Core modification of cytisine: A modular synthesis

Article abstract of DOI:10.1002/anie.201100441

Getting down to the core: A novel, modular, and more robust synthesis of cytisine, a partial agonist selective for the α4β2 nicotinic acetylcholine receptor, also allows modification of the core structure, as exemplified by the first azacytisine and a cytisine-varenicline hybrid. Key steps include Stille coupling of heteroarylstannanes with a bromolactam motif and an in situ epimerization/alkylative cyclization to complete the tricyclic core (see scheme). Copyright

Full text of DOI:10.1002/anie.201100441

Communications
DOI: 10.1002/anie.201100441
Alkaloid Synthesis
Core Modification of Cytisine: A Modular Synthesis**
Christoph Hirschhꢀuser, Claire A. Haseler, and Timothy Gallagher*
(ꢀ)-Cytisine (1) has occupied a pivotal role in nicotinic
pharmacology as a partial agonist at the a4b2 subtype of
neuronal nicotinic acetylcholine receptor (nAChR).[1] More
recently because of this partial agonist profile, cytisine
provided the discovery lead for varenicline (2), Pfizerꢀs
smoking cessation agent which was launched in 2006.[2,3]
Recent interest in the chemistry of cytisine (1) has been
high because of a lack of understanding as to the molecular
basis of this underpinning partial agonist profile. In terms of
analogues, manipulation of cytisine itself is readily achievable
at C3 and C5,[4] C10,[5] and N12[6] but interest in cytisine as a
target for total synthesis has also been significant.[7] Synthesis
offers opportunities for variation within other regions of the
Figure 1. Cytisine (1), varenicline (2), and new core-modified deriva-
tives 3 and 4.
heterocyclic scaffold,[8] and critically, access to more funda-
mentally, core-modified cytisine congeners.
To date, these have encompassed, for example, norcyti-
sines (involving deletion of C8[9] or C13),[10] however, these
specific modifications lead to substantial loss of nicotinic
binding affinity. New insights linked to the detail of the ligand
binding site can now offer a rationalization of these and
related results. Crystallographic data for a number of ligands
bound to acetylcholine binding protein (AChBP) have been
recently published.[11] These protein complexes closely resem-
ble the extracellular binding domain of the a4b2 nicotinic
receptor[12] where part of the ligand binding site is charac-
terized by a highly conserved “aromatic box” comprised of
tyrosine and tryptophan residues. Taking this further, Abin-
Carriquiry et al.[13] have very recently derived a model for
cytisine (and 3- and 5-halocytisines) complexed to the AChBP
binding site (Figure 1). This binding model was validated by
correlation to experimental binding affinities and shows the
bispidine moiety (and associated ammonium) of cytisine to be
locked within the conserved “aromatic box”; this allows the
lack of activity of norcytisines, where the bispidine scaffold
has been modified, to be explained. In contrast, the pyridone
component of 1 (and by analogy the quinoxaline ring of
varenicline, 2) is directed into a variable region of the binding
site suggesting this as an area of molecular space that provides
an opportunity for achieving (and rationalizing) nicotinic
subtype selectivity.
Our goal was to develop a modular synthetic approach
towards cytisine to address, within the context of this variable
region, the “need to develop an approach to cytisine that
delivers a late-stage intermediate that is equipped … for
analogue preparation.”[7] To challenge the strategy, a ring-
extended derivative and a representative azacytisine, exem-
plified by ligands 3 and 4, respectively, were targeted. In
addition, these variants provide potential for new receptor
binding interactions, unavailable by established and more
conventional derivatizations of 1, and an opportunity to
elucidate those structural features of cytisine that contribute
to its selective partial agonist profile.
We have recently reported an efficient and convergent
entry to cytisine (as well as unnatural (+)-cytisine and other
lupin alkaloids) based on the intramolecular addition of a
lactam enolate to a pyridone in order to establish the tricyclic
[14]
ꢀ
framework by construction of the C6 C7 bond. This early
approach, however, failed to secure core variants such as 3
and 4.[15] As a result, an alternative and ultimately more
robust and flexible strategy to cytisine (1) has been defined,
the advantages of which are exemplified here by the
successful synthesis of prototype analogues 3 and 4.
The synthesis of cytisine (1) is shown in Scheme 1 where
ꢀ
formation of the C6 C7 bond, based on bromoalkene 6b as a
pivotal intermediate, plays a central role. Bromide 6b (readily
available from lactam 5a)[16] underwent smooth Stille cou-
pling to the 6-stannylpyridine 7 to provide adduct 8 in 99%
yield. Hydrogenation followed by desilylation generated
lactams cis-9 and trans-9. The diastereoselectivity of this
[*] Dr. C. Hirschhꢀuser, C. A. Haseler, Prof. Dr. T. Gallagher
School of Chemistry, University of Bristol
Bristol BS8 1TS (UK)
Fax: (+44)117-929-8611
E-mail: t.gallagher@bristol.ac.uk
Homepage: http://www.chm.bris.ac.uk/org/gallagher/
indextcg.htm
transformation was not
a concern. Alcohol activation
(through mesylation) and thermal cyclization (through N al-
kylation and O demethylation),[17] with concomitant in situ
epimerization, provided tricyclic lactam 10 in 91% overall
yield. The latter was readily converted to 1 in two steps:
selective lactam reduction and N debenzylation.[14] The key
[**] Funding from the Rotary Foundation (to C.H.) and the EPSRC (to
C.A.H.) is acknowledged.
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/anie.201100441.
5162
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2011, 50, 5162 –5165

mixture of cis and trans diastereomers. This mixture was
subjected to mesylation followed immediately by in situ
epimerization/cyclization to provide lactam 14 in 59% overall
yield. Selective lactam reduction and N debenzylation yielded
the target ring-extended isoquinolone 3, the first example of a
cytisine–varenicline hybrid, in 58% yield over two steps.
The third case study reported here describes the first
example of an aza-variant of cytisine, which represents a
fundamental modification within the pyridone moiety that is
only accessible by synthesis. The 4-aza isomer 4 was chosen as
the exemplar and the synthesis of this target is shown in
Scheme 3. Again, access to azacytisine 4 was also not
Scheme 1. Synthesis of cytisine (1): a) TBSCl, Et3N, CH2Cl2, 37 h, 88%;
b) 1. BuLi, PhSeCl, THF, ꢀ788C to 08C, 2. THF/MeOH/H2O, NaIO4,
14 h, 50%; c) 1. Br2, CH2Cl2, 458C, 3 h, 2. Et3N, CH2Cl2, 458C, 17 h,
63%; d) 7, [Pd(PPh3)4], CuCl, LiCl, THF, 708C, 15 h, 99%; e) 1. H2,
Pd/C, MeOH, 6 h, 2. TBAF, THF/H2O, 508C, 36 h, 85% (cis/trans
1.5:1.0); f) 1. MeSO2Cl, Et3N, CH2Cl2, 08C, 20 min, 2. PhMe/DMF
(9:1), 1108C, 21 h, then LiHMDS (in THF) was added at RT, 1108C,
18 h, 91%; g) see Ref. [14]. Bn=benzyl, DMF=N,N’-dimethylform-
amide, HMDS=hexamethyldisilazide, TBAF=tetra-n-butylammonium
fluoride, TBS=tert-butyldimethylsilyl, THF=tetrahydrofuran.
feature exploited here was facile interconversion of the
mesylates derived from cis- and trans-9 under basic condi-
tions, thereby channelling both isomers through the cycliza-
tion step to provide tricycle 10 in high yield.
Scheme 3. Synthesis of 4-azacytisine (4): a) 6b, [Pd(PPh3)4], CuCl, LiCl,
THF, 708C, 15 h, 86%; b) LiAlH4, THF, 08C, 15 min, 89% (cis/trans
2.7:1.0); c) TBAF, THF/H2O (9:1), 508C, 36 h, 89% (cis/trans 1.5:1.0);
d) 1. MeSO2Cl, Et3N, CH2Cl2, 08C to RT, 2. Et3N, PhMe/DMF (2:1),
1108C, 15 h, 79%; e) LiAlH4, THF, 15 h, then TBAF, THF/H2O (9:1),
508C, 16 h, 49%; f) 1. MeSO2Cl, Et3N, CH2Cl2, 08C to RT, 2. Et3N,
PhMe/DMF (4:1), 1108C, 15 h, 73%; g) 1. 1-chloroethyl chloroformate,
ClCH2CH2Cl, 808C, 17 h, 2. MeOH, 658C, 3 h, 61%.
The ring-extended derivative 3 represents a bridge
between cytisine (1) and varenicline (2), and such a hybrid
offers a new vehicle for exploration of the partial agonist
behavior associated with cytisine. The synthesis of the ring-
extended isoquinolone 3 based on extrapolation of the
approach devised for cytisine is shown in Scheme 2. Stille
coupling of bromide 6b with the requisite isoquinoline-based
stannane 11 (see the Supporting Information) smoothly
provided adduct 12, which was reduced and simultaneously
deprotected to afford alcohols 13 as an inseparable 1.7:1
achievable using our earlier[14] chemistry as the requisite
pyrazine precursor (and also the corresponding pyrimidine-
based substrate) was not tolerant to (and decomposed under)
the conditions needed for lactam enolate formation.[15]
6-Methoxypyrazine tributylstannane 15 was conveniently
prepared in two steps from 2,6-dichloropyrazine by consec-
utive nucleophilic substitution with NaOMe and Bu3SnLi.
Stille coupling of 15 with building block 6b gave adduct 16
(86%), reduction of which was best achieved using LiAlH4 to
give lactam 17 as a 2.7:1 mixture of cis/trans diastereomers.
Desilylation of 17 with commercially available tetrabutylam-
monium fluoride (TBAF; containing 5% H2O) in THF led to
epimerization at C3 and a 1:1 mixture of diastereomeric
alcohols 18. This was suppressed (but not totally) by addition
of water and conducting the reaction in THF/H2O (9:1); this
produced 18 as a 1.5:1 cis/trans mixture. With the increased
acidity of the proton at C3, epimerization/cyclization of the
mesylates derived from alcohols 18 was possible under
exceptionally mild conditions (with Et3N) providing tricycle
Scheme 2. Synthesis of isoquinolone 3: a) 6b, [Pd(PPh3)4], CuCl, LiCl,
THF, 708C, 24 h, 92%; b) H2, Pd/C, MeOH, 22 h, 100% (cis/trans
1.7:1.0); c) 1. MeSO2Cl, Et3N, CH2Cl2, 08C, 20 min, 2. PhMe, 1108C,
15 h, then LiHMDS (in THF) was added at RT, 1108C, 24 h, 59%;
d) BH3, THF, 3 h, 85%; e) Pd(OH)2, MeOH, HCl, H2, 5 h, 68%.
Angew. Chem. Int. Ed. 2011, 50, 5162 –5165
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
5163

Communications
S. Demphel, J. Gopisetti, C. Coulon, M. Ottaviani, L. A. Wrenn,
S. R. Letchworth, G. A. Bohme, J. Benavides, M.-C. Lasne, M.
Bottlaender, F. Dollꢃ, Bioorg. Med. Chem. 2003, 11, 5333 – 5343.
A small number of C4-substituted cytisine derivatives have been
reported; see Ref. [4c] and [8a].
19 in 79% overall yield. However, selective reduction of the
lactam carbonyl in the presence of the pyrazinone moiety (i.e.
19!21) was not successful, despite screening a range of
reagents and conditions.[18] This issue was solved by LiAlH4
reduction of lactam 17 (as a 2.7:1 mixture of cis and trans
isomers), that is, prior to unmasking of the pyrazinone. This
was, however, relatively inefficient, which we attributed to
formation of aluminium complexes on basic workup. Optimal
conditions involved LiAlH4 reduction of 17 and direct
treatment (with minimal workup) of the crude product with
TBAF yielding the diastereomerically pure alcohol 20 in 49%
yield (based on cis-17). Cyclization of 20 provided the
N-benzyl 4-azacytisine 21 which was best deprotected under
dealkylative conditions using 1-chloroethyl chloroformate
followed by methanolysis to give the target 4-azacytisine 4 in
61% yield.[19]
In conclusion, we have defined a new, highly modular
approach to the synthesis of cytisine, that is significantly more
robust and adaptable, and allows access to previously
inaccessible core-modified derivatives, exemplified by the
cytisine–varencline hybrid 3 and the first example of an
azacytisine, that is, pyrazinone 4. This chemistry features the
use of heteroarylstannanes, which are coupled with bromide
6b (as the critical, common and late-stage intermediate), and
unmasked (to release the pyridone moiety). Key to the
efficient construction of the tricyclic core scaffold was use of a
novel in situ epimerization/cyclization sequence that largely
negated a requirement for cis/trans stereocontrol. Application
of this methodology to other cytisine variants together with
ligand-based studies exploiting the acetylcholine binding
protein (AChBP) are underway.[20]
[5] a) N. Houllier, S. Gouault, M.-C. Lasne, J. Rouden, Tetrahedron
2006, 62, 11679 – 11686; b) J. Rouden, A. Ragot, S. Gouault, D.
Cahard, J.-C. Plaquevent, M.-C. Lasne, Tetrahedron: Asymmetry
2002, 13, 1299 – 1305; c) F. Frigerio, C. A. Haseler, T. Gallagher,
Synlett 2010, 729 – 730.
[6] C. Canu Boido, B. Tasso, V. Boido, F. Sparatore, Farmaco 2003,
58, 265 – 277.
[7] D. Stead, P. OꢀBrien, Tetrahedron 2007, 63, 1885 – 1896.
[8] For the synthesis of C4-substituted derivatives see: a) B. T.
OꢀNeill, D. Yohannes, M. W. Bundesmann, E. P. Arnold, Org.
Lett. 2000, 2, 4201 – 4204; b) Ref. [4c]; c) P. Durkin, P. Magrone,
S. Matthews. C. Dallanoce, T. Gallagher, Synlett 2010, 18, 2789 –
2791.
[9] D. Yohannes, K. Procko, L. A. Lebel, C. B. Fox, B. T. OꢀNeill,
Bioorg. Med. Chem. Lett. 2008, 18, 2316; as well as Ref. [8c].
[10] D. Yohannes, C. P. Hansen, S. R. Akireddy, T. A. Hauser, M. N.
Kiser, N. J. Gurnon, C. S. Day, B. Bhatti, W. S. Caldwell, Org.
Lett. 2008, 10, 5353 – 5356.
[11] a) A. B. Smit, N. I. Syed, D. Schaap, J. van Minnen, J. Klumper-
man, K. S. Kits, H. Lodder, R. C. van der Schors, R. van Elk, B.
Sorgedrager, K. Brejc, T. K. Sixma, W. P. M. Geraerts, Nature
2001, 411, 261 – 268; b) K. Brejc, W. J. van Dijk, R. V. Klaassen,
M. Schuurmans, J. van der Oost, A. B. Smit, T. K. Sixma, Nature
2001, 411, 269 – 276; c) P. H. N. Celie, S. E. van Rossum-Fikkert,
W. J. van Dijk, K. Brejc, A. B. Smit, T. K. Sixma, Neuron 2004,
41, 907 – 914; d) Y. Bourne, T. T. Talley, S. B. Hansen, P. Taylor, P.
Marchot, EMBO J. 2005, 24, 1512 – 1522; e) S. B. Hansen, G.
Sulzenbacher, T. Huxford, P. Marchot, P. Taylor, Y. Bourne,
EMBO J. 2005, 24, 3635 – 3646.
[12] C. D. Dellisanti, Y. Yao, J. C. Stroud, Z. Z. Wang, L. Chen, Nat.
Neurosci. 2007, 10, 953 – 962.
[13] J. A. Abin-Carriquiry, M. P. Zunini, B. K. Cassels, S. Wonnacott,
F. Dajas, Bioorg. Med. Chem. Lett. 2010, 20, 3683 – 3687.
[14] D. Gray, T. Gallagher, Angew. Chem. 2006, 118, 2479 – 2483;
Angew. Chem. Int. Ed. 2006, 45, 2419 – 2423.
[15] T. Gallagher, I. Derrick, P. M. Durkin, C. A. Haseler, C.
Hirschhꢄuser, P. Magrone, J. Org. Chem. 2010, 75, 3766 – 3774.
Key step of our earlier method was the intramolecular 1,6-
addition of a lactam enolate to a 2-pyridone (22a!23a). This
was unsuccessful for 22b!23b and 22c!23c where evidence
was obtained for addition of the HMDS anion to the heterocycle.
Received: January 18, 2011
Published online: April 21, 2011
Keywords: acetylcholine receptor · cytisine · heterocycles ·
.
pyridone · varenicline
[1] A. A. Jensen, B. Frølund, T. Lijefors, P. Krogsgaard-Larsen, J.
Med. Chem. 2005, 48, 4705 – 4745. For leading references to
nicotinic pharmacophore studies see: a) R. A. Glennon, M.
Dukat, Bioorg. Med. Chem. Lett. 2004, 14, 1841 – 1844; b) J. E.
Tønder, P. H. Olesen, Curr. Med. Chem. 2001, 8, 651 – 674.
[2] a) J. W. Coe, P. R. Brooks, M. G. Vetelino, M. C. Wirtz, E. P.
Arnold, J. Huang, S. B. Sands, T. I. Davis, L. A. Lebel, C. B. Fox,
A. Shrikhande, J. H. Heym, E. Schaeffer, H. Rollema, Y. Lu,
R. S. Mansbach, L. K. Chambers, C. C. Rovetti, D. W. Schulz,
F. D. Tingley, B. T. OꢀNeill, J. Med. Chem. 2005, 48, 3474 – 3477;
b) J. W. Coe, H. Rollema, B. T. OꢀNeill, Annu. Rep. Med. Chem.
2009, 44, 71 – 101.
[3] Natural cytisine has also been used extensively within Eastern
Europe for smoking cessation. J. F. Etter, Arch. Intern. Med.
2006, 166, 1553 – 1559.
[16] All compounds described here are racemic. The synthesis of (R)-
5a and nonnatural (+)-cytisine has been reported; see Ref. [14]
and also: M. H. Kim, Y. Park, B. S. Jeong, H. G. Park, S. S. Jew,
Org. Lett. 2010, 12, 2826 – 2829.
[17] A related sequence involving configurationally stable cis-piper-
idines has been described, see Ref. [8a] and T. Honda, R.
Takahashi, H. Namiki, J. Org. Chem. 2005, 70, 499 – 504.
[18] For conditions used see the Supporting Information. Selective
[4] a) P. Imming, P. Klaperski, M. T. Stubbs, G. Seitz, D. Gꢁndisch,
Eur. J. Med. Chem. 2001, 36, 375 – 388; b) E. Marriꢂre, J.
Rouden, V. Tadino, M.-C. Lasne, Org. Lett. 2000, 2, 1121 – 1124;
c) S. K. Chellappan, Y. Xiao, W. Tueckmantel, K. J. Kellar, A. P.
Kozikowski, J. Med. Chem. 2006, 49, 2673 – 2676; d) A. P.
Kozikowski, S. K. Chellappan, Y. Xiao, K. M. Bajjuri, H. Yuan,
K. J. Kellar, P. A. Petukhov, ChemMedChem 2007, 2, 1157 –
1161; e) G. Roger, B. Lagnel, J. Rouden, L. Besret, H. Valette,
=
thionation of the lactam C O of 19 using Lawessonꢀs reagent
failed (see the Supporting Information).
5164
www.angewandte.org
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2011, 50, 5162 –5165

[19] Debenzylation of 21 under H2 in the presence of Pd/C and HCl/
MeOH led to pyrazinone reduction and attempts at oxidative
debenzylation with ceric ammonium nitrate (CAN) in MeCN/
H2O caused degradation.
[20] Initial pharmacological evaluation of the ligands reported here
has focused on the ring-extended isoquinolone 3, as the first
annulated cytisine and because of its structural relationship to
varenicline. Using a whole-brain (heteromeric) assay with
200 pm [3H]epibatidine, racemic isoquinolone 3 was a potent
ligand with an IC50 of 85 nm (cf. (ꢀ)-cytisine 7.7 nm; A. Kennett,
S. Wonnacott, C. Hirschhꢄuser, C. A. Haseler, T. Gallagher,
unpublished results). Coe et al. have reported that varenicline is
three-fold more potent than cytisine at a4b2 in rat cortex; see
Ref. [2a].
Angew. Chem. Int. Ed. 2011, 50, 5162 –5165
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
5165