A. Foster a, , Jerry L. Jeffrey a, Wayne H. Miller a, Noriyuki Kurose b,
Kenichi Matsumura b, Tamio Fujiwara b
a GlaxoSmithKline Research & Development, Infectious Diseases Therapeutic Area Unit, Five Moore Drive, Research Triangle Park, NC 27709, USA
b Shionogi Research Laboratories, Shionogi & Co., Ltd, 12-4 Sagisu 5-chome Fukushima-ku, Osaka 553-0002, Japan
a r t i c l e i n f o
a b s t r a c t
Article history:
Received 7 July 2011
Revised 15 August 2011
Accepted 17 August 2011
Available online 8 September 2011
A series of naphthyridinone HIV-1 integrase strand-transfer inhibitors have been designed based on a
psdeudo-C2 symmetry element present in the two-metal chelation pharmacophore. A combination of
two distinct inhibitor binding modes resulted in potent inhibition of the integrase strand-transfer reac-
tion in the low nM range. Effects of aryl and N1 substitutions are disclosed including the impact on pro-
tein binding adjusted antiviral activity.
Ó 2011 Published by Elsevier Ltd.
Keywords:
HIV Integrase
Naphthyridinone
Antiviral
HAART
Inhibion of HIV-1 integrase has become an important compo-
nent of highly active antiretroviral therapy (HAART) regimens
since the approval of raltegravir in late 2007.1 Although this new
agent possesses many excellent attributes, raltegravir is suscepti-
ble, as is many antivirals, to resistance. At least three distinct sig-
nature mutation pathways (N155H, Q148H/R/K with E138K or
G140S/A and Y143R/C) are associated with virological failures
and loss of efficacy.2,3 There are a number of reasons for resistance,
among which is a lack of patient compliance with dosing regimens.
Raltegravir requires a twice daily dose of 400 mg which, while
effective and safe, is inconvenient and ultimately leads to missed
doses and suboptimal drug levels which may lead to a more rapid
onset of resistance.4 As a result, there continues to be a need to
pursue additional integrase inhibitors with superior resistance
profiles and more convenient dosing regimens.
(a)
(b)
OH
O
O
O
O
OH
N
R
N
N
H
N
H
3
N
1
N
H
7
H
NTD type A
NTD type B
O
O
OH
O
N
H
N
N
Raltegravir
O
O
OH
F
N
O
N
H
N
H
N
N
H
N
⇑
Corresponding author.
E-mail address: brian.a.johns@gsk.com (B.A. Johns).
NTD type A/B
Corresponding address at present address: Viamet Pharmaceuticals, Inc., Morr-
Figure 1. (a) Two-metal pharmacophore model. (b) Pseudo-symmetric naphthy-
isville, NC (E.P.G.), Chimerix, Inc., Durham, NC (S.A.F.).
ridinone (NTD) integrase inhibitors NTD type A, B and A/B.
0960-894X/$ - see front matter Ó 2011 Published by Elsevier Ltd.
doi:10.1016/j.bmcl.2011.08.082

6462
B. A. Johns et al. / Bioorg. Med. Chem. Lett. 21 (2011) 6461–6464
Table 2
NTD type A/B 3-amido SAR
Raltegravir exacts its inhibitory activity on the second biochem-
ical step catalyzed by the integrase enzyme whereby the 30 pro-
cessed viral DNA is joined with the host chromosomal DNA in a
process called strand-transfer.5 Several classes of integrase
strand-transfer inhibitors have been reported over the past few
years. They all appear to rely on binding to the highly conserved
catalytic core of the integrase enzyme which consists of two
Mg2+ ions held in place by a DD(35)E triad of protein carboxylate
side chains (asp116, asp64 and glu152) typical of the endonuclease
class of enzymes.6 This two-metal pharmacophore is shown in
graphical format in Figure 1. Structurally, inhibitors typically con-
sist of a planar metal chelating motif. Integral to this model is a
central phenolate-like moiety that coordinates both electropositive
Mg2+ ions. In addition, a pendant aromatic group is also required
for potent activity. The central metal-binding core has an interest-
ing pseudo-symmetry about the phenolate group due to the double
chelation requirements (Fig. 1a). This symmetry element was an
area that we desired to explore during our investigations of a series
of naphthyridinone (NTD) integrase inhibitors that have recently
been reported from our laboratories.7 As shown in Figure 1b, the
NTD carboxamide core can in principle coordinate the active site
metals in either the type A or type B arrangement which differ in
how the requisite aromatic tail is tethered. In the type A arrange-
ment, the 3-carboxamide serves to position the benzyl group while
in the type B structures a 7-benzyl substitution places the flexible
aromatic tail in a similar orientation when the metal binding motif
is flipped 180°. Our objective in the current study was to examine
what effects a combination of the two binding modes would have
on the structure activity relationship (SAR) for the series.
OH
O
O
R3
X
N
N
H
N
H
Compd
X
R3
(
lM)
T.I.
IC50
EC50
PAEC50
0.43
4
5
6
7
H
F
0.013
0.019
711
F
F
F
F
0.006
0.035
0.032
0.008
0.12
0.28
ND
1728
117
90
H
H
Cl
0.039
0.48
F3C
8
F
F
F
0.007
0.010
0.006
0.007
0.013
0.009
0.44
0.46
0.33
1728
1053
1538
OMe
OMe
9
10
As a starting point we sought to determine the baseline activity
for the type A scaffold. The N1 protio example (1) was highly po-
tent in the integrase strand transfer (INST) biochemical assay
(IC50 = 13 nM) and also showed good activity (EC50 = 110 nM) in a
HIV pseudo-typed single round antiviral assay suggesting the
NTD core was efficiently binding the integrase active site as de-
signed (Table 1). Two additional examples of the type A scaffold
were also constructed that contained N1 alkyl substituents. The
isopropyl (2) and cyclopentyl (3) analogs were also potent inhibi-
MeO
OMe
OMe
11
12
F
0.008
0.020
0.021
0.065
1.10
0.36
680
79
H
tors of the strand-transfer event and showed sub-lM antiviral
activity. The cyclopentyl analog was examined using therapeuti-
cally relevant levels of added human serum albumin to obtain a
protein-adjusted EC50 (PAEC50). The activity was shifted by 14 fold
in the presence of HSA and resulted in a value 1.5 lM. All of the
analogs in the series had a reasonable therapeutic index (TI) clearly
our study, a literature report appeared exploiting similar structures
with consistent data to our results.8
While the results showed sufficient potency with a very limited
number of examples, the publication of similar compounds tem-
pered our interest at an early stage. Furthermore, the lack of read-
ily accessible sites on the NTD type A chemotype limited the
optimization of the potency and improvement of the drug-like
properties. At the same time, other efforts within our laboratories
were examining the NTD type B motif which were showing im-
proved potency and had additional positions that were easily ame-
nable for SAR examinations. As such, we were interested to
determine if combining both type A and type B pendant aryl
groups would provide improved potency.
The parent 7-phenylmethyl and the corresponding (4-fluoro)-7-
phenylmethyl counterpart were examined during this SAR combi-
nation study. Table 2 shows a variety of C3 amide modifications
while maintaining the N1 position unsubstituted. The NTD type
A/B combination analogs 4 and 5 which differ by the presence of
a fluoro group on the 7-benzyl moiety showed approximately
one order of magnitude improvement in antiviral potency when
compared to the type A analog 1. Addition of the 4-fluoro substitu-
ent to the 7-benzyl analog resulted in a two-fold increase in po-
tency. Further modifications of the C3 amide showed a 4-fold
decrease in activity for 2,4-disubstituted electron deficient analog
(7). Alternatively, the electron rich methoxy ether derivatives
(8–10) retained a similar potency as the initial 4-fluoro benzyl
separating antiviral activity from cytotoxicity. During the course of
Table 1
NTD type A activity
O
O
OH
N
N
H
F
N
R
Compd
R
(
lM)
T.I.13
10
11
12
IC50
EC50
0.11
0.40
PAEC50q
1
2
0.013
0.001
ND
87
30
H
ND
3
0.006
0.11
1.54
79

B. A. Johns et al. / Bioorg. Med. Chem. Lett. 21 (2011) 6461–6464
6463
Table 3
O
O
OEt
O
NTD type A/B N1 substituent SAR
N
N
Cl
OEt
NaOEt
EtOH
O
OEt
R1
OH
O
O
N
NH
R1
X
N
DCE, Δ
N
H
O
N
F
EtO
O
R1
X
1
X
24a
24b
R
R
= -H
23a R1 = -H
R1CHO,
BH3•DMS
AcOH, DCM
Compd
X
R1
(
lM)
T.I.
1 = -alkyl
IC50
EC50
PAEC50
0.53
23b
R
1 = -alkyl
OH
R2
13
14
F
0.017
0.018
0.008
0.070
1707
99
CH3
O
O
H2N
OEt
OH
O
O
R2
N
N
H
0.64
N
H
110 o
neat
C
N
N
R1
15
16
H
H
0.034
0.028
0.030
0.012
0.73
0.14
465
R1
4-19
1 = -alkyl
= -H
25b
25a
R
R
X
1129
X
1
NH2
LHMDS
MeI, DMF
1 = -Me
25c
R
17
H
0.028
0.009
0.23
1427
Scheme 1. Synthesis of compounds 4–19.
SO2Me
O
18
19
H
F
0.032
0.014
0.008
0.003
0.29
0.14
1772
5600
O
OEt
N
N
X
N
OEt 1) EtOH,
glyoxal
O
O
R
NH2
N
N
N
H
2) NaBH3CN
O
26
R = -OH
23
20
21
F
F
0.004
0.019
0.010
0.026
0.13
0.22
1149
171
HNR3R4
EDCI, DMF
27 R = -NR3R4
O
X
OH
N
O
O
N
N
X
N
H
R2
O
O
N
O
NR3R4
22
F
0.008
0.003
0.045
2875
20-22
O
Scheme 2. Synthesis of N1 acetamide derivatives 20–22.
carboxamide 5 in both the enzyme and antiviral assays. The highly
electron rich example 11 consisting of a 3,4-dimethoxy benzyl
moiety had similar enzyme potency but fell off approximately
two to three fold in the whole cell assay. Finally, some loss in po-
N1 benzylic group was each tested with both having approxi-
mately 10 nM antiviral activity with 12- and 25-fold shifts respec-
tively upon the addition of HSA. Finally, a set of polar amido and
reverse-amido substituents were constructed consisting of either
a pyrrolidinonyl ethyl group (18 and 19) or an acetamide substitu-
ent (20–22). The pyrrolidinone analogs showed low nM antiviral
activity but had higher fold shifts in the presence of HSA. The
N,N-dimethyl amide displayed slightly better potency than the
more lipophilic piperidine derived amide 21, while both shifted
near 10-fold when albumin was added. The morpholinyl amide
was the most potent analog having an antiviral EC50 of 3 nM and
a protein-adjusted EC50 of 45 nM.
tency was observed for the
a-methyl analog 12 which was more
pronounced in the antiviral assay. Some differences in the effects
of protein binding were noticed but most of the analogs main-
tained a potency of 300–500 nM with added HSA.
Next we sought to survey the effects of N1 substitution on the
activity of the A/B combination analogs. To this end, we began with
simple N1 alkyl groups (13–15) which had little effect on intrinsic
potency but did cause a slightly increased shift when HSA was
added (Table 3). An electron rich (16) and electron deficient (17)
In order to confirm activity in a fully replicative competent as-
say, several compounds were tested in an MT4 cell antiviral assay
(Table 4).9 Similar levels of activity were observed as well as robust
therapeutic indexes in the examples tested. Additionally, com-
pound 19 was assayed in a 7-day PBMC assay resulting in an
EC50 of 11 nM.
Table 4
MT4 antiviral activity
Compd
EC50
(lM)
T.I.
108
453
>400
347
4
0.037
0.066
0.020
0.015
13
19
22
The new chemotype was made beginning with a previously de-
scribed7 4-benzyl-3-amino picolinate ester 23a. This material

6464
B. A. Johns et al. / Bioorg. Med. Chem. Lett. 21 (2011) 6461–6464
could be alkylated via a reductive amination or used as is for N1
protio analogs (Scheme 1). The aniline 23 was acylated with ethyl
chloromalonate under base free conditions in refluxing 1,2-DCE.
This material then was subjected to Dieckmann cyclization
conditions to provide the naphthyridinone core 25. In the case of
the N1-methyl analog 13, a late stage alkylation was performed
with methyl iodide with strong base to provide the N1-methyl
NTD core 25c. The C3 ester present in the various intermediate
25 structures was converted to the desired benzyl amides by heat-
ing with the neat benzylamines to provide compounds 4–19.
Alternatively as shown in Scheme 2, the reductive amination to
install the N1 acetamide moiety employed glyoxal to arrive at the
precursor acid 26 which was coupled with simple amines and car-
ried through to the target analogs 20–22 as illustrated in Scheme 1
(e.g., 25a ? 4).
Overall, the type A/B combination naphthyridinone analogs ap-
pear to show significant improvement in antiviral activity when
compared to the C3 benzyl carboxamide type A series while having
similar levels of enzymatic activity in a cell-free biochemical assay.
The lack of dramatic SAR outlined with the C3 amide modification
may suggest that the type B binding mode may be dominant when
the two pharmacophores are fused into a single motif, however
this cannot be unambiguously concluded from the data herein.
However, it is clear that this combination is able to deliver potent
activity which is largely maintained in the presence of human ser-
um albumin. It also appears to be likely that the N1 substituent is
not only tolerated in this series but can contribute to either direct
antiviral potency or to diminishing the shift when challenged with
serum proteins.
References and notes
1. Taiwo, B.; Murphy, R. L.; Katlama, C. Drugs 2010, 70, 1629.
2. Fransen, S.; Gupta, S.; Danovich, R.; Hazuda, D.; Miller, M.; Witmer, M.;
Petropoulos, C. J.; Huang, W. J. Virol. 2009, 83, 11440.
3. Hazuda, D. J. Braz. J. Infect. Dis. 2010, 14, 513.
4. Bangsberg, D. R.; Ragland, K.; Monk, A.; Deeks, S. G. AIDS 2010, 24, 2835.
5. Hazuda, D.; Iwamoto, M.; Wenning, L. Annu. Rev. Pharmacol. Toxicol. 2009, 49,
377.
6. Johns, B. A.; Svolto, A. C. Expert. Opin. Ther. Patents 2008, 18, 1225.
7. Boros, E. E.; Edwards, C. E.; Foster, S. A.; Fuji, M.; Fujiwara, T.; Garvey, E. P.;
Golden, P. L.; Hazen, R. J.; Jeffrey, J. L.; Johns, B. A.; Kawasuji, T.; Kiyama, R.;
Koble, C. S.; Kurose, N.; Miller, W. H.; Mote, A. L.; Murai, H.; Sato, A.; Thompson,
J. B.; Woodward, M. C.; Yoshinaga, T. J. Med. Chem. 2009, 52, 2754.
8. Egbertson, M.; Melamed, J. Y.; Langford, H. M.; Young, S. D. WO 03/062204.
9. MT4 and PBL assays were performed according to: Kobayashi, M.; Yoshinago,
T.; Seki, T.; Wakasa-Morimoto, C.; Brown, K. W.; Ferris, R.; Foster, S. A.; Hazen,
R. J.; Miki, S.; Suyama-Kagitani, A.; Kawauchi-Miki, S.; Taishi, T.; Kawasuji, T.;
Johns, B. A.; Underwood, M. R.; Garvey, E. P.; Sato, A.; Fujiwara, T. Antimicrob.
Agents Chemother. 2011, 55, 813.
10. Recombinant HIV-1 integrase strand transfer assay: Boros, E. E.; Johns, B. A.;
Garvey, E. P.; Koble, C. S.; Miller, W. H. Bioorg. Med. Chem. Lett. 2006, 16, 5668.
11. Pseudo-type HIV assay (PHIV): Garvey, E. P.; Johns, B. A.; Gartland, M. J.; Foster,
S. A.; Miller, W. H.; Ferris, R. G.; Hazen, R. J.; Underwood, M. R.; Boros, E. E.;
Thompson, J. B.; Weatherhead, J. G.; Koble, C. S.; Allen, S. H.; Schaller, L. T.;
Sherrill, R. G.; Yoshinaga, T.; Kobayashi, M.; Wakasa-Morimoto, C.; Miki, S.;
Nakahara, K.; Noshi, T.; Sato, A.; Fujiwara, T. Antimicrob. Agents Chemother.
2008, 52, 901.
12. PAEC50 is determined using 40 mg/mL of purified human serum albumin in the
pseudo-type HIV assay described in the Ref. 10
13. Therapeutic index (CC50/EC50).