A scalable synthesis of the antidepressant agomelatine by a tandem allylic chlorination-isomerization process

Article abstract of DOI:10.1002/ejoc.201402886

A concise, scalable, and industrially applicable process for the synthesis of the antidepressant agomelatine is described. The process relies on a tandem allylic chlorination-isomerization sequence, on a tetralone-derived allyl carbinol, as the key transformation. The target compound is obtained in five steps from commercially available 7-methoxy-1-tetralone, in 52.3 % overall yield after final recrystallization. A tandem allylic rearrangement-isomerization sequence on a tetralone-derived allyl carbinol was applied for the preparation of agomelatine, a high-potential antidepressant agent with improved pharmacokinetic profile. The active pharmaceutical ingredient (API) was obtained in 52.3 % overall yield and excellent purity (> 99.5 %) by a simple process amenable to industrial-scale applications.

Full text of DOI:10.1002/ejoc.201402886

SHORT COMMUNICATION
DOI: 10.1002/ejoc.201402886
A Scalable Synthesis of the Antidepressant Agomelatine by a Tandem Allylic
Chlorination–Isomerization Process
Christos I. Stathakis,*[a] Efstratios Neokosmidis,[a] and Theocharis V. Koftis*[a]
Keywords: Antidepressants / Agomelatine / Total synthesis / Isomerization / Industrial chemistry
A concise, scalable, and industrially applicable process for
the synthesis of the antidepressant agomelatine is described.
The process relies on a tandem allylic chlorination–isomer-
ization sequence, on a tetralone-derived allyl carbinol, as the
key transformation. The target compound is obtained in five
steps from commercially available 7-methoxy-1-tetralone, in
52.3% overall yield after final recrystallization.
ability to resynchronize the disturbed circadian rhythms,
hence improving sleep disorders.[3c,3d]
Introduction
Melatonin is a very important neurohormone, involved
in several pathophysiological processes such as sleep, sea-
sonal disorders, depression, and aging.[1] Despite its well-
recognized significance, the therapeutic usage of melatonin
is strongly hampered by its extremely fast metabolism
(plasma half-life about 15 min).
Agomelatine, a bioisosteric analogue of melatonin in
which the indole core has been replaced by a naphthalene
core (Scheme 1), has been developed by the pharmaceutical
company Servier and demonstrates an improved pharmaco-
kinetic profile.[2] It is marketed under the trade name Val-
doxan and prescribed for the treatment of major depressive
disorder. Agomelatine compares favorably with other drugs
of the same class, as a result of its dual mechanism of ac-
tion.[3a,3b] It performs as an agonist of MT1 and MT2 mela-
tonergic receptors and in the same time as a 5-HT2C recep-
tor antagonist, a combination which, in addition to the
antidepressant properties, provides agomelatine with the
In view of its distinctive biological properties and its high
marketing potential, much effort has been put into the de-
velopment of industrially applicable synthetic routes
towards this active pharmaceutical ingredient (API).[4]
Early work by the inventors utilized a Reformatsky reaction
on 7-methoxy-1-tetralone (1) for installation of the side
chain, a process that proved to be hardly reproducible and
thus unreliable (Scheme 2).[2]
Scheme 2. Summary of previously existing methods for the indus-
trially applicable synthesis of agomelatine.
In a revised procedure by the same laboratory, the side
arm was introduced effectively by a Knoevenagel condensa-
tion with monomalononitrile, providing intermediate 3;
however, aromatization to nitrile 4 was not easy to repro-
duce and required prolonged heating at elevated tempera-
tures.[5a] In addition, environmental concerns arose from
the utilization of toxic compounds such as allyl meth-
acrylate, heptanoic acid, and benzylamine. In an alternative
route, via carbinol 5, low temperatures (–70 °C) were re-
Scheme 1. Structures of melatonin and its bioisosteric analogue,
agomelatine.
[a] Pharmathen S.A.,
9th km Thermi-Thessaloniki, Thessaloniki 57001, Greece
E-mail: hkoftis@pharmathen.com
cstathakis@pharmathen.com
www.pharmathen.gr
Supporting information for this article is available on the
WWW under http://dx.doi.org/10.1002/ejoc.201402886.
6376
© 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Eur. J. Org. Chem. 2014, 6376–6379

A Scalable Synthesis of the Antidepressant Agomelatine
quired, which decreased the cost effectiveness.[5b] Despite
the large number of synthetic approaches, disadvantages
like those mentioned above indicated that the optimum in-
dustrially applicable synthesis of agomelatine was yet to be
achieved.
Results and Discussion
In our laboratory, we envisioned that agomelatine could
be prepared from chloride 6, through an aromatization, am-
monolysis, and acetylation sequence (Scheme 3). Intermedi-
ate 6, in turn, could be accessed from carbinol 7 by allylic
substitution, as the key step, by using a chlorination agent
such as HCl, SOCl2, PCl3 or a similar compound, and
isomerization of the double bond. Although such allylic
substitution reactions are very common for quaternary all-
ylic alcohols, for those derived from additions to 1-tetral-
ones there are very few examples in the literature.[6] Carb-
inol 7 should be easily obtained by addition of a vinylic
organometallic species to 7-methoxy-1-tetralone (1), the key
starting material for most of the syntheses of agomelatine
that have appeared in the literature so far.
Scheme 4. Preparation and reactivity of carbinol 7.
Next, we proceeded with the exploration of the reactivity
of the aforementioned carbinol 7, which could possibly
open additional pathways towards our final target. First,
we examined the prospect of an allylic chlorination, accord-
ing to our retrosynthetic design. Treatment of alcohol 7
with concentrated hydrochloric acid in THF resulted in a
mixture of allylic chlorides 10 and diene 9 in the ratio 7:1
(Scheme 4). We observed that allylic chlorides 10 slowly iso-
merized to chloride 6 upon being stirred in chloroform at
ambient temperature (70% isolated yield for two steps).
Luckily, when the treatment with HCl was performed in
chloroform or dichloromethane, the endocyclic vinyl chlor-
ide 6 was directly obtained in excellent yield (Ͼ 96% iso-
lated yield), while the formation of the undesired diene 9
was minimal (Ͻ 4%).
Diene 9, even though considered an undesired side prod-
uct according to our synthetic design, in general terms,
could be regarded as a useful intermediate in the synthesis
of agomelatine or of structurally related compounds. There-
fore, we explored the possibility of chemoselectively con-
Scheme 3. Proposed retrosynthesis of agomelatine.
For the realization of our synthetic plan, we first exam- verting carbinol 7 to diene 9 and found that using catalytic
ined the addition of a vinyl Grignard reagent to tetralone amounts of I2 in the presence of quinoline provides the eli-
1. We were surprised to find out that, when vinylmagnesium mination product in 87% yield (Scheme 4).[7b,8] Efforts to
chloride was employed, the major product of the reaction oxidize compound 9 to the corresponding aromatic deriva-
was the reduction of tetralone 1 to the corresponding tive 11 using DDQ or Pd catalysis in the presence of allyl
alcohol 8 (Scheme 4), while the expected addition product methacrylate, as the hydrogen acceptor, were unsuccessful,
7 was only formed in 15–20% yield. To our delight, this mainly because of the high propensity of diene 9 to partici-
trend was completely reversed by simply changing the orga- pate in Diels–Alder reactions.[9] To our delight, direct treat-
nometallic component to vinylmagnesium bromide.[7] After ment of carbinol 7 with DDQ afforded naphthalene deriva-
much experimentation, the optimum conditions were found tive 11 (60% yield; Scheme 4), a known intermediate in the
to be the slow addition of a THF solution of tetralone 1 synthesis of agomelatine analogues, which is currently ac-
(1.0 m) to vinylmagnesium bromide (0.8–1.0 m in THF) at cessed by much more laborious routes.[10]
25 °C. Under these conditions, carbinol 7 was formed in
Having demonstrated the high potential of carbinol 7 as
80–85% yield, while the undesired side product 8 was sup- a key intermediate en route to agomelatine and related
pressed to less than 4%. The process was performed suc- chemical entities, we proceeded with the realization of our
cessfully with 200 g of tetralone 1.
retrosynthetic plan. The next step involved the oxidation
Eur. J. Org. Chem. 2014, 6376–6379
© 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.eurjoc.org
6377

C. I. Stathakis, E. Neokosmidis, T. V. Koftis
SHORT COMMUNICATION
of chloride 6 to the corresponding aromatic derivative 12
With the access to naphthalene chloride 12 secured, we
(Table 1), which was initially attempted by using Pd in the moved on with the introduction of the amine group. Most
presence of a hydrogen acceptor.[11] In accordance with pre- processes appearing in the literature involve a two-step pro-
vious reports by the inventors at Servier laboratories, we cedure, such as the Gabriel amine synthesis using potassium
faced difficulties to drive the reaction to completion even phthalimide and hydrolysis,[14a] azide substitution, and sub-
with prolonged heating (Ͼ 48 h) at temperatures as high as sequent reduction[14b] or similar processes using other
120 °C, conditions that are not industrially friendly (entries amino-group surrogates.[14c] We found that, by heating com-
1 and 2). In addition, the progress of the reaction seemed pound 12 in equal volumes of EtOH and aqueous NH3 at
to be highly dependent on the quality of the Pd catalyst 100–105 °C in an autoclave for 6 h, the corresponding am-
used.
monium chloride 13 is obtained in good yield (Scheme 5).
The major side product of the transformation, bis-alkylated
ammonium salt 14, was formed in less than 10% yield and
could effectively be removed from the reaction mixture dur-
ing work-up to provide compound 13 in greater than 99%
purity.
Table 1. Attempts to oxidize chloride 6 to naphthalene chloride 12.
Oxidant Additive
Solvent T [°C], t
Conver-
sion[a]
Entry
[h]
1
2
Pd/C
Pd(OH)
A.M.[b]
A.M.[b]
toluene 120, 48
toluene 120, 48
65%
55%
2
3
4
DDQ
DDQ
–
–
toluene 25, 0.5
CH2Cl2 25, 0.5
100%
100%
NaNO2/
O2
NaNO2/
O2
NaNO2/
O2
NaNO2/
O2
5[c]
6[c]
7[c]
DDQ
toluene 90, 4
0%
E.A.[b]
90, 4
40%
Ͻ 5%
Scheme 5. Preparation of ammonium chloride 13.
DDQ
DDQ
acetone 90, 4
ACN[b]
ACN[b] 90, 4
It is important that the whole sequence for the conver-
sion of tetralone 1 to ammonium salt 13 can be operated
without any intermediate purification. The nonpolar impu-
rities carried over from the initial steps do not affect the
efficacy of the following transformations, thus a simple ex-
traction during the work-up of the ammonolysis reaction
mixture (Scheme 5) is sufficient to provide a high-purity
material in the aqueous phase. By a simple slurry wash of
the solid residue, after evaporation of the volatiles, ammo-
nium chloride 13 is obtained in excellent purity (Ͼ 99%)
and in 58–61% overall yield (four steps from tetralone 1)
as an off-white to light brown solid.
The synthesis of the final API was completed by acetyl-
ation of ammonium salt 13 to the corresponding acetamide
by using Ac2O and AcONa in EtOH heated to reflux
(Scheme 6).[5a] This procedure provides a very clean conver-
sion to crude agomelatine and was applied successfully with
166 g of intermediate 13. After a simple recrystallization,
the pure final API was afforded in greater than 99.5% pu-
8[c]
DDQ
90, 4
100%
9[c]
DDQ
NaNO2
60%
[a] Determined by HPLC analysis of in process reaction aliquots.
[b] A.M.: allyl methacrylate, E.A.: ethyl acetate, ACN: acetonitrile.
[c] 10 mol-% DDQ and 10 mol-% NaNO2 is used. The reaction is
performed in a stainless steel high-pressure reactor.
A solution to the problems described above was found
by using DDQ (2,3-dichloro-5,6-dicyanobenzoquinone) as
the oxidant.[12] The reaction proceeded smoothly at ex-
tremely mild conditions (25 °C, 30 min) in a range of or-
ganic solvents, such as toluene or dichloromethane (entries
3 and 4). However, even though very efficient, this method
raised both environmental and cost effectiveness considera-
tions due to the utilization of stoichiometric amounts of
the genotoxic and expensive DDQ. Therefore, we sought a
catalytic version of this reaction and were pleased to find
out that a modification of the procedure described by Xu
et al. performed perfectly on our substrate.[13] According to
the latter, the catalytic system consisting of 10 mol-% DDQ
and 10 mol-% NaNO2, as the co-oxidant, in the presence
of O2 gas, in toluene as the solvent, could successfully ac-
complish the dehydrogenation of dihydroarenes. In our
hands, the strict implementation of these conditions met
with failure (entry 5). While screening for alternative sol-
vents (entries 6–8), we discovered that a simple change from
toluene to acetonitrile resulted in full conversion of chloride
6 to aromatic compound 12 (entry 8). For the same reaction
time, the conversion was only 60% (entry 9) in the absence
of O2 gas.
Scheme 6. Completion of the synthesis and purification of the final
API.
6378
www.eurjoc.org
© 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Eur. J. Org. Chem. 2014, 6376–6379

A Scalable Synthesis of the Antidepressant Agomelatine
Clin. Psychopharmacol. 2002, 17, 239–247; d) S. Kasper, G.
Hajak, K. Wulff, W. J. Hoogendijk, A. L. Montejo, E. Smer-
aldi, J. K. Rybakowski, M. A. Quera-Salva, A. M. Wirz-Jus-
tice, F. Picarel-Blanchot, F. J. Bayle, J. Clin. Psychiatry 2010,
71, 109–120.
rity. The overall molar yield of the synthetic scheme was
calculated to be 52.3% (72.2% w/w).
Conclusions
[4] More than one hundred patents have emerged from the litera-
ture over the last decade concerning the synthesis and
recrystallization to specific crystalline forms of agomelatine.
[5] a) J.-C. Souvie, I. Gonzales Blanco, G. Thominot, G. Chapuis,
S. Horvath, G. Damien, EP1564202B1, Sept. 10, 2008; b) S.
Zhou, F. Jian, L. Gan, WO2011006387A1, Jan. 20, 2011.
[6] K. W. Hunt, A. W. Cook, R. J. Watts, C. T. Clark, G. Vigers,
D. Smith, A. T. Metcalf, I. W. Gunawardana, M. Burkard,
A. A. Cox, M. K. Geck Do, D. Dutcher, A. A. Thomas, S.
Rana, N. C. Kallan, R. K. DeLisle, J. P. Rizzi, K. Regal, D.
Sammond, R. Groneberg, M. Siu, H. Purkey, J. P. Lyssikatos,
A. Marlow, X. Liu, T. P. Tang, J. Med. Chem. 2013, 56, 3379–
3403.
[7] a) L. F. Silva Jr., R. S. Vasconcelos, M. A. Nogueira, Org. Lett.
2008, 10, 1017–1020; b) D. Soorukram, P. Knochel, Org. Lett.
2007, 9, 1021–1023.
[8] C. Symmes Jr., L. D. Quin, J. Org. Chem. 1979, 44, 1048–1056.
[9] The mixture of diastereomers of the Diels–Alder adducts, com-
pounds A, were isolated from the reaction mixture as the major
products.
In summary, we have described the synthesis of ago-
melatine, an important API with great potential for the
treatment of major depression, by using a widely applicable
key starting material, 7-methoxy-1-tetralone (1). Our syn-
thesis is among the most efficient in terms of overall yield
and cost effectiveness and is well suited for large industrial
production. The experimental conditions are mild and envi-
ronmentally friendly. In addition, common and easily ac-
cessible chemicals are involved, while the most expensive of
those, namely DDQ, was used only in catalytic amounts.
Efforts for further optimization of the synthetic process are
currently underway.
Supporting Information (see footnote on the first page of this arti-
cle): Full experimental details for the preparation and copies of the
1H and 13C NMR spectra of compounds 6, 7, 12, 13, and synthetic
agomelatine are provided.
Acknowledgments
The authors are grateful to Dr. Thanos Andreou for fruitful dis-
cussions, especially concerning the catalytic oxidation of chloride 6
to the corresponding aromatic chloride 12.
[10] S. Yous, M. Ettaoussi, A. Sabaouni, P. Berthelot, M. Spedding,
P.
Delagrange,
D.-H.
Caignard,
M.
Millan,
[1] a) R. J. Reiter, Endocr. Rev. 1991, 12, 151–180; b) I. Nir, Recept.
Channels 2003, 9, 379–385; c) Y. Touitou, Exp. Gerontol. 2001,
36, 1083–1100.
WO 2009022063A3, Feb. 19, 2009.
[11] J.-C. Souvie, I. Gonzales Blanco, EP 1564205 B1, Feb. 11,
2005.
[2] a) J. Andreux, R. Houssin, S. Yous, B. Guardiola, D. Lesieur,
EP0447285B1, May 12, 1993; b) S. Yous, J. Andrieux, H. E.
Howell, P. Depreux, P. Morgan, P. Renard, B. Pfeiffer, D. Le-
sieur, B. Guardiola-Lemaitre, J. Med. Chem. 1992, 35, 1484–
1486; c) P. Depreux, D. Lesieur, H. A. Mansour, P. Morgan,
H. E. Howell, P. Renard, D.-H. Caignard, B. Pfeiffer, P. De-
lagrange, B. Guardiola, S. Yous, A. Demarque, G. Adam, J.
Andrieux, J. Med. Chem. 1994, 37, 3231–3239.
[3] a) A. Hale, R. M. Corral, C. Mencacci, J. S. Ruiz, C. A. Severo,
V. Gentil, Int. Clin. Psychopharmacol. 2010, 25, 305–314; b) P.
Lemoine, C. Guilleminault, E. Alvarez, J. Clin. Psychiatry
2007, 68, 1723–1730; c) H. Loo, A. Hale, H. D’Haenen, Int.
[12] D. R. Buckle, S. J. Collier, M. D. McLaws “2,3-Dichloro-5,6-
dicyano-1,4-benzoquinone” in e-EROS Encyclopedia of Rea-
gents for Organic Synthesis 2005.
[13] W. Zhang, H. Ma, L. Zhou, Z. Sun, Z. Du, H. Miao, J. Xu,
Molecules 2008, 13, 3236–3245.
[14] a) G. Zhang, D. Chen, Y. Ma, X. Yang, S. Zhou, L. Chen,
WO2010012208A1, Jul. 23, 2009; b) Z. Xianyi, Y. Jianbin, G.
Hongjun, C. Daqing, CN102531956A, Dec. 21, 2010; c) A. Ba-
rozza, M. Veronese, J. Roletto, P. Paissoni, EP2703383, Aug.
02, 2013.
Received: July 8, 2014
Published Online: September 8, 2014
Eur. J. Org. Chem. 2014, 6376–6379
© 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.eurjoc.org
6379