Synthesis of a hexahydropyrimido[1,2-a]azepine-2-carboxamide derivative useful as an HIV integrase inhibitor

Article abstract of DOI:10.1016/j.tetlet.2007.09.072

The hexahydropyrimido[1,2-a]azepine-2-carboxamide derivative 1 could be obtained by three synthetic strategies, which allowed access to multigram amounts of material of high purity and ee. Two strategies involved alternative approaches to the bicyclic pyrimidone core, with the most efficient one being a two-step sequence from commercially available starting materials exploiting a little precedented cyclisation reaction. The remaining steps to 1 included an efficient crystallisation of an intermediate as a single stereoisomer. An alternative strategy employing a chiral starting material led to products of low optical purity but allowed the assignment of the configuration of the stereogenic centre of 1.

Full text of DOI:10.1016/j.tetlet.2007.09.072

Available online at www.sciencedirect.com
Tetrahedron Letters 48 (2007) 8379–8382
Synthesis of a hexahydropyrimido[1,2-a]azepine-2-carboxamide
derivative useful as an HIV integrase inhibitor
Marco Ferrara,* Benedetta Crescenzi, Monica Donghi, Ester Muraglia, Emanuela Nizi,
Silvia Pesci, Vincenzo Summa and Cristina Gardelli
Department of Medicinal Chemistry, IRBM-MRL Rome, Via Pontina, Km 30600, 00040 Pomezia (Rome), Italy
Received 10 July 2007; accepted 11 September 2007
Available online 15 September 2007
Abstract—The hexahydropyrimido[1,2-a]azepine-2-carboxamide derivative 1 could be obtained by three synthetic strategies, which
allowed access to multigram amounts of material of high purity and ee. Two strategies involved alternative approaches to the
bicyclic pyrimidone core, with the most efficient one being a two-step sequence from commercially available starting materials
exploiting a little precedented cyclisation reaction. The remaining steps to 1 included an efficient crystallisation of an intermediate
as a single stereoisomer. An alternative strategy employing a chiral starting material led to products of low optical purity but
allowed the assignment of the configuration of the stereogenic centre of 1.
ꢀ 2007 Elsevier Ltd. All rights reserved.
HIV/AIDS is currently responsible for three million
deaths every year and will continue to be a major threat
as it is estimated that 40 million people are infected
worldwide with HIV and every year four million new
infections occur. Moreover, the increasing resistance to
currently available drugs to treat HIV/AIDS as well as
their side effects renders the discovery of new drugs
imperative. Of particular value would be the identifi-
cation of an HIV integrase inhibitor as there are no
approved drugs that inhibit this enzyme, which proved
to be vital for the replication of the virus.1 Bicyclic pyr-
imidones have recently been identified as potent HIV
integrase inhibitors endowed with desirable preclinical
features and amongst them hexahydropyrimido[1,2-
a]azepine-2-carboxamide derivative 1 proved particu-
larly interesting.2
carboxylate (DMAD) by an oxadiazoline formation/
rearrangement sequence.3,4
To access the seven-membered ring amidoxime 3, N,O-
bis-protected hydroxylamine was N-alkylated with com-
mercially available bromonitrile 4 and the resulting
hydroxylamino nitrile was N-deprotected under acidic
conditions to afford an intermediate, which required
basic conditions to cyclise to the desired O-benzyl-pro-
tected amidoxime (Scheme 2). This cyclisation reaction
however did not take place as cleanly as observed in
the case of the six-membered analogue,3 but was instead
accompanied by side reactions such as conversion of the
nitrile into the ethyl ester. Hydrogenolytic removal of
the benzyl group followed by treatment of 3 with
DMAD afforded bicyclic intermediate 5 by double
Michael addition.5 Upon heating, a rearrangement took
place yielding the N1-alkylated bicyclic pyrimidone 2,
which was protected at the phenolic oxygen as benzoate
to give 6 to ease isolation of the product.4,6
Initial efforts toward 1 were based on the synthetic route
developed for its ring contracted tetrahydro-4H-pyri-
do[1,2-a]pyrimidines analogues3 using 2 as a key inter-
mediate, which required functionalisation at the
pseudobenzylic position and conversion of the ester into
a benzylamide (Scheme 1). Intermediate 2 in turn could
be derived from amidoxime 3 and dimethyl acetylenedi-
As the synthesis described above leads to the hexahydro-
pyrimido[1,2-a]azepine derivative 6 in less than 20%
yield over seven steps, it was decided to seek a more
efficient synthetic route. A recently reported alternative
preparation of N1-alkylated pyrimidones employs
N-alkylated amidoximes and DMAD as starting materi-
als.7 Thus commercially available caprolactam amidox-
ime (7) was reacted with DMAD in chloroform and the
Keywords: Hexahydropyrimido[1,2-a]azepine; Integrase inhibitor;
1,2,4-Oxadiazoline; Amidoxime.
*
Corresponding author. Tel.: +39 0691093488; e-mail: marco_
ferrara@merck.com
0040-4039/$ - see front matter ꢀ 2007 Elsevier Ltd. All rights reserved.
doi:10.1016/j.tetlet.2007.09.072

8380
M. Ferrara et al. / Tetrahedron Letters 48 (2007) 8379–8382
O
N
OH
F
O
N
H
OH
N
OH
NH
N
N
+ MeO2C
CO2Me
O
O
N
N
2
CO2Me
3
N
1
O
Scheme 1.
O
OBz
CO2Me
O
N
CO2Me
CO2Me
f, g
a-d
e
N
N
Br
CN
3
4
N
6
5
Scheme 2. Reagents and conditions: (a) BocNHOBn 1.0 equiv, NaH 1.3 equiv, NaI 0.07 equiv, DMF, 85 ꢁC (87%); (b) HCl satd solution in EtOH,
rt; (c) NEt3 to pH = 10, 1,4-dioxane, rt; (d) Pd/C cat., H2, MeOH; (e) DMAD 1 equiv, NEt3 1 equiv, CHCl3, 0 ꢁC (33%, four steps); (f) o-xylene,
reflux; (g) Bz2O 1.3 equiv, pyridine, rt (59% from 5).
two geometrical isomers of mono-Michael adducts
which formed were separated and subjected indepen-
dently to the rearrangement conditions (Scheme 3).8
The adduct (Z)-8 cyclized to the 1,2,4-oxadiazole 9,
which, in contrast to its regioisomer 5, even after pro-
longed heating in refluxing xylene did not rearrange to
a pyrimidone. On the other hand heating (E)-8 in reflux-
ing xylene led to the formation of the N1-alkylated
pyrimidone 2 and of 1,2,4-oxadiazole 9 in 7:3 ratio,
respectively, while no traces of N3-alkylated pyrimidone
11 were identified.9 The observed overall behaviour was
in good agreement with the data reported for acyclic
amidoximes where the formation of the N1-alkylated
pyrimidone was attributed to an intermediate analogous
to 10, formed by a [1,3]-shift of the Michael adduct
(E)-8, which does not appear to be prone to undergo a
Claisen [3,3]-rearrangement leading to 11.7 In order to
maximise the yield of 2, the Michael addition had to
proceed selectively towards the formation of the E
adduct. After a solvent screening, acetonitrile was found
to favour the formation of the desired adduct, which
took place with good selectivity (Scheme 3).
With this new optimised synthetic route in hand, it was
possible to access the O-protected pyrimidone 6 in only
three steps and more than 30% overall yield from com-
mercially available starting materials, thus improving
substantially the previous synthetic route (Scheme 4).
For the installation of the nitrogen-substituted pseudo-
benzylic stereogenic center present in 1, pyrimidone 6
was monobrominated under radical conditions and the
bromide displaced with enantiopure phenylethylamine
(Scheme 5). The diastereoselectivity of the latter reaction
was dependent on the reaction conditions and a
diastereomeric ratio of 2:1 could be achieved by employ-
ing a mixture of methanol and water at low temperature.
O
1. DMAD 1.2 eq., CH3CN, r.t.
2. o-xylene, reflux
3. Bz2O 1.2 eq., pyridine, r.t.
OBz
NH
NOH
N
N
CO2Me
7
31% yield
6
Scheme 4.
MeO2C
NH
CO2Me
CO2Me
O
N
O
N
N
CO2Me
(Z)-8
NH
NOH
9
DMAD
+
Solvent
O
N
7
NH
OH
[1,3]
N
O
+ 9
CO2Me
N
CO2Me
MeO2C
NH
CO2Me
Solvent
Ratio (
Z
)-8:(
E)-8
(E)-8
O
2
CHCl3
MeOH
Benzene
CH3CN
18:82
65:35
15:85
11:89
[3,3]
N
10
CO2Me
CO2Me
OH
N
N
O
11
Scheme 3.

M. Ferrara et al. / Tetrahedron Letters 48 (2007) 8379–8382
8381
F
O
O
N
H3N
O
O
N
OBz
OH
F
F
d, e
a
b, c
N
N
N
6
H
H
N
N
N
CO2Me
Br
O
12
H2N
CF3CO2
O
13
14
NH
Ph
f, g
1
Scheme 5. Reagents and conditions: (a) NBS 2.0 equiv, AIBN 0.45 equiv, CCl4, 80 ꢁC (50%); (b) (R)-1-phenylethylamine 4.5 equiv, MeOH/H2O 7:3,
ꢀ30 ꢁC to rt; (c) 4-fluorobenzylamine 3 equiv, MeOH, 70 ꢁC (45% from 12); (d) CH2O 6 equiv, NaCNBH3 6.3 equiv, AcOH to pH = 5, CH2Cl2, rt;
(e) Pd/C cat., H2, TFA, MeOH; (f) methyl chlorooxoacetate 2.0 equiv, i-Pr2EtN 4 equiv, CHCl3, rt; (g) excess 2 N Me2NH in MeOH (38% from 13).
A similar behaviour was observed when this reaction
was performed on tetrahydropyridopyrimidone
reaction with DMAD proceeded efficiently and with
high selectivity in favour of the required E-adduct.
Heating of the Michael adducts 17 followed by amide
formation afforded the benzylamide 18, which was
obtained in six steps and 14% overall yield from 15.
However, when 18 was analyzed against an enantio-
merically pure sample of the same compound obtained
by N-a-methyl-benzyl deprotection/N-Boc protection
of 13 (synthesised as described in Scheme 5), unfortu-
nately the compound proved to have 20% ee indicating
that partial racemisation must have occurred along the
synthetic route, possibly in the rearrangement step
which requires high temperatures. As 18 obtained from
L-(ꢀ)-a-amino-e-caprolactam was enriched in the same
enantiomer as the sample derived from 13, the configu-
ration of the stereogenic of 1, previously unknown,
could be assigned as S. To the best of our knowledge,
none of the steps in the synthesis from L-(ꢀ)-a-amino-
e-caprolactam is known from the literature to involve
inversion of configuration at the stereogenic center
under examination. When the Boc protecting group
for the amine in 15 was replaced by a phenylfluorenyl
group, which is known to hamper the epimerisation of
configurationally labile stereogenic centers and to
reduce the reactivity of the nitrogen,12 the pyrimidine
formation step failed thus forcing us to use the synthetic
route employing a-amino-e-caprolactam only for the
synthesis of racemic pyrimidones.
analogues, suggesting that also in our case the reaction
proceeds via an achiral intermediate, possibly with a
p-quinone-like structure, which undergoes a diastereo-
selective attack of the chiral amine.3 (R)-1-Phenylethyl-
amine was found to be the enantiomer leading to the
product enriched with the diastereoisomer required for
the synthesis of 1. Initially the diastereoisomers had to
be separated through a lengthy purification by prepara-
tive RP-HPLC hampering the overall efficiency of the
synthetic route, however, a practical and effective crys-
tallisation technique was later developed. Indeed, amide
13, obtained by reaction with 4-fluorobenzylamine,
could be crystallised from acetonitrile as a single diaste-
reoisomer and in 45% overall yield from 12.10 Reductive
methylation of the secondary amine followed by
hydrogenolytic removal of the chiral auxiliary afforded
ammonium salt 14, which was converted in the final
product 1 by acylation with methyl chlorooxoacetate
and amide formation with dimethylamine. Purification
by preparative RP-HPLC and subsequent crystallisation
from acetonitrile and water afforded 1 in high purity and
>99.5% ee.11
Given the requirement for homochiral 1, an attractive
possibility was the utilization of enantiomerically pure
commercially available a-amino-e-caprolactam 15 as
the starting material in an extension of the above route.
This alternative approach involves the presence of the
chiral centre substituted with the pendant nitrogen from
the beginning, differently from the synthesis described
above where it is installed at a later stage, namely, once
the pyrimidine ring has been constructed. Accordingly,
the amine was Boc-protected, the lactam was converted
to the corresponding thiolactam and then to the cyclic
amidoxime 16 in good overall yield (Scheme 6). The
In summary, three alternative synthetic methods have
been developed for the synthesis of the hexahydropyrim-
ido[1,2-a]azepine-2-carboxamide derivative 1. In the
most efficient route, the pyrimidone core could be
accessed in two steps from commercially available
starting materials employing a cyclisation method
unprecedented for bicyclic systems. The synthesis could
be completed by further eight xsteps including an
O
OH
F
d
e, f
a, b, c
NH
NOH
NH
NH
N
H
O
N
O
N
CO2Me
N
NHBoc
CO2Me
O
NH2
15
NHBoc
NHBoc
18
16
17
Scheme 6. Reagents and conditions: (a) Boc2O 1.3 equiv, i-Pr2EtN 2 equiv, CH2Cl2, rt (91%); (b) P2S5 0.37 equiv, Me3SiOSiMe3 1.67 equiv, CH2Cl2,
rt (88%); (c) NH2OHÆHCl 3 equiv, i-Pr2EtN 3 equiv, MeOH, reflux (53%); (d) DMAD 1.1 equiv, CH3CN, rt (73%); (e) o-xylene, reflux (54%); (f) 4-
fluorobenzylamine 2.5 equiv, MeOH, reflux (81%).

8382
M. Ferrara et al. / Tetrahedron Letters 48 (2007) 8379–8382
7. Colarusso, S.; Attenni, B.; Avolio, S.; Malancona, S.;
Harper, S.; Altamura, S.; Koch, U.; Narjes, F. ARKIVOC
2006, 479–495.
8. The C@C bond geometry of (Z)- and (E)-8, which
appeared as single isomers at the C@N bond, was
determined by NMR, measuring the vicinal heteronuclear
coupling constant MeO2CC@CH [J = 9.3 Hz for (E)-8,
4.3 Hz for (Z)-8].
efficient crystallisation of an intermediate as single
stereoisomer allowing access to the final compound in
high purity and ee. Multigram amounts of 1 could be
obtained via this route. An alternative strategy
employing a chiral starting material led to products of
low optical purity but allowed the assignment of the
configuration of the stereogenic centre of 1.
9. Samples of 6 obtained from the two alternative routes
proved to have identical properties and the regiochemistry
of the alkylation was assigned at N1 rather than at N3 by
1H–13C-HMBC analysis as CH2N correlates with C6@O.
1
References and notes
10. Selected data for 13: H NMR (400 MHz, DMSO-d6) d:
10.23 (s, 1H), 7.42–7.33 (m, 6H), 7.28 (t, J = 6.6 Hz, 2H),
7.21–7.11 (m, 5H), 4.75 (dd, J = 5.8, 13.1 Hz, 1H), 4.55–
4.25 (m, 2H), 3.81 (s, 2H), 3.61–3.47 (m, 3H in part
obscured by water signal), 1.86–1.75 (m, 1H), 1.72–1.63
(m, 2H), 1.48–1.30 (m, 3H), 1.24 (d, J = 6.5 Hz, 3H). 13C
NMR (100 MHz, DMSO-d6, 300 K) d: 167.97, 161.11 (d,
JC–F = 242 Hz), 161.05 (d, JC–F = 242 Hz), 158.72, 150.75,
148.20, 146.03, 137.46, 134.98, 129.16 (d, JC–F = 8 Hz),
127.88, 126.52, 126.26, 124.43, 114.81 (d, JC–F = 27 Hz),
114.59 (d, JC–F = 26 Hz), 58.47, 55.98, 43.81, 41.21, 40.65,
33.31, 26.27, 25.18, 25.35. MS m/z: 451 (M+H)+.
1. (a) Hazuda, D. J.; Felock, P.; Witmer, M.; Wolfe, A.;
Stillmock, K.; Grobler, J. A.; Espeseth, A.; Gabryelski, L.;
Schleif, W.; Blau, C.; Miller, M. D. Science 2000, 287,
646–650; (b) Hazuda, D. J.; Young, S. D.; Guare, J. P.;
Anthony, N. J.; Gomez, R. P.; Wai, J. S.; Vacca, J. P.;
Handt, L.; Motzel, S. L.; Klein, H. J.; Dornadula, G.;
Danovich, R. M.; Witmer, M. V.; Wilson, K. A.; Tussey,
L.; Schleif, W. A.; Gabryelski, L. S.; Jin, L.; Miller, M. D.;
Casimiro, D. R.; Emini, E. A.; Shiver, J. W. Science 2004,
305, 528–532.
11. Selected data for 1: 1H NMR (400 MHz, DMSO-d6,
300 K, mixture of rotamers) d: 12.29 (br s, 0.1H), 11.95 (br
s, 0.9H), 9.30 (br s, 0.9H), 8.51–8.45 (m, 0.1), 7.41–7.35
(m, 1.8H, in part overlaid by the following signal), 7.35–
7.33 (m, 0.2H, in part overlaid by the previous signal),
7.22–7.12 (m, 2H), 5.45–5.25 (m, 0.1H), 4.94 (dd, J = 5.7,
J = 14.0 Hz, 0.9H), 4.84–4.79 (m, 0.1H), 4.57–4.43 (m,
2H), 3.54 (dd, J = 14.0, J = 11.0 Hz, 0.9H), 3.28–3.18 (m,
0.1H), 3.05 (s, 0.3H), 2.92 (s, 2.7H), 2.90 (s, 5.4H), 2.81 (s,
0.3H), 2.76 (s, 0.3H), 1.78–2.19 (m, 5H), 1.41–1.27 (m,
1H).13C NMR (100 MHz, DMSO-d6, 300 K, only major
rotamer reported) d: 168.01, 165.80, 165.03, 161.30 (d,
JC–F = 243 Hz), 157.68, 149.67, 145.94, 134.59, 129.56 (d,
JC–F = 8.5 Hz), 124.72, 115.10 (d, JC–F = 21 Hz), 55.88,
42.42, 41.56, 36.24, 32.79, 32.34, 28.83, 27.10, 26.15. MS
2. Crescenzi, B.; Gardelli, C.; Donghi, M.; Ferrara, M.; Pace,
P.; Kinzel, O.; Muraglia, E.; Rowley, L.; Fiore, F.;
Gonzalez Paz, O.; Fonsi, M.; Stillmock, K. A.; Witmer,
M. V.; Hazuda, D. J.; Summa, V. Abstracts of Papers,
232nd ACS National Meeting, San Francisco, CA, USA,
Sept 10–14, 2006.
3. Kinzel, O. D.; Monteagudo, E.; Muraglia, E.; Orvieto, F.;
Pescatore, G.; Rico Ferreira, M.; Rowley, M.; Summa, V.
Tetrahedron Lett. 2007, 48, 6552–6555.
4. (a) Culbertson, T. P. J. Heterocycl. Chem. 1979, 16, 1423–
1424; (b) Cooper, J.; Irwin, W. J. J. Chem. Soc., Perkin
Trans. 1 1976, 2038–2045; (c) Summa, V.; Petrocchi, A.;
Matassa, V. G.; Taliani, M.; Laufer, R.; De Francesco, R.;
Altamura, S.; Pace, P. J. Med. Chem. 2004, 47, 5336–5339.
5. A similar transformation has been reported recently:
Naidu, B. N.; Sorenson, M. E. Org. Lett. 2005, 7, 1391–
1393.
20
m/z: 460 (M+H)+. ½aꢁD ꢀ72 2 (c = 0.1, CHCl3). Ee
calculated by HPLC (stationary phase: column Chiralpak
AD, 4.6 · 250 mm. Mobile phase: 50/50 n-hexane with
0.2% TFA/EtOH with 3% MeOH, 1 mL/min. Retention
time: 23.25 min. Enantiomer has retention time:
11.44 min).
6. Selected data for 6: 1H NMR (400 MHz, DMSO-d6) d 8.07
(d, J = 7.3 Hz, 2H), 7.78 (t, J = 7.6 Hz, 1H), 7.62 (t,
J = 7.6 Hz, 2H), 4.31–4.29 (m, 2H), 3.74 (s, 3H), 3.07–3.04
(m, 2H), 1.82–1.65 (m, 6H). 13C NMR (75 MHz, 300 K,
DMSO-d6) d 162.8, 162.7, 162.4, 157.0, 140.3, 135.2, 134.2,
129.6, 128.9, 127.5, 52.5, 43.1, 36.1, 28.5, 26.1, 23.7. MS
m/z: 343 (M+H)+.
12. See: Greene, T. W.; Wuts, P. G. M. In Protective Groups in
Organic Synthesis, 3rd ed.; John Wiley & Sons: New York,
1999.