Optimization of a 1,5-dihydrobenzo[b][1,4]diazepine-2,4-dione series of HIV capsid assembly inhibitors 2: Structure-activity relationships (SAR) of the C3-phenyl moiety

Article abstract of DOI:10.1016/j.bmcl.2013.03.074

Detailed structure-activity relationships of the C3-phenyl moiety that allow for the optimization of antiviral potency of a series of 1,5-dihydrobenzo[b][1,4]diazepine-2,4-dione inhibitors of HIV capsid (CA) assembly are described. Combination of favorable substitutions gave additive SAR and allowed for the identification of the most potent compound in the series, analog 27. Productive SAR also transferred to the benzotriazepine and spirobenzodiazepine scaffolds, providing a solution to the labile stereocenter at the C3 position. The molecular basis of how compound 27 inhibits mature CA assembly is rationalized using high-resolution structural information. Our understanding of how compound 27 may inhibit immature Gag assembly is also discussed.

Full text of DOI:10.1016/j.bmcl.2013.03.074

Bioorganic & Medicinal Chemistry Letters 23 (2013) 3401–3405
Contents lists available at SciVerse ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
Optimization of a 1,5-dihydrobenzo[b][1,4]diazepine-2,4-dione series
of HIV capsid assembly inhibitors 2: Structure–activity relationships
(SAR) of the C3-phenyl moiety
,
⇑
Lee D. Fader , Serge Landry, Sylvie Goulet, Sébastien Morin, Stephen H. Kawai à, Yves Bousquet,
Isabelle Dion, Oliver Hucke, Nathalie Goudreau, Christopher T. Lemke, Jean Rancourt, Pierre Bonneau,
Steve Titolo, Ma’an Amad, Michel Garneau, Jianmin Duan, Stephen Mason §, Bruno Simoneau
Boehringer Ingelheim (Canada) Ltd, Research and Development, 2100 Cunard Street, Laval, Québec, Canada H7S 2G5
a r t i c l e i n f o
a b s t r a c t
Article history:
Available online 3 April 2013
Detailed structure–activity relationships of the C3-phenyl moiety that allow for the optimization of anti-
viral potency of a series of 1,5-dihydrobenzo[b][1,4]diazepine-2,4-dione inhibitors of HIV capsid (CA)
assembly are described. Combination of favorable substitutions gave additive SAR and allowed for the
identification of the most potent compound in the series, analog 27. Productive SAR also transferred to
the benzotriazepine and spirobenzodiazepine scaffolds, providing a solution to the labile stereocenter
at the C3 position. The molecular basis of how compound 27 inhibits mature CA assembly is rationalized
using high-resolution structural information. Our understanding of how compound 27 may inhibit
immature Gag assembly is also discussed.
Keywords:
HIV
Capsid Assembly Inhibitor
Immature Gag assembly Inhibitor
Benzodiazepine
Ó 2013 Elsevier Ltd. All rights reserved.
The ever present need for new agents having distinct resistance
profiles for treating HIV infected patients has served as the back-
drop for discovery of drugs that intervene at clinically novel points
in the HIV replication cycle.1,2 Toward this goal, the preceding
companion Letter documented our effort to optimize a series of
1,5-dihydrobenzo[b][1,4]diazepine-2,4-dione (BD) inhibitors of
HIV capsid (CA) assembly, exemplified by compound 1.3 This study
dealt with removal of a liability inherent to the scaffold through a
rational, structure-guided approach. Herein we detail structure–
activity relationships (SAR) of the C3-phenyl moiety developed in
the original 1,5-dihydrobenzo[b][1,4]diazepine-2,4-dione series
and the transposition of this knowledge to the benzotriazepine
(BT) and spirobenzodiazepine (SBD) scaffolds described in the pre-
ceding account.
O
O
N5
N1
CF3
3
1
The original synthetic route to analogs of compound 1 proved to
be well suited to introduce diversity at the C3 position of the BD
core (Scheme 1).4 In one approach, deprotonation of phenylacetic
acid 2 with NaHMDS and quenching of the resulting enolate with
NCCO2Et provided malonate derivative 3. This intermediate was
then used in the direct acylation of the previously described phen-
ylenediamine derivative 4 and cyclization to the 7-membered ring
analog 5 was then promoted by addition of Li- or NaHMDS. In
many cases, arylation of the N5 position with Ph3Bi mediated by
Cu(OAc)2 and base yielded final compounds. In other cases, depro-
tection or functional group interconversion was necessary and
these details have been described elsewhere for compounds.5 It
was also found that the previously described benzodiazepine 64
was a suitable coupling partner in the Pd-catalyzed C-arylation
reaction with aryl iodides and bromides, which allowed for parallel
synthesis and late stage introduction of diversity.6
⇑
Corresponding author. Tel.: +1 203 791 6766.
E-mail address: lee.fader@boehringer-ingelheim.com (L.D. Fader).
Present address: Boehringer Ingelheim Pharmaceuticals, Inc., 900 Ridgebury
Road, Ridgefield, CT 06877, United States.
à
Present address: Concordia University, Deptartment of Chemistry and Biochem-
To assess the ability of our compounds to inhibit assembly of
CA, compounds were tested in either our capsid assembly assay
(CAA),4,5,7,8 capsid disassembly assay (CDA)5,8 or both, since
istry, 7141 Sherbrooke Street West, Montreal, Québec, Canada H4B 1R6.
§
Present address: Bristol–Myers Squibb, Virology, 5 Research Parkway, Walling-
ford, CT 06492, United States.
0960-894X/$ - see front matter Ó 2013 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.bmcl.2013.03.074

3402
L. D. Fader et al. / Bioorg. Med. Chem. Lett. 23 (2013) 3401–3405
R2
R4
in antiviral potency. This trend held for a number of different aryl
H
N
O R1
R2
R4
O R1
NH2
NH
CF3
CF3
b
groups and so the diversity within the meta- and para-biaryl motifs
was expanded in parallel. In both cases, a number of variations
were identified that led to similar levels of antiviral potency rela-
tive to the prototypical compounds 15 and 16. For example, in
the meta-series the 3-pyridinyl, 1-methyl-4-pyrazolyl and 1-
methyl-5-pyrazolyl groups all served as suitable replacements for
the 4-pyridinyl group. Similarly, we discovered that the 4-oxazolyl
(20), 3-pyridinyl (21) and the 2-methoxy-4-pyrimidinyl (22) ana-
logs showed antiviral potency similar to the 4-pyridinyl analog
(16).
EtO
R3
R3
+
N
X
O
2
3
X = H
4
5
a
c and d (if needed)
R1
R2
X = CO2Et
R3
X
R4
R1
R2
R4
At this point in our study we had shown that simple substitu-
ents on the 3-phenyl substituent and inclusion of a biaryl motif
each independently led to modest potency improvements. We next
investigated whether or not we could combine these two ap-
proaches in an effort to further improve antiviral potency. Among
numerous permutations attempted, the most productive effort to
optimize this series of inhibitors is outlined in Table 3. Here, the
meta-attachment of the 1-methyl-4-pyrazolyl modification (18)
was selected for study. When we added the hydroxyl substitution
at the para- and ortho-positions to give compounds 23 and 24,
respectively, these modifications led to significant improvements
in antiviral potency, with EC50 values below 200 nM for the first
time in this series. During this phase of the study we measured a
number of in vitro ADME properties and calculated various drug-
like properties. Compounds 23 and 24 displayed very similar pro-
files, with the exception of very poor caco-2 permeability for com-
pound 24. We noted that the calculated logP narrowly obeyed
Lipinski’s rule of five10 and hypothesized that combining two hy-
droxyl groups with the biaryl motif may serve the dual purpose
of decreasing AlogP11 while improving antiviral potency. Gratify-
ingly, compound 25 was more potent in the viral replication assay,
had a lower AlogP value and showed a promising in vitro ADME
profile, including improved caco-2 permeability relative to com-
pound 24. Encouraged by this outcome, we explored combinations
of the 2,4-dihydroxy substitution pattern with other biaryl frame-
works and observed similar levels of potency. In the case of unsub-
stituted pyrazole 26, AlogP was lower but again poor caco-2
permeability was observed. Finally, addition of a number of alter-
native third substituents to compound 24 followed by a limited
exploration of the pyrazole ring substituents eventually led us to
analog 27, which was about eightfold more potent in the viral rep-
lication assay than the related disubstituted analog 13 and, in some
respects, showed a promising in vitro ADME profile (see Table 4).
Compounds such as pyrazole 27 possess a chiral center that we
have shown is unstable at slightly elevated pH. The preceding Let-
ter documented our discovery of scaffold replacements designed to
address this problem. With these promising solutions in hand, we
7
O
O
O
O
N
N
N
N
CF3
CF3
R3
e
X = Br, I
6
8-17
Scheme 1. Reagents and conditions: (a) NC–CO2Et, NaHMDS; (b) compounds 3 + 4,
160 °C, 25 min; (c) Ph3Bi, Cu(OAc)2, Pyr, DCM; (d) functional group
interconversion; see Ref. 5; (e) compounds 6 + 7, Pd[P(t-Bu)3]2, Cs2CO3, THF, 80 °C.
lW @
compounds that typically inhibit assembly of CA also promote its
disassembly. Ultimately, SAR was followed using a viral replication
assay, which allowed comparison of all compounds. The explora-
tion of the SAR began with mono-substitution of the 3-phenyl moi-
ety (Table 1). For these studies, compounds 1 and 8 served as
control compounds where the N1 substituent was methyl or ethyl,
respectively.9 Positional scans revealed productive SAR and that a
range of substitutions where tolerated. The case of a methoxy sub-
stituent is illustrative and shows that while the ortho- and para-
methoxy analogs (9 and 11, respectively) are equivalent to control
compound 1 in terms of intrinsic (CAA) and antiviral potency
(effective concentration corresponding to 50% inhibition of viral
replication, EC50), placement of a methoxy group at the meta-posi-
tion led to a modest improvement in intrinsic and antiviral potency
(two and threefold, respectively). A range of substituents were
tested at each position and clear SAR was observed. Among those
evaluated, a para-hydroxyl substituent proved to robustly improve
antiviral potency. It was also revealed that combining substituents
often proved to be additive for improving antiviral potency, as
illustrated by the comparison of compound 12 and compound 13.
As our investigation progressed, we began to explore aryl sub-
stitution on the 3-phenyl ring and a summary of this effort is pre-
sented in Table 2. A positional scan revealed that ortho-arylation of
compound 8 to give compound 14 led to a complete loss of activity
against capsid assembly, while meta- and para-arylation to give
compounds 15 and 16, respectively, led to modest improvements
Table 1
Effect of simple substitution of the C3-Ph
R1
R2
O
O
N
CF3
R3
N
X
X
R1
R2
R3
CAA IC50
(lM)
CDA IC50
(lM)
EC50
(
lM)
CC50 (lM)
1
8
9
10
11
12
13
Me
Et
Me
Me
Me
Et
H
H
OMe
H
H
H
H
H
OMe
H
H
H
H
H
OMe
OH
1.2
0.68
0.48
—
—
—
3.0
2.6
2.7
1.2
3.2
1.0
0.53
—
—
—
—
—
—
42
0.73
0.71
0.55
1.1
0.40
0.59
H
H
H
OMe
0.18
0.11
Et
OH

L. D. Fader et al. / Bioorg. Med. Chem. Lett. 23 (2013) 3401–3405
3403
Table 2
Exploration of biaryl substitutions
R1
R2
O
O
N
N
CF3
R3
R1
R2
H
R3
H
CAA IC50
>7.2
(l
M)
CDA IC50
—
(lM)
EC50
—
(lM)
CC50
—
(lM)
14
N
15
16
17
18
H
H
H
H
H
0.53
0.39
0.48
0.63
0.19
0.21
—
0.79
0.57
1.0
17
15
20
20
N
H
N
N
N
H
H
0.13
0.71
N
N
19
H
H
0.67
0.30
0.99
30
N
O
20
21
22
H
H
H
H
H
H
0.76
0.40
0.45
0.47
0.32
0.27
1.1
>7.0
7.0
N
N
N
0.40
0.62
>20
N
OMe
Table 3
Substitution of the m-biaryl
R1
O
R2
R4
N
N
CF3
R3
O
R1
R2
R3
R4
CDA IC50
M)
EC50
M)
AlogP Caco-2 permeability
(Â10À6 cm/s)
HLM/RLM metabolism (t1/2
min)
,
CYP inh. 3A4/2D6
(lM)
(
l
(
l
N
N
23 OH
H
H
H
H
H
0.17
0.12
0.10
0.05
0.16
0.18
0.19
0.14
0.13
4.8
4.8
4.5
4.3
5.6
4.3
150/—
120/—
111/11
>300/15
86/67
1.7/>30
5.7/>30
4.5/13
N
24
H
OH
OH
OH
<0.1
7.7
N
N
N
25 OH
26 OH
NH
<0.1
9.9
4.8/8.9
3.6/>30
N
N
27
H
OMe OH
0.070
N
transposed some of the knowledge from the current study onto
these alternative scaffolds. For example, the triazepine (TP) analog
28 showed nearly identical IC50 and EC50 values to its benzodiaze-
pine counterpart, compound 24. In the case of the spiro-
benzodiazepine (SBD) scaffold, the meta-positions of the 3-
phenyl group are non-equivalent and so both compounds 29 and
30 were made and compared to compound 24. Compound 29
showed antiviral potency within 2-fold of compound 24, while
compound 30 was fourfold less potent. This is consistent with all
available X-ray crystallographic data for biaryl analogs related to
compound 24, such as compound 31, which binds exclusively in
a conformation that orients the pyridine ring toward the base of
helix 7 and away from helix 1 (Fig. 1). Regioisomer 29 was de-
signed to mimic this bound conformation, while compound 30
would be expected to mimic a conformation that orients the pyri-
dine ring toward helix 1. Indeed, compound 30 showed similar po-
tency to the corresponding analog that lacks the pyrazol ring,3
indicating there is no benefit to antiviral potency when adding
the heterocycle at this position. It also became clear that the ben-
efit realized through transition to the BT or SBD scaffolds was offset

3404
L. D. Fader et al. / Bioorg. Med. Chem. Lett. 23 (2013) 3401–3405
Table 4
Transposition of this knowledge to the benzotriazepine (BT) and spirobenzodiazepine
(SBD) scaffolds
O
N
N
CF3
X
O
–X–
CDA IC50
(lM)
EC50
0.24
(l
M)
AlogP
N
N
N
28
29
30
0.12
0.13
0.20
5.5
N
OH
N
Figure 2. (A) ribbon representation of
a single CA monomer from of the CA
hexamer (PDB ID 4MGE13). The bound position of compound 31 (superposed from
PDB ID 4E917) is shown in spheres (blue = nitrogen, red = oxygen and cyan = car-
bon), and helices 1 and 2, which have been shown to shift upon binding of
compound 31, have been highlighted orange. (A) semitransparent surface of the CA
monomer is overlaid, with the surface corresponding to helices 1 and 2 again
highlighted orange; (B) Hexameric CA generated from the monomer described in A.
Alternating CA monomers has been darkened in order to better distinguish
individual CA molecules, and thereby better define the intermolecular CA–CA
interfaces.
0.30
0.78
5.3
5.3
OH
OH
N
N
of this series was discontinued due to the lack of a path forward
that did not involve a compromise in compound quality.12
We have previously shown that compound 27 exerts its anti-
viral effect by (a) decreasing virion production, presumably
through inhibition of Gag assembly, and (b) causing an imma-
ture defect in the virions that are produced, presumably through
inhibition of capsid assembly.7 In this study we also reported
that upon binding to CANTD, compound 31 and related com-
pounds induce a shift in helices 1 and 2. A single monomer from
the hexameric CA structure of Yeager and co-workers,13 with the
bound position of compound 31 superposed, is shown in Fig-
ure 2A. A hexameric CA model constructed from this superposi-
tion illustrates how the shift in helices 1 and 2 might affect
assembly of the mature capsid (Fig. 2B). The protein surfaces
formed by helices 1 and 2 are localized at the NTD–NTD inter-
faces near the center of the CA hexamer. Since perturbation of
the NTD–NTD interaction has a similar impact on virion pheno-
type in vitro to that observed in our previous study,14–16 we con-
clude that the mode of action of our capsid assembly inhibitors
may involve disruption of the NTD–NTD interaction in the con-
text of mature CA assembly.
N
O
N
N
F3C
OH
O
31
A molecular level explanation of how compound 27 inhibits
viral replication by preventing assembly of immature Gag is
not available. Based on the available information at the time,
we speculated that perturbation of analogous NTD–NTD interac-
tions involved in assembly of both immature Gag and mature CA
could account for our observations.7 However, the recent 8 Å res-
olution cryo-electron microscopy structure of the Mason–Pfizer
monkey virus immature Gag assembly is not consistent with this
hypothesis.17 In this study, it was shown that the role of CANTD
is to hold hexameric building blocks of the immature Gag lattice
together and that the NTD–NTD interface involved in this inter-
action is primarily made through helices 5–7. It is noteworthy
that this model for immature retroviral capsid assembly is com-
pletely consistent with structure-based alanine-scanning muta-
genesis of the HIV-1 CA surface, where mutations on the
surface surrounding helix 5 altered immature particle produc-
tion, presumably by disrupting important interactions during
Gag assembly.15 In contrast to the assembly of mature CA,
helices 1 and 2 are not involved in any protein–protein interac-
tions in the immature Gag lattice. Consequently, it is challenging
7
Figure 1. X-ray co-crystal structure of compound 31 bound to CANTD
.
The protein
surface is coloured by hydrophobicity (green = hydrophobic, white = neutral, and
red = hydrophilic) and has been cut away to reveal the deeply bound compound.
The backbone trace of the protein is shown as a grey coil.
by an increase in lipophilicity, as measured by AlogP. Taken to-
gether, the results indicate SAR from the BD scaffold readily trans-
ferred to the BT and SBD scaffolds, but that such solutions require
careful attention to drug-like properties such as lipophilicity in or-
der to occupy attractive chemical space.
Compound 27 is the most potent molecule identified in this ser-
ies of capsid assembly inhibitors. Improving antiviral potency was
also accompanied by increases in lipophilicity, a general trend that
made balancing potency with desirable ADME-PK properties very
difficult. For example, in rat pharmacokinetic experiments, com-
pound 27 had reasonable iv parameters (total clearance, CLtot = 50%
liver blood perfusion rate, QH; half life, t1/2 = 4.8 h, steady state vol-
ume distribution, Vss = 15 L/kg) but showed very low bioavailabil-
ity. After profiling a number of compounds, further optimization

L. D. Fader et al. / Bioorg. Med. Chem. Lett. 23 (2013) 3401–3405
3405
to implicate the shift in helices 1 and 2 caused by binding of
References and notes
compound 27 to the CANTD in inhibition of immature Gag
assembly.
1. UNAIDS Report on the Global Aids Epidemic, 2012. Available online at: http://
www.unaids.org/en/media/unaids/contentassets/documents/epidemiology/2012/
gr2012/20121120_UNAIDS_Global_Report_2012_en.pdf (Accessed: December
5, 2012).
2. For a comprehensive review of emerging HIV targets see Adamson, C. S.; Freed,
E. O. Antiviral Res. 2010, 119.
3. Fader, L. D.; Landry, S.; Morin, S.; Kawai, S. H.; Bousquet, Y.; Hucke, O.;
Goudreau, N.; Lemke, C. T.; Bonneau, P.; Titolo, S.; Mason, S.; Simoneau, B.
Bioorg. Med. Chem. Lett. 2013, 23, 3396.
4. Fader, L. D.; Bethell, R.; Bonneau, P.; Bös, M.; Bousquet, Y.; Cordingley, M. G.;
Coulombe, R.; Deroy, P.; Faucher, A.-M.; Gagnon, A.; Goudreau, N.; Grand-
Maître, C.; Guse, I.; Hucke, O.; Kawai, S. H.; Lacoste, J.-E.; Landry, S.; Lemke, C.
T.; Malenfant, E.; Mason, S.; Morin, S.; O’Meara, J.; Simoneau, B.; Titolo, S.;
Yoakim, C. Bioorg. Med. Chem. Lett. 2011, 21, 398.
5. Detailed experimental procedures for most of the compounds in the present
manuscript can be found in: Simoneau, B.; Deroy, P.; Fader, L.; Faucher, A.-M.;
Gagnon, A.; Grand-Maitre, C.; Kawai, S.; Landry, S.; Mercier, J.-F.; Rancourt, J.
PCT Int. Appl. WO 2011/100838.
6. Kawatsura, M.; Hartwig, J. F. J. Am. Chem. Soc. 1999, 121, 147.
7. Lemke, C. T.; Titolo, S.; von Schwedler, U.; Goudreau, N.; Mercier, J.-F.;
Wardrop, E.; Faucher, A.-M.; Coulombe, R.; Banik, S. S. R.; Fader, L.; Gagnon, A.;
Kawai, S. H.; Rancourt, J.; Tremblay, M.; Yoakim, C.; Simoneau, B.; Archambault,
J.; Sundquist, W. I.; Mason, S. W. J. Virol. 2012, 86, 6643.
8. IC50 and EC50 values are typically the mean of two determinations.
9. As described in our earlier work, the methyl or ethyl modifications at the N1
position gave compounds with very similar intrinsic and antiviral potency. See
Ref. 4.
10. Lipinski, C. A.; Lombardo, F.; Dominy, B. W.; Feeny, P. J. Adv. Drug Deliv. Rev.
1997, 23, 3.
In conclusion, we have described detailed structure–activity
relationships that allow for the optimization of antiviral potency
a series of 1,5-dihydrobenzo[b][1,4]diazepine-2,4-dione inhibi-
tors of HIV capsid (CA) assembly. Simple substitutions and
introduction of a biaryl motif at the 3 position of the benzodi-
azepine scaffold led to improvements in potency. Combination
of favorable substitutions gave additive SAR and allowed for
the identification of the most potent compound in the series,
analog 27. Productive SAR also transferred to the benzotriaze-
pine and spirobenzodiazepine scaffolds, providing a solution to
the labile stereocenter at the C3 position. Unfortunately, all
meaningful improvements in potency were also accompanied
by increases in lipophilicity, which had a detrimental impact
on ADME-PK properties across the series. The discovery of com-
pound 27 has enabled detailed study of the mechanism of ac-
tion of capsid assembly inhibitors. The molecular basis of how
compound 27 inhibits mature CA assembly can be rationalized
using high-resolution structural information. However, our
understanding of how compound 27 inhibits immature Gag
assembly is not as well understood. Nonetheless, this series of
compounds provides a promising proof of concept for both ma-
ture CA assembly and immature Gag assembly as targets for
therapeutic intervention in the treatment of HIV-infected in-
fected individuals.
11. Ghose, A. K.; Viswanadhan, V. N.; Wendoloski, J. J. J. Phys. Chem. A 1998, 102, 3762.
}
12. Tarcsay, Á.; Nyíri, K.; Keseru, G. M. J. Med. Chem. 2012, 55, 1252.
13. Owen Pornillos, O.; Ganser-Pornillos, B. K.; Kelly, B. N.; Hua, Y.; Whitby, F. G. C.;
Stout, D.; Sundquist, W. I.; Hill, C. P.; Yeager, M. Cell 2009, 137, 1282.
14. Specifically, mutant virus having one of four CA mutations at the NTD–NTD
interface (A22D, E28A/E29A, M39D and A42D) produces virions that lack CA
cones (see Refs. 15,16). Similarly, virus produced in the presence of compound
27 lack CA cones, although virion morphology is also altered in other ways (see
Ref. 7).
Acknowledgments
The authors wish to acknowledge Jean-François Mercier and
Elizabeth Wardrop for IC50 and EC50 value determinations, respec-
tively. We are also grateful to Michael Bös, Richard Bethell and Mi-
chael Cordingley for leadership and guidance throughout the
course of this work.
15. von Schwedler, U. K.; Stray, K. M.; Garrus, J. E.; Sundquist, W. I. J. Virol. 2003, 77,
5439.
16. von Schwedler, U. K.; Stemmler, T. L.; Klishko, V. Y.; Li, S.; Albertine, H. H.;
Davis, D. R.; Sundquist, Wesley I. EMBO J. 1998, 17, 1555.
17. Bharat, T. A. M.; Davey, N. E.; Ulbrich, P.; Riches, J. D.; de Marco, A.; Rumlova,
M.; Sachse, C.; Ruml, T.; Briggs, J. A. G. Nature 2012, 487, 385.