Concise synthesis of the antidepressive drug candidate GSK1360707 by a highly enantioselective gold-catalyzed enyne cycloisomerization reaction

Article abstract of DOI:10.1002/chem.201101346

Out of depression: Indeed, no depressing results are obtained from a gold-catalyzed enyne cycloisomerization controlled by phosphoramidite ligands with TADDOL-related but acyclic backbones (see scheme; Cbz=benzyloxycarbonyl). The "triple-reuptake inhibito

Full text of DOI:10.1002/chem.201101346

DOI: 10.1002/chem.201101346
Concise Synthesis of the Antidepressive Drug Candidate GSK1360707 by a
Highly Enantioselective Gold-Catalyzed Enyne Cycloisomerization ACTHUNTGRENNGRU eaction
Henrik Teller and Alois Fꢀrstner*[a]
Depression is a serious chronic illness that exacts a signifi-
cant toll on health and productivity in the entire western
world. Although various lines of treatment are established,
medication with antidepressive agents is most common,
which usually act by increasing the levels of the excitatory
neurotransmitters serotonin, noradrenaline, and dopamine.[1]
This can be achieved by down-regulating either the metabo-
lization of these key aminergic messengers or their reuptake
from the synaptic cleft by the presynaptic neurons. Whereas
most of the approved drugs interfere with the serotonin re-
uptake, more advanced medication attempts to interfere
with two or preferably even all three of these crucial neuro-
transmitters simultaneously to better address the anhedonic
symptoms of depression and/or shorten the time to onset of
clinical efficiency.[1] GSK1360707 represents a particularly
promising “triple-reuptake inhibitor”, which was recently
progressed into clinical trials because it had shown promis-
ing bioavailability and pharmacokinetics, together with a
well-balanced activity profile in animal experiments.[2]
So far, three different syntheses of GSK1360707 and salts
thereof were disclosed in patents and/or the open literature
(Scheme 1). The lengthy original approach (route A, 13
steps total, 11 linear steps, 5% overall yield) was deemed in-
appropriate for scale up as it required the use of diazomalo-
Scheme 1. Published syntheses of the triple-reuptake inhibitor
nate and comprised a low-yielding resolution as the final
step.[3] A much improved second generation synthesis fol-
lows a significantly shorter sequence (Scheme 1, route B, 5
steps, 21% overall yield), which could be performed on a
pilot plant scale to produce multikilogram amounts of the
drug candidate.[4] However, high cost of goods, not least for
the advanced starting material, a fairly low atom economy,
the troublesome production of polychlorinated biphenyl de-
rivatives as byproducts in the necessary Suzuki reaction, and
the need for a chromatographic separation of the enantio-
mers on a chiral stationary phase in the penultimate step
are still far from ideal.[4] Therefore, yet another synthesis
was pursued, which was based on the formation of the tricy-
GSK1360707 ((ꢀ)-1), all of which require a resolution step; Ms=meth-
AHCTUNGERTGaNNUN nesulfonyl; Boc=tert-butyloxycarbonyl; Ns=2-nitrophenylsulfonyl.
clic skeleton by a noble-metal-catalyzed enyne cycloisomeri-
zation (route C).[5] In adaptation of chemistry originally de-
scribed by our group,[6] the development team managed to
form the key intermediate 2 from substrate 3 in excellent
yield with the aid of PtCl2 (3.5 mol%) as the catalyst (7
steps, 45(58)% overall yield).[7] PtCl2 could also be replaced
by gold-based catalysts, but attempts to perform the rear-
rangement asymmetrically with enantiopure LAu+ tem-
plates were plagued by low enantiomeric excesses (ee), de-
spite the considerable number of well-established chiral li-
gands (L) that were tested. The best results were obtained
with a combination of AuCl·SMe2 (10 mol%), (R)-tol-
BINAP (5 mol%) and AgBF4 (10 mol%), which afforded
compound 2 in 59% ee.[5] Prompted by this recent disclo-
sure,[5] we now present our approach to the triple-reuptake
inhibitor GSK1360707, which is not only even shorter and
significantly more efficient, but also provides the desired
product in excellent optical purity. It attests to the power
[a] H. Teller, Prof. A. Fꢀrstner
Max-Planck-Institut fꢀr Kohlenforschung
45470 Mꢀlheim/Ruhr (Germany)
Fax : (+49)208 306 2994
E-mail: fuerstner@kofo.mpg.de
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/chem.201101346.
7764
ꢁ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2011, 17, 7764 – 7767

COMMUNICATION
and the flexibility of the ligand-design concept for asymmet-
ric gold catalysis recently disclosed by this laboratory.[8]
Our approach capitalized on the simple notion that re-
placement of the N-nosyl (Ns) group in 2[5] by a benzyloxy-
carbonyl (Cbz) moiety might streamline the end part of the
synthesis. Whereas compound 2 had to be transformed into
the target in a stepwise manner using Et3SiH and trifluoro-
acetic acid to reduce the enamide followed by cleavage of
the nosyl group with thioglycolic acid in basic medium,[5] we
envisaged a straightforward catalytic hydrogenation/hydro-
genolysis step for the release of the final product in a single
operation. To this end, propargyl amine was acylated with
CbzCl, followed by N-alkylation of the resulting compound
8 with the substituted allyl chloride 5 (Scheme 2).[9] As ex-
pected, the subsequent Sonogashira coupling with 1,2-di-
chloro-4-iodobenzene 4 under standard conditions also pro-
ceeded smoothly[10] to give enyne 10 as the required cycliza-
tion precursor in excellent yield.
by crafting chiral pockets with an effective threefold symme-
try.[8] The environment in the resulting binding site is degen-
erate upon rotation of L and should therefore translate into
good enantioselectivity. In fact, readily accessible phosphor-
amidite ligands,[13] preferably those comprising 2,3-dime-
thoxy-1,1,4,4-tetraarylbutane-1,4-diol entities as TADDOL-
related units[14] with an acyclic backbone, were shown to
form such C3-symmetric ligand spheres about a AuI center
and turned out to impart excellent levels of asymmetric in-
duction onto [2+2] and [4+2] cycloadditions of ene–allene
substrates.[8]
At first, however, applications of these ligands to the cy-
cloisomerization of enyne 10 met with limited success.
Under the standard conditions of our earlier studies, product
14 was obtained with only moderate ee values independent
of the chosen phosphoramidite.[15] Even L3, which had been
the most effective ligand in the ene–allene series,[8] gave
compound 14 with no more than 44% ee (entry 3). Gratify-
ingly though, it was found that changing the solvent from
the previously used CH2Cl2 to toluene resulted in a dramatic
improvement, with 14 now being formed with 90% ee. This
favorable outcome stands in marked contrast to the report
by the Glaxo chemical development group, which did not
find any significant effects on the outcome of their reaction
when changing the solvent, the counterion, or the tempera-
ture.[5] In our case, however, the
Asymmetric gold catalysis is particularly exigent because
AuI strongly favors linear dicoordination.[11,12] This handicap
forces the substrate to approach the active center trans to
the chosen chiral ligand L, which itself can be attached to
the metal center only by single-point binding and hence is
ꢀ
able to rotate freely about the L Au bond. In a previous
communication we showed that these challenges can be met
positive effect of toluene
turned out to be general as can
be seen from the data compiled
in Table 1. Even more surpris-
ing is the observation that the
switch from CH2Cl2 to toluene
even inverted the sense of in-
duction when the parent
TADDOL ligand L1 was
used.[16]
Previous crystallographic in-
formation had suggested that
the C3-symmetric chiral pocket
grafted by a phosphoramidite
with a TADDOL backbone car-
rying unsubstituted phenyl rings
(Ar=Ph) would not reach over
the big AuI cation by much. To
extend the binding site and
ensure better chiral recognition,
tert-butyl groups had been at-
tached to the 4-position of
these decisive substituents as
shown in ligands L2 and L3.[8]
An obvious alternative is the
annulation of a second ring;
this design is reduced to prac-
Scheme 2. Enantioselective synthesis of GSK1360707: a) CbzCl, NaHCO3, EtOH/H2O (1:1), 08C!RT, quant;
b) Compound 5, NaH, THF/DMF (1:1), 08C!RT, 91%; c) Compound 4, [(Ph3P)2PdCl2] (2.5 mol%), CuI (5
mol%), Et3N, DMF, 95%; d) [LAuCl] cat., AgBF4 cat., see Table 1; e) Pd black (2.5 mol%), H2 (1 atm),
Na2CO3, EtOAc/MeOH (1:1), 91%; f) HCl in ether (1m), Et2O, quant; g) [(PhO)3PAuCl] (5.5 mol%), AgBF4
tice in ligands L4–L7 carrying
2-naphthyl rather than (4-tert-
butyl)phenyl moieties. All of
(5 mol%), MeOH/CH2Cl2 (1:1), 49% (95% based on recovered starting material). Cbz=benzyloxycarbonyl. them performed exceedingly
Chem. Eur. J. 2011, 17, 7764 – 7767
ꢁ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
7765

A. Fꢀrstner et al.
Table 1. Enantioselective cycloisomerization of enyne 10 to product (+)-14 (indicated by “+”) or its enantio-
mer (indicated by “ꢀ”) using complexes of the general type [LAuCl] as precatalysts.[a]
ly, both antipodes of 1-phenyl-
AHCTUNGTERGeNNUN thylamine afforded the same
Entry
Ligand
(L)
L·AuCl
A
ee [%]
enantiomer of 14 with apprecia-
ble levels of induction (Table 1,
compare entries 4 to 5 and 5 to
6), indicating that the match/
CH2Cl2
toluene
1
L1
5.5
5.5
5.5
5.5
5.5
+48
ꢀ26
mismatch
between
the
TADDOL and amine plays a
minor role. (S,S)-2,6-Diphenyl-
piperidine[18] gave essentially
the same outcome as its acyclic
congener bis(S)-1-phenylethyl-
2
3
4
5
L2
L3
L4
L5
ꢀ23
ꢀ44
ꢀ48
AHCTUNGTERGaNNUN mine (compare entries 6 to 8
and 7 to 9). For practical rea-
sons, however, ligand L6 was
preferred because bis(S)-1-phe-
nylethylamine is commercially
available. With 2.75 mol% of
the readily prepared, air stable
and hence very practical com-
plex L6·AuCl in combination
with 2.5 mol% of AgBF4 in tol-
uene at 08C,[19] we were able to
obtain the desired cycloisomeri-
zation product (+)-14 in 88%
yield with a well reproducible
ee value of 95%.
ꢀ90
ꢀ84
+90
6
7
L6
L6
5.5
2.75
+93
+95[b,c,d]
When the reaction was per-
formed in MeOH as co-solvent,
the rather unstable tertiary
ether 15 was obtained as the
only product. We tentatively in-
terpret this result as an indica-
tion for the “non-classical”
character of the reactive inter-
mediate primarily formed in
8
9
L7
L7
5.5
5.5
+53
+94
+95[b]
[a] Unless indicated otherwise, all reactions were performed at ambient temperature using AgBF4 (5 mol%)
as co-catalyst; the conversion was quantitative in all cases investigated. [b] At 08C. [c] With only 2.5 mol% of
AgBF4 as co-catalyst. [d] Product 14 was isolated in 88% yield on a 1 mmol scale.
noble-metal-catalyzed
cycloisomerization
enyne
reac-
well in the current application, with L6 and L7 both giving
the desired product 14 in up to 95% ee. In this context it
should be noted, that the level of induction is only marginal-
ly improved upon lowering the temperature (Table 1, com-
pare entries 6 to 7 and 8 to 9), whereas the reaction rate is
decreased. It is also noteworthy that replacement of AgBF4
by non-hygroscopic AgNTf2 shut down the reaction (<5%
conversion after 24 h, as determined by NMR analysis).
In line with the rationale previously outlined,[8] the data
compiled in Table 1 shows that all TADDOLꢂs with an acy-
clic backbone performed significantly better than their tradi-
tional analogues carrying an acetal moiety. This observation
is true for all cases in which direct comparisons can be made
and is independent of the chosen solvent (Table 1, compare
entries 2 to 3 and 4 to 6).
tions.[6,20] For gold in particular, it has been computed that
the putative cyclopropyl carbenes are actually highly distort-
ed and largely resemble the canonical form of a gold-stabi-
lized homoallyl cation;[21] this particular resonance extreme
12 is then intercepted by the external nucleophile at its terti-
ary cation site, at which the charge density is most stabi-
lized.[22] It is interesting to note, however, that 12 has an
achiral scaffold, whereas the bond formation that converts
12 into 13 is enantio-determining. This may help rationalize
the high enantioselectivity observed with our system, be-
cause induction occurs only after the substrate has been co-
valently linked to the gold template and hence deeply im-
mersed into the chiral pocket.[23]
Finally, the choice of the N-Cbz group paid dividends;
stirring a solution of enamide 14 in a mixed solvent system
(EtOAc/MeOH, 1:1) under an atmosphere of hydrogen in
the presence of palladium black and Na2CO3, ensured the
deprotection and saturation of the double bond in a single
operation.[24] Under these conditions, no signs of cyclopro-
Another striking result pertains to the fact that the abso-
lute configuration of 14 is determined solely by the stereo-
chemistry of the TADDOL unit of the ligand, whereas the
configuration of the amine part is not decisive.[17] Specifical-
7766
www.chemeurj.org
ꢁ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2011, 17, 7764 – 7767

Concise Synthesis of the Antidepressive Drug GSK1360707
COMMUNICATION
[10] K. Sonogashira, J. Organomet. Chem. 2002, 653, 46–49.
pane opening or reductive cleavage of the chlorine substitu-
ents were noticed, and (ꢀ)-1 was obtained in a spectroscopi-
cally and analytically pure form after simple extraction; as
expected, the ee value was fully preserved (as determined
by HPLC analysis). However, since the free amine is a
liquid, it was preferable to convert the crude material into
an appropriate salt such as the crystalline hydrochloride
(ꢀ)-1·HCl;[25] recrystallization then allows trace amounts of
metal impurities to be removed, which are strictly regulated
in any drug candidate. Overall, the route to the triple-reup-
take inhibitor GSK1360707 presented herein requires no
more than 5 operations, provides this valuable antidepres-
sive agent with 95% optical purity in an overall yield of no
less than 69%, and hence clearly surpasses all known routes
in terms of efficiency and stereoselectivity. We are currently
further extending the scope of the chiral gold catalysts de-
scribed herein and will report our results in due course.
[11] For pertinent reviews on asymmetric gold catalysis, see: a) A.
Pradal, P. Y. Toullec, V. Michelet, Synthesis 2011, 1501–1514; b) N.
Bongers, N. Krause, Angew. Chem. 2008, 120, 2208–2211; Angew.
Chem. Int. Ed. 2008, 47, 2178–2181; c) R. A. Widenhoefer, Chem.
Eur. J. 2008, 14, 5382–5391.
[12] a) D. J. Gorin, F. D. Toste, Nature 2007, 446, 395–403; b) A. Fꢀrst-
ner, P. W. Davies, Angew. Chem. 2007, 119, 3478–3519; Angew.
Chem. Int. Ed. 2007, 46, 3410–3449; c) E. Jimꢃnez-NfflÇez, A. M.
Echavarren, Chem. Rev. 2008, 108, 3326–3350; d) A. S. K. Hashmi,
Chem. Rev. 2007, 107, 3180–3211; e) Y. Yamamoto, J. Org. Chem.
2007, 72, 7817–7831; f) V. Michelet, P. Y. Toullec, J.-P. GenÞt,
Angew. Chem. 2008, 120, 4338–4386; Angew. Chem. Int. Ed. 2008,
47, 4268–4315; g) Z. Li, C. Brouwer, C. He, Chem. Rev. 2008, 108,
3239–3265; h) A. Arcadi, Chem. Rev. 2008, 108, 3266–3325; i) A. R.
Chianese, S. J. Lee, M. R. Gagnꢃ, Angew. Chem. 2007, 119, 4118–
4136; Angew. Chem. Int. Ed. 2007, 46, 4042–4059.
[13] J. F. Teichert, B. L. Feringa, Angew. Chem. 2010, 122, 2538–2582;
Angew. Chem. Int. Ed. 2010, 49, 2486–2528.
[14] Review: D. Seebach, A. K. Beck, A. Heckel, Angew. Chem. 2001,
113, 96–142; Angew. Chem. Int. Ed. 2001, 40, 92–138.
[15] Phosphoramidites incorporating BINOL- or 3,3’-disubstituted
BINOL moieties instead of the TADDOL backbones gave even
lower ee values in the cycloisomerization of enyne 10; details will be
disclosed in a future full paper. For successful applications of such li-
gands in asymmetric gold catalysis, see: a) I. Alonso, B. Trillo, F.
López, S. Montserrat, G. Ujaque, L. Castedo, A. Lledós, J. L. Mas-
careÇas, J. Am. Chem. Soc. 2009, 131, 13020–13030; b) A. Z. Gon-
zµlez, F. D. Toste, Org. Lett. 2010, 12, 200–203; c) A. Z. Gonzµlez,
D. Benitez, E. Tkatchouk, W. A. Goddard, F. D. Toste, J. Am.
Chem. Soc. 2011, 133, 5500–5507.
Acknowledgement
Generous financial support by the MPG and the Fonds der Chemischen
Industrie (Kekulꢃ stipend to H. T.) is gratefully acknowledged. We thank
all analytical departments of our Institute for excellent support, and R.
Leichtweiß for many ee value determinations.
[16] Precedent for such changes in the sense of induction by changing
the solvent is discussed in the following reviews and literature cited
therein: a) M. Bartók, Chem. Rev. 2010, 110, 1663–1705; b) T.
Tanaka, M. Hayashi, Synthesis 2008, 3361–3376.
Keywords: asymmetric catalysis · cycloisomerization · drug
design · gold · phosphoramidites
[17] The presence of two peripheral aryl wings in the amine moiety, how-
ever, is important for obtaining high induction. For example, re-
placement of the 2,6-diphenylpiperidine unit in L7 by unsubstituted
piperidine led to only 50% ee.
[18] Prepared according to: D. J. Aldous, M. D. Dutton, P. G. Steel, Tet-
rahedron: Asymmetry 2000, 11, 2455–2462. One should note, how-
ever, that the wrong enantiomer of the diphenylprolinol ligand used
in the borane reduction is depicted in Scheme 3 and Table 1 of the
above publication.
[19] It is important to avoid any excess of AgBF4 as this may result in
decomposition of the chiral cationic catalyst and give racemic back-
ground reactions. This is ensured by using 2.75 mol% of [LAuCl]
but only 2.5 mol% of AgBF4.
[20] For a general discussion, see: A. Fꢀrstner, Chem. Soc. Rev. 2009, 38,
3208–3221.
[1] a) M.-L. Wong, J. Licinio, Nat. Rev. Drug Discov. 2004, 3, 136–151;
b) C. B. Nemeroff, J. Psychiatr. Res. 2007, 41, 189–206; c) R. H. Bel-
maker, G. Agam, N. Engl. J. Med. 2008, 358, 55–68; d) M. Schubert-
Zsilavecz, H. Stark, Pharm. Unserer Zeit 2004, 33, 282–287 and ref-
erences cited therein.
[2] F. Micheli, P. Cavanni, D. Andreotti, R. Arban, R. Benedetti, B.
Bertani, M. Bettati, L. Bettelini, G. Bonanomi, S. Braggio, R. Car-
letta, A. Checchia, M. Corsi, E. Fazzolari, S. Fontana, C. Marchioro,
E. Merlo-Pich, M. Negri, B. Oliosi, E. Ratti, K. D. Read, M. Roscic,
I. Sartori, S. Spada, G. Tedesco, L. Tarsi, S. Terreni, F. Visentini, A.
Zocchi, L. Zonzini, R. Di Fabio, J. Med. Chem. 2010, 53, 4989–5001.
[3] B. Bertani, R. Di Fabio, F. Micheli, G. Tedesco, S. Terreni (Glaxo
Group), WO 2008031772A1, 2008.
[4] V. I. Elitzin, K. A. Harvey, H. Kim, M. Salmons, M. J. Sharp, E. A.
Tabet, M. A. Toczko, Org. Process Res. Dev. 2010, 14, 912–917.
[5] N. M. Deschamps, V. I. Elitzin, B. Liu, M. B. Mitchell, M. J. Sharp,
E. A. Tabet, J. Org. Chem. 2011, 76, 712–715.
[21] C. Nieto-Oberhuber, M. P. MuÇoz, E. BuÇuel, C. Nevado, D. J.
Cµrdenas, A. M. Echavarren, Angew. Chem. 2004, 116, 2456–2460;
Angew. Chem. Int. Ed. 2004, 43, 2402–2406.
[22] For a related study, which also strongly advocated the highly cation-
ic character of the reactive intermediates in gold-catalyzed enyne cy-
cloisomerizations by interception with tethered nucleophiles, see: A.
Fꢀrstner, L. Morency, Angew. Chem. 2008, 120, 5108–5111; Angew.
Chem. Int. Ed. 2008, 47, 5030–5033.
[23] For a detailed discussion on strereochemical issues related to gold-
stabilized “non-classical” cations, see: D. Garayalde, E. Gómez-
Bengoa, X. Huang, A. Goeke, C. Nevado, J. Am. Chem. Soc. 2010,
132, 4720–4730.
[24] Inspection of the NMR spectra shows that the Cbz group is quickly
cleaved; the hydrogenation of the resulting imine required stirring
overnight at ambient temperature.
[25] Various other salts of 1 are also known in the literature, see ref. [3]–
[5].
[6] a) A. Fꢀrstner, H. Szillat, F. Stelzer, J. Am. Chem. Soc. 2000, 122,
6785–6786; b) A. Fꢀrstner, F. Stelzer, H. Szillat, J. Am. Chem. Soc.
2001, 123, 11863–11869; c) A. Fꢀrstner, H. Szillat, B. Gabor, R.
Mynott, J. Am. Chem. Soc. 1998, 120, 8305–8314.
[7] The yields given in Scheme 1 and in the Experimental Part of ref.
[5] are not consistent; based on the yield reported in the described
procedures, we calculate an overall yield of only 45% rather than
58% as stated by the authors.
[8] H. Teller, S. Flꢀgge, R. Goddard, A. Fꢀrstner, Angew. Chem. 2010,
122, 1993–1997; Angew. Chem. Int. Ed. 2010, 49, 1949–1953.
[9] Since the literature procedure for the preparation of 5 is rather inef-
ficient (R. M. Wilson, R. K. Thalji, R. G. Bergman, J. A. Ellman,
Org. Lett. 2006, 8, 1745–1747), a modified protocol was developed
that gave multigram amounts of 5 starting from cheap 3-chloro-2-
(chloromethyl)-1-propene and NaOMe; for details, see the Support-
ing Information.
Received: May 3, 2011
Published online: May 26, 2011
Chem. Eur. J. 2011, 17, 7764 – 7767
ꢁ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
7767