Evaluation of a 3-amino-8-azabicyclo[3.2.1]octane replacement in the CCR5 antagonist maraviroc

Article abstract of DOI:10.1016/j.bmcl.2010.01.080

The bicyclic 5-amino-3-azabicyclo[3.3.0]octanes were shown to be effective replacements for the 3-amino-8-azabicyclo[3.2.1]octane found in the CCR5 antagonist maraviroc.

Full text of DOI:10.1016/j.bmcl.2010.01.080

Bioorganic & Medicinal Chemistry Letters 20 (2010) 1674–1676
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
Evaluation of a 3-amino-8-azabicyclo[3.2.1]octane replacement in the CCR5
antagonist maraviroc
a,
Rémy C. Lemoine a, , Ann C. Petersen , Lina Setti a, Thomas Baldinger a,à, Jutta Wanner a, , Andreas Jekle b,
*
Gabrielle Heilek b, André deRosier b, Changhua Ji b, David M. Rotstein a
a Department of Medicinal Chemistry, Roche Palo Alto, 3431 Hillview Avenue, Palo Alto, CA 94304, USA
b Department of Viral Diseases, Roche Palo Alto, 3431 Hillview Avenue, Palo Alto, CA 94304, USA
a r t i c l e i n f o
a b s t r a c t
Article history:
The bicyclic 5-amino-3-azabicyclo[3.3.0]octanes were shown to be effective replacements for the
3-amino-8-azabicyclo[3.2.1]octane found in the CCR5 antagonist maraviroc.
Ó 2010 Elsevier Ltd. All rights reserved.
Received 3 December 2009
Revised 5 January 2010
Accepted 11 January 2010
Available online 21 January 2010
Keywords:
3-Amino-8-azabicyclo[3.2.1]octane
5-Amino-3-azabicyclo[3.3.0]octane
CCR5 antagonist
Maraviroc
Replacement
In addition to its role as a chemokine receptor on inflammatory
leukocytes, CCR5 is also the primary co-receptor for macrophage-
tropic HIV-1. Since its discovery, CCR5 has proven to be an extraor-
dinary target for the pharmaceutical industry, with demonstrated
therapeutic applications in HIV-treatment, as well as potential
applications in various chronic or acute inflammatory diseases. Over
the last 10 years or so, many CCR5 antagonists chemotypes have
been explored.1 These efforts led to the discovery, optimization,
and clinical development of several compounds. For example,
maraviroc (SelzentryÒ) was the first CCR5 antagonist approved for
the treatment of HIV-1 infection. Maraviroc contains the four phar-
macophore elementsfound in most series of CCR5 antagonists: a ter-
tiary amine, two hydrophobic groups in the western portion of the
molecule(henceforthreferredtoasthetail), and oneheadheteroaryl
groupin theeasternportionof themolecule(thehead). In maraviroc,
the basic amine is part of the rigid 8-azabicyclo[3.2.1]octane bicyclic
system. In our efforts to identify a new series of CCR5 antagonists, we
were, in part, interested in finding a replacement of this system. Syn-
thetically, maraviroc’s triazole ring was derived from a free amine
group at position 3, so, we focused our search on systems that would
have a free amine group appropriately positioned. We identified the
exo-6-amino-3-azabicyclo[3.1.0]hexane (template 1) and the 5-
amino-3-azabicyclo[3.3.0]octane (template 2) as possible replace-
ments for the 3-amino-8-azabicyclo[3.2.1]octane (Fig. 1).
Conformational analysis showed that the 4-substituted 3-
[1,2,4]triazole system derived from 3-aminotropane and the equiv-
alent triazole system derived from template 1 did not overlap well
(Fig. 2). Indeed, its lowest energy conformation presented the tria-
zole ring almost in the same plane as that of the pyrrolidine ring
while in the tropane series, the triazole ring was almost perpendic-
ular to the piperidine ring. Sampling of higher energy conforma-
tions identified conformations in which the triazole ring and the
pyrrolidine ring were close to the adequate perpendicular orienta-
tion. However, they were at least 5.6 kcal higher in energy than the
lowest energy parallel conformation, an energy barrier that was as-
sumed to be too high to overcome.
The 4-substituted endo- or exo-5-[1,2,4]triazole systems derived
from template 2 showed decent overlap with the equivalent triazole
derived from 3-aminotropane (Fig. 2), with the triazole rings almost
perpendicular to the pyrrolidine ring. Since one of the key interac-
tions between CCR5 and an antagonist involves a salt bridge be-
tween the tertiary amine of the antagonist and a glutamic acid in
the CCR5 binding site, we wanted to evaluate the possibility of a dif-
ference in pKa between the basic amine of the tropane ring and that
of template 2 that would affect the binding energy. The calculated
pKa values4 of the tertiary amine of the 4-substituted dimethyl-
1,2,4-triazole derived from 3-aminotropane and that derived from
* Corresponding author. Tel.: +1 650 855 5774; fax: +1 650 855 5237.
E-mail address: remy.lemoine@roche.com (R.C. Lemoine).
Present address: Discovery Chemistry, Hoffmann-La Roche, 340 Kingsland Street,
Nutley, NJ 07110, USA.
à
Present address: DSM Nutritional Products, Wurmisweg 576, 4303 Kaiseraugst,
Switzerland.
0960-894X/$ - see front matter Ó 2010 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2010.01.080

R. C. Lemoine et al. / Bioorg. Med. Chem. Lett. 20 (2010) 1674–1676
1675
F
H
HN
F
template 1
NH2
H
O
HN
H
N
HN
HN
template 2(exo)
HN
N
N
N
NH2
NH2
NH2
H
H
H
Maraviroc
template 2(endo)
Figure 1. The 3-amino-8-azabicyclo[3.2.1]octane core in maraviroc and its potential replacements, template 1 and template 2.
O
Boc
NH
2
(b)
(a)
NBoc
H
H
N
Boc
NH2
OH
OTs
3
(e), (f)
(c)
(g)
(d)
H
H
H
H
H
H
N
Boc
N
Boc
N
Boc
NH2
exo-1
endo-1
NHBn
(f)
H
H
H
H
N
N
4
Boc
Boc
Scheme 1. Reagents and conditions: (a) allyl bromide, NaH, DMF, 0 °C to rt,
overnight (73%); (b) Co2(CO)8, H2O, DME, reflux, 4 h (74%); (c) NaBH4, MeOH, 0 °C,
1 h (>95%); (d) TsCl, DMAP, pyridine, CH2Cl2, reflux, 36 h (52%); (e) NaN3, DMF,
50 °C (78%); H2 (1 atm), Pd(OH)2, MeOH, rt, overnight (>95%); (f) H2 (1 atm),
Pd(OH)2, MeOH, rt, overnight (>95%); (g) BnNH2, NaBH(OAc)3, CH2Cl2, RT (43% after
recrystallization with isopropylacetate/heptane 60/40).
2
Figure 2. Overlaps (overlaps were performed in MOE using C1, N1, C2 and N2 as
anchoring points and were not manually modified) of the lowest energy confor-
mations (conformations generated in MAESTRO (OPLS-2005 force field, CHCl3 as
3
solvent and distance-dependent dielectric constant of 2 to mimic the hydrophobic
nature of the CCR5 binding site) and processed using MOE) of the 4-substituted 3,5-
dimethyl-[1,2,4]triazole derived from the 3-aminotropane system (blue), the N-
methyl exo template 1 (red), the N-methyl exo template 2 (orange) and the N-
methyl endo template 2 (yellow).
noticing the selectivity of the ketone reduction we formed the
amine endo-1 by reacting ketone 2 with benzylamine under reduc-
tive N-alkylation conditions to give N-benzylamine 4 as a 90/10
mixture of endo/exo isomers. The pure endo N-benzylamine was
obtained after two recrystallizations. The N-benzyl group was then
hydrogenolyzed to give endo-1.
Triazoles exo-5 and endo-5 were prepared in six steps (Scheme
2). The exo-1 and endo-1 amines were acylated, the pyrrolidine
nitrogen protecting group was exchanged from a boc group to an
N-benzyl group, the acetamides were treated with phosphorous
pentachloride to give the chloroimidate adducts, which reacted
with acetic hydrazide to give the corresponding acylhydrazidoim-
idate intermediates. These were cyclized under acidic conditions to
give the 4-substituted 1,2,4-triazoles. Finally, the N-benzyl groups
were hydrogenolyzed to give exo-5 and endo-5.
The pyrrolidine was treated under reductive conditions with N-
boc-(S)-3-amino-3-(3-fluoro-phenyl)-propionaldehyde 6 prepared
according to literature procedures6 to give exo-7 and endo-7. The
boc group was hydrolyzed and the amine was acylated with tetra-
hydrofuran-3-carboxylic acid, 4-tetrahydropyrancarboxylic acid,
3,3-difluorocyclobutanecarboxylic acid or 4,4-difluoro-cyclohex-
anecarboxylic acid to give exo-8(a–d), and endo-8(a–d).
For comparative purposes, we prepared the piperidine com-
pounds 9a,d, and the tropane compounds 10(a–d), following the
same sequence.
the N-methyl template 2 were virtually the same (9.47 0.48,
9.42 0.48, respectively). So, we anticipated that the only difference
between the two ring systems would be structural and conforma-
tional. From our conformational analysis, it seemed that the major
differences between the two systems were in the distances between
the basic nitrogen and the triazole system, and the orientation of
both exocyclic exit vectors. However, we hypothesized that the
entropic character of the linker propyl chain could ultimately com-
pensate for the differences in conformational overlap, by allowing
anadjustmentof thethree-dimensional positionsandrelative orien-
tations of the pharmacophores in the somewhatflexible CCR5 recep-
tor binding site.
Conformational analysis alone was not enough to determine
which of the two template 2 amines (exo or endo) would be the
best replacement for the 3-amino-8-azabicyclo[3.2.1]octane,
therefore, compounds in both sub-series were prepared.
Amines exo-1 and endo-1 were prepared (Scheme 1) from a
common intermediate, ketone 2, which itself was synthesized
through a reductive Pauson–Khand cyclization5 of N-boc-N-allyl-
propargylamine. Reduction of ketone 2 with sodium borohydride
gave selectively the endo alcohol 3. Conversion of the alcohol to
the tosylate, displacement of the tosylate with sodium azide and
subsequent hydrogenolysis of the azide gave amine exo-1. It is
noteworthy to point out that the tosylation required more drastic
conditions than usual and gave the tosylate only in 52% yield indi-
cating a substantial steric hindrance of the endo alcohol 3. After
Compounds were tested in two assays, first as inhibitors of
RANTES binding, one of the natural chemokine ligands of CCR5,
then as inhibitors of HIV-1 entry (Table 1).7

1676
R. C. Lemoine et al. / Bioorg. Med. Chem. Lett. 20 (2010) 1674–1676
N N
N
N N
exo-1
endo-1
exo-5
endo-5
NH2
NHAc
H
NHAc
H
N
(g)
(b), (c)
(a)
(d), (e), (f)
H
H
H
H
H
H
H
H
N
Boc
N
Boc
N
Bn
N
H
N
Bn
(h)
R
exo-7
endo-7
O
NHBoc
H
HN
H
O
F
(i), (j)
N
N
N
exo-8
endo-8
a
R =
N
N
N
N
N
H
N
H
N
N
F
R
F
F
b
c
O
HN
(h), (i), (j)
(h), (i), (j)
HN
9
O
F
N
N
N
N
F
R
O
F
HN
10
HN
N
d
N
N
N
N
N
N
F
Scheme 2. Reagents and conditions (yields in parentheses are first for the endo- and then for the exo-bicyclic systems): (a) Ac2O, pyridine, CH2Cl2, rt, 3 h (51%, 38%); (b) HCl
4 M in dioxane, CH2Cl2, rt, 2 h; (c) PhCHO, NaBH(OAc)3, CH2Cl2, rt, overnight (56%, 75% over two steps); (d) PCl5, CH2Cl2, 0 °C, 2 h; (e) AcNHNH2, THF/CH2Cl2, 0 °C to rt,
overnight; (f) AcOH catalytic, ClCH2CH2Cl, reflux, 2 h (67%, 50% over three steps); (g) H2 (1 atm), Pd(OH)2, MeOH, rt, overnight; (h) 6, NaBH(OAc)3, CH2Cl2, rt, 2 h (53%, 54%
over two steps); (i) HCl 4 M in dioxane, CH2Cl2, rt, 2 h; (j) RCOOH, EDCI, HOBt, DIPEA (or Et3N), CH2Cl2 (20–85% over two steps).
modeling, template 2 might be an effective replacement of the
Table 1
conformationally restricted 8-azabicyclo[3.2.1]octane found in
RANTES binding inhibition, and antiviral activity of exo-8(a–d), endo-8(a–d), 9a,d,
and 10(a–d)
maraviroc. It is known from the SAR that led to the discovery of
maraviroc8 and that of our own discovery program,9 that as far
Compounds
Bindinga,c (nM)
Antiviralb,c (nM)
as antiviral activity was concerned, hydrophobic groups in the tail
were better tolerated than more polar ones (e.g., 10b and 10d com-
pared to 10a and 10c), that a larger hydrophobic group was better
tolerated than smaller ones (10d compared to 10b), and that the
tetrahydrofuran ring was better tolerated than the tetrahydropy-
ran ring (10a compared to 10c). We observed similar properties
in the case of the compounds using template 2 (i.e., exo/endo-8a
compared to exo/endo-8c, and exo/endo-8b compared to exo/
endo-8d). However, compounds with hydrophobic groups in the
exo sub-series were surprisingly less potent than the compounds
with polar groups. This indicated to us that the series had a slightly
different binding mode than that maraviroc, and that more SAR
exploration would be required to further optimize both the endo
and the exo sub-series. However, no other compounds were pre-
pared in this new series of CCR5 antagonists.
Maraviroc
Exo-8a
Exo-8b
Exo-8c
Exo-8d
Endo-8a
Endo-8b
Endo-8c
Endo-8d
9a
1.4
35
6
43
—
18
6
30
8
—
38
7
2.8
40
200
P625
81
210
74
P625
41
P625
P625
62
9d
10a
10b
10c
5
13
5
24
220
5
10d
Binding inhibition (IC50) of [128I]-RANTES to CCR5-expressing CHO cells.
Replication inhibition (IC50) of R5 HIVNLBal in JC53-BL cells.
Values are means of at least two experiments.
a
b
c
Based on functional activity, we showed that the bicyclic exo
and endo 5-amino-3-azabicyclo[3.3.0]octanes were effective, but
not equal, replacements for the conformationally restricted 3-ami-
no-8-azabicyclo[3.2.1]octane of the CCR5 antagonist maraviroc.
The introduction of this template into other series of CCR5 antag-
onists will be described in due course.
It is noteworthy to point out the discrepancy between the RAN-
TES binding inhibition and the activity in the antiviral assay. This
indicated to us that binding inhibition was not necessarily corre-
lated with functional inhibition. We hypothesized that upon bind-
ing, the antagonist induces a change of conformation in the
receptor, which in turn prevents the binding of HIVNLBal to CCR5.
Mechanistically, it is still not clear to us why one antagonist would
induce a conformational change leading to a better antiviral activ-
ity than another antagonist. The RANTES binding inhibition assay
was thus used as the first line assay, while we only compared com-
pounds with each other based on their functional antiviral activity.
The piperidine compounds 9a and 9d were devoid of any mea-
surable antiviral activity (upper limit of the assay = 625 nM) while
the direct maraviroc-like equivalents showed reasonable antiviral
activity for 10a (62 nM) and single digit nanomolar activity for
10d (5 nM). Interestingly, exo/endo-8a and exo/endo-8d were still
active in the antiviral assay. We attributed some of the activity to
the increased rigidity of template 2 compared to the piperidine
ring. This was the first indication that, as we had anticipitated from
References and notes
1. Lemoine, R.; Wanner, J. Curr. Topics Med. Chem., in press.
2. MOE software version 2006.08 available from the Chemical Computing Group Inc.,
1010 Sherbrooke Street West, Montreal, Quebec, Canada [www.chemcomp.com].
3. MAESTRO software version 8.5 available from Schrodinger Software Inc., 101 SW
Main Street Suite 1300, Portland, OR 97204, USA [www.schrodinger.com].
4. pKas values were calculated using Moka Linux version 20090616 T.
5. Lee, H.-Y.; An, M.; Sohn, J.-H. Bull. Korean Chem. Soc. 2003, 24, 539.
6. Stupple, P. A. PCT Int. Appl. WO2005033107.
7. For details on these assays, please see: Melville, C. R.; Rotstein, D. M. PCT Int.
Appl. WO2008119663.
8. Wood, A.; Armour, D. Prog. Med. Chem. 2005, 43, 239.
9. Rotstein, D. M.; Melville, C. R.; Padilla, F.; Cournoyer, D.; Lee, E. K.; Lemoine, R.;
Petersen, A. Setti, L. Q.; Wanner, J.; Chen, L.; Filonova, L.; Loughhead, D.; Manka,
J.; Lin, X.-F.; Gleason, S.; Sankuratri, S.; Ji, C.; de Rosier, A.; Dioszegi, M.; Heilek,
G.; Jekle, A.; Berry, P.; Mau, C.-I.; Weller, P. Bioorg. Med. Chem. Lett., in
preparation.