An efficient and scalable synthesis of the spirocyclic glycine transporter inhibitor GSK2137305

Article abstract of DOI:10.1021/op100210s

An efficient and scalable synthesis of a glycine transporter inhibitor is presented. The key steps in the synthetic sequence are the formation of a spirocyclic imidazolidinone from an -amino nitrile and a cyclic ketone and an arylation of 4-methyl imidazo

Full text of DOI:10.1021/op100210s

Organic Process Research & Development 2011, 15, 44–48
An Efficient and Scalable Synthesis of the Spirocyclic Glycine Transporter Inhibitor
GSK2137305
Jonathan P. Graham,* Nathalie Langlade, John M. Northall, Alastair J. Roberts, and Andrew J. Whitehead
Chemical DeVelopment, GlaxoSmithKline R&D, Gunnels Wood Road, SteVenage, Hertfordshire, SG1 2NY, U.K.
Abstract:
condensation of amino nitriles with ketones to generate imi-
dazolidinone heterocycles,3 which encouraged us to evaluate
this approach further.
An efficient and scalable synthesis of a glycine transporter inhibitor
is presented. The key steps in the synthetic sequence are the
formation of a spirocyclic imidazolidinone from an r-amino nitrile
and a cyclic ketone and an arylation of 4-methyl imidazole under
‘ligandless’ Ullmann coupling conditions.
Hence, employing standard Strecker conditions the desired
amino nitrile 12 was formed from 4-bromobenzaldehyde (3)
in ca. 4 h at ambient temperature.4 The product was then
extracted into tert-butylmethyl ether (TBME) and washed with
water to remove all cyanide waste. Since the reaction and waste
streams remain basic throughout, the potential for hydrogen
cyanide formation is avoided. Subjection of purified 12 to the
literature conditions for the cyclisation, a neat reaction at
elevated temperature with catalytic sodium methoxide,3a did
generate imidazolidinone 7, albeit as a minor component in the
reaction mixture (Table 1, entry 1). Encouraged by this
preliminary result, we sought to rapidly optimize the reaction
conditions. By running the reaction in methanol a significantly
cleaner reaction profile was obtained, with imidazolidinone 7
now observed as the major component (entry 2). Crystallisation
occurred upon cooling the methanol solution to 0-5 °C, with
7 isolated in 14% yield and with high purity. A similar reaction
profile was obtained with ethanol, and the lower solubility of 7
in ethanol led to a higher isolated yield (entry 3). The use of
stoichiometric base offered no advantage (entry 4), and little
or no desired product was observed in the absence of base or
with sodium acetate or a range of organic bases (entries 5-9).
A further improvement in both reaction profile and recovery
was obtained on switching to 1-butanol, with little difference
in reaction profile when run at either 90 or 120 °C (entries 10
and 11), but other solvents screened proved less effective (entries
12-16).
Introduction
GSK2137305 (1, Figure 1) is a potent and selective Glycine
transporter type-1 (GlyT1) inhibitor and, thus, has potential for
treating neurological and neuropsychiatric disorders.1 A key
structural feature of GSK2137305 is the spirocyclic imidazalone
core.
The initial route for the synthesis of GSK2137305, used to
prepare the first ca. 50 g of drug substance, is depicted in
Scheme 1.2 A number of factors made this approach unsuitable
as the long-term supply route capable of delivering substantial
quantities of 1. For example, the eight-step synthesis provided
crude 1 in <5% yield and had to be further purified by
preparative HPLC to remove a regioisomeric impurity 2, which
originates from the copper-catalysed cross-coupling of aryl
bromide 8 with 4-methyl imidazole. Furthermore, safety
concerns due to the evolution of hydrogen cyanide during the
Bucherer-Bergs reaction to form hydantoin 4 precluded the
rapid operation of this chemistry in large scale equipment. We
herein report the identification and development of a new,
concise route to 1 that enabled its safe and rapid manufacture
on a multikilogram scale.
To allow efficient processing on a plant scale, it was desirable
to telescope the Strecker reaction with the imidazolidinone
formation, since amino nitrile 12 proved to be difficult to isolate
and showed limited stability. Hence, after aqueous washes of
the TBME solution of 12, the reaction was dried by azeotropic
distillation and concentrated to ca. 2 volumes by vacuum
distillation. Although the imidazolidinone formation reaction
Results and Discussion
It was postulated that it might be possible to go directly to
the spirocyclic imidazolidinone 7 by conducting a Strecker
reaction with 4-bromobenzaldehyde (3) and condensing the
resulting amino nitrile 12 with cyclopentanone (Scheme 2).
Initial reaction of the amine 12 with cyclopentanone would form
the hemiaminal 13, which could then undergo an intramolecular
Pinner reaction to form the cyclic imidate 14. Base catalysed
rearrangement of 14 would then provide the desired imidazo-
lidinone 7. Indeed, there is limited literature precedent for the
* To whom correspondence should be addressed. Telephone: +44 (0)1438
763688. Fax: +44 (0)1438 764869. E-mail: jonathan.p.graham@gsk.com.
(1) (a) Blunt, R.; Cooper, D. G.; Porter, R. A.; Wyman, P. A. Patent
Application WO 2009/034061 A1, 2009. (b) For a recent review on
inhibitors of GlyT1 for the treatment of schizophrenia, see: Bridges,
T. M.; Williams, R.; Lindsley, C. W. Curr. Opin. Mol. Ther. 2008,
10, 591.
(2) The chemistry as detailed in ref 1a was adapted by Simone Spada
and Roberto Profeta at GSK Verona for the preparation of the first
50 g of GSK2137305.
Figure 1. GSK2137305 (1).
44
•
Vol. 15, No. 1, 2011 / Organic Process Research & Development
10.1021/op100210s  2011 American Chemical Society
Published on Web 11/10/2010

Scheme 1. Initial route to GSK2137305a
a Reagents and conditions: (a) KCN, (NH4)2CO3, EtOH, H2O, 68%; (b) H2SO4, 81%; (c) SOCl2, MeOH, 91%; (d) NH3 (aq), 66%; (e) cyclopentanone, H-Y zeolite,
methanol, 63%; (f) NBS, CH2Cl2, 56%; (g) 4-methyl imidazole (6 equiv), CuI (0.6 equiv), L-histidine (1.2 equiv), K2CO3, DMSO, 50%; (h) KOtBu, DMF, 81%; (i)
preparative HPLC, ca. 65%.
Scheme 2. Postulated preparation of imidazolidinone 7
Table 1. Solvent and base screen for reaction of amino
nitrile 12 with cyclopentanone
Entrya
Solvent
Base
Productb (%)
c
1
-
NaOMe
NaOMe
NaOEt
NaOEtd
-
6
2
MeOH
EtOH
50 (14)
54 (32)
50
3
4
EtOH
5
EtOH
<5
6
EtOH
NaOAc
Et3N
<5
7
EtOH
<5
8
EtOH
Pyridine
DBU
<5
9
EtOH
14
10
11
12
13
14
15
16
1-BuOH
1-BuOHe
2-BuOH
EtOAc
Toluene
THF
NaOEt
NaOEt
NaOEt
NaOEt
NaOEt
NaOEt
NaOEt
63 (45)
60
25
<5
<5
<5
Acetonitrile
<5
a Reactions run in Radley’s carousel tubes as 0.5 M solutions of amino nitrile
12, with 2.4 equiv of cyclopentanone and 0.1 equiv of base, and heated with a
block temperature of 90 °C for 6-12 h. b Percentage a/a 7 in the reaction mixture,
determined by HPLC. Isolated yields in parentheses. c Reaction performed neat,
with
temperature.
5
equiv of cyclopentanone. d 1.1 equiv of base used. e 120 °C block
to transfer to plant the process was investigated by reaction
calorimetry, with the exhaust gas monitored by mass spectros-
copy due to the potential hazard of hydrogen cyanide evolution.
We were not able to establish the absence of hydrogen cyanide
during all of the process due to fragmentation of TBME in the
mass spectrometer (a fragment with mass 27 masks that of
HCN). However, the ratio between the main TBME signal and
the fragment remained constant and no gas was evolved so we
could infer that little or no hydrogen cyanide was generated.
Hence, the construction of the aryl imidazolidinone spirocycle
7 from cheap, readily available starting materials employing
an operationally simple and safe process was readily achieved.
Although the yield was modest, it represented approximately a
doubling of the yield when compared to the initial route and
tolerated the carryover of TBME, it did slow the reaction, so
upon dilution with 1-BuOH and treatment with cyclopentanone
and sodium ethoxide solution the reaction typically took about
12 h to reach completion at 80 °C. Upon cooling the reaction
the desired product crystallised, providing imidazolidinone 7
in 35-40% isolated yield from 4-bromobenzaldehyde. Prior
(3) (a) David, A. C.; Levy, A. L. J. Chem. Soc. 1951, 3479. (b) An
imidazolidinone impurity was observed from the reaction of 1-amino-
1-cyanocyclohexane in DMF under basic conditions, presumably
formed by cyclisation/rearrangement with cyclohexanone, formed
under the reaction conditions by adventitious water: Wendelin, W.;
Kern, W.; Zmo¨lnig, I.; Schramm, H. Monatsh. Chem. 1981, 1091.
(4) Addition of ethanol to the procedure of Gaertner, M. Photochem.
Photobiol. Sci. 2007, 159. avoided material forming a gum in the
reaction without adversely affecting the reaction profile.
Vol. 15, No. 1, 2011 / Organic Process Research & Development
•
45

Scheme 3. New route to GSK2137305a
a Reagents and conditions: (a) (i) NaCN, aq. NH3, NH4OAc, EtOH, 25 °C, 4 h; (ii) cyclopentanone, 1-BuOH, NaOEt, 80 °C, 35-40%; (b) DDQ, EtOAc, 60 °C,
80-85%; (c) 4-methyl imidazole, CuI, K2CO3, DMSO, 130 °C, 50-55%; (d) (i) NMP, aq. KOH; (ii) 3-aminobenzotrifluoride, chloroacetyl chloride, NMP, 75-80%.
significantly shortened the supply time for the preparation of
the key intermediate 7.
trivial. Additionally, a ca. 4:1 mixture of the 4- and 5-methyl
imidazole regioisomers (9 and 10, respectively) was formed in
the reaction, a ratio similar to other known examples of the
coupling of 4-methyl imidazole with aryl bromides that do not
possess ortho-substituents,6,7 with the 4-methyl isomer formed
as the major product due to steric factors.6a The undesired
5-isomer 10 converts through to the corresponding regioisomer
of drug substance (2), and this impurity needed to be controlled
to very low levels. In the initial route, preparative HPLC was
used to remove this isomer from drug substance, and this was
unsuitable for the long-term supply of GSK2137305 (1).
Improving the regioselectivity of the Ullmann coupling reaction
was considered; however, given the literature precedent for
coupling of 4-methyl imidazole with aryl bromides, it was
regarded that any significant improvements within a short time
frame would be unlikely. Instead, the focus remained on
developing facile chemistry that is easy to scale and to gain
the desired purification through control of the crystallisation
processes.
An initial screen of the CuI catalysed coupling of 8 with
4-methyl imidazole in different solvents and with a range of
commercially available ligands identified several useful leads,
particularly for alcoholic solvents with diamine ligands.8
However, it proved difficult to quickly identify a facile method
to isolate 9 free from the catalyst, ligand, and other impurities.9
We reasoned that the presence of 4-methyl imidazole might
render an additional ligand superfluous and that ‘ligandless’
conditions would simplify the isolation of 9. Gratifyingly, the
With a new, efficient route to imidazolidinone 7 in place,
all that remained to complete the four-stage synthesis of
GSK2137305 (1) was the oxidation of 7, an Ullmann coupling
with 4-methyl imidazole and an N-alkylation reaction (see
Scheme 3). However, there were still many issues that needed
to be addressed with these three steps in order to develop a
route suitable for further scale-up. The oxidation of imidazo-
lidinone 7 had initially been performed using N-bromosuccin-
imide (NBS) in dichloromethane. However, it was shown that
the imidazolone product 8 competitively reacted with NBS,
generating a range of decomposition products and limiting the
isolated yield of 8 to about 50%. When 2,3-dichloro-5,6-
dicyano-p-benzoquinone (DDQ) was used as oxidant, complete
conversion to 8 was observed after 1 h at 60 °C in ethyl acetate.
Furthermore, unlike the NBS process, the product remained
stable to the reaction conditions. Excess DDQ was destroyed
with aqueous sodium sulfite, and following removal of the
aqueous waste and concentration of the organic phase, imi-
dazalone 8 crystallised from solution and was isolated in
80-85% yield.
The next challenge was to develop a scalable coupling of
aryl bromide 8 with 4-methyl imidazole. There has been a
resurgence of interest in recent years in the Ullmann-type
copper-catalysed arylation of nucleophiles, in particular around
the N-arylation of azoles, since these have proved difficult to
couple under palladium catalysed conditions.5 Indeed, the
chemistry utilised in the initial route used a CuI/L-histidine
system to affect the coupling. Although this was successful in
preparing 9, high loadings of CuI (0.6 equiv) and L-histidine
(1.2 equiv) were employed and several side reactions competed,
including the coupling of L-histidine with the aryl bromide. This
meant that the isolation of 9 from the reaction mixture was not
(6) (a) Altman, R. A.; Koval, E. D.; Buchwald, S. L. J. Org. Chem. 2007,
72, 6190. (b) Zhu, L.; Guo, P.; Li, G.; Lan, J.; Xie, R.; You, J. J.
Org. Chem. 2007, 72, 8535. (c) Huang, W.; Shakespear, W. C.
Synthesis 2007, 2121. (d) Kiyomori, A.; Marcoux, J.; Buchwald, S. L.
Tetrahedron Lett. 1999, 40, 2657. (e) Lo, Y. S.; Nolan, J. C.; Maren,
T. H.; Welstead, W. J., Jr.; Gripshover, D. F.; Shamblee, D. A. J. Med.
Chem. 1992, 35, 4790.
(7) A recent example of the coupling of an aryl iodide with 4-methyl
imidazole under mild conditions has reported a 17:1 ratio of regioi-
somers; see: Zhu, L.; Guo, P.; Li, G.; Lan, J.; Luo, L.; You, J. J. Org.
Chem. 2009, 74, 2200.
(5) For recent reviews on the copper-catalysed arylation of nucleophiles,
see: (a) Monnier, F.; Taillefer, M. Angew. Chem., Int. Ed 2008, 47,
3096. (b) Sorokin, V. I. Mini-ReV. Org. Chem. 2008, 5, 323.
46
•
Vol. 15, No. 1, 2011 / Organic Process Research & Development

‘ligandless’ CuI catalysed reaction between aryl bromide 8 and
4-methyl imidazole proceeded cleanly in DMSO at 130 °C to
afford a ca. 4:1 mixture of regioisomers 9 and 10,10 albeit with
a slightly extended reaction time of ca. 36 h.11 Addition of
aqueous L-cysteine solution and isopropyl acetate allowed the
removal of the inorganics, including copper, in the aqueous
layer, and 9 was crystallised from the organic phase in 50-55%
yield as a ca. 20:1 mixture of regioisomers. Although we had
identified milder conditions from our initial screen, these
‘ligandless’ conditions meant that the process was cheap,
operationally simple, and suitable for supplying 9 in high purity
on a multikilogram scale.
The final chemistry stage required the N-alkylation of
imidazalone 9. The reaction proved to be more efficient using
2-chloro-N-[3-(trifluoromethyl)phenyl]acetamide rather than the
bromo analogue 11, with the optimal reaction profiles obtained
in a polar aprotic solvent and with a strong base. Using NMP
as solvent, it was possible to directly telescope the alkylation
with the preparation of 2-chloro-N-[3-(trifluoromethyl)pheny-
l]acetamide, from 3-aminobenzotrifluoride and chloroacetyl
chloride.12 Crystallisation of 1 was achieved by addition of water
to the reaction mixture, with 1 isolated in 75-80% yield in
high purity and <1% of the imidazole regioisomer 2. Further
upgrade in the purity of 1 could be readily obtained by
recrystallisation from a DMSO/1-propanol solvent system in
80% yield.
of the chemistry and isolations in the oxidation, Ullmann
coupling, and alkylation steps. The scope and utility of the key
reaction to form the imidazolidinone structure from amino nitrile
and carbonyl compounds are currently being explored and will
be reported in due course.
Experimental Section
3-(4-Bromophenyl)-1,4-diazaspiro[4.4]nonan-2-one 7. A
slurry of 4-bromobenzaldehyde (3) (29.6 kg, 1 equiv) in ethanol
(75 L) was charged to a solution of ammonium acetate (37.5
kg, 3.0 equiv) and sodium cyanide (8.7 kg, 1.1 equiv) in water
(27 L) and aqueous ammonia (35% w/w, d ) 0.88; 63 L),
washing in with further ethanol (15 L). The resultant reaction
mixture was stirred at 20 ( 3 °C for ca. 4 h until the reaction
was deemed complete by HPLC analysis. TBME (150 L) was
added, and the aqueous layer was removed. The organic phase
was washed with water (2 × 150 L), aqueous sodium chloride
solution (20% w/w; 2 × 150 L), and further water (60 L). The
organic phase was azeotropically dried by atmospheric distil-
lation and concentrated to ca. 50 L at 20-30 °C under reduced
pressure. To the concentrate of amino nitrile 12 were charged
1-BuOH (300 L), cyclopentanone (36 L, 2.5 equiv), and NaOEt
(21% wt in ethanol, 6 L, 0.1 equiv), and the reaction was
warmed to 80 ( 5 °C for ca. 12 h until deemed complete by
HPLC analysis. The reaction was cooled to 20 ( 3 °C, and the
resulting crystals were isolated by filtration, washed with cold
ethanol (5 ( 3 °C; 2 × 60 L), and dried under vacuum at 50
( 5 °C to provide an off-white solid of 7 (16.3 kg, 35% yield)
All stages of the synthetic sequence performed as expected
during a pilot plant campaign, with a total of 19.5 kg of 1
prepared in a 12% overall yield, representing a 4-fold increase
in yield compared to the initial route.
1
with 97% area purity by HPLC. H NMR (400 MHz; d6-
DMSO) δ 8.54 (s, 1H), 7.53 (d, J ) 8.4 Hz, 2H), 7.44 (d, J )
8.4 Hz, 2H), 4.53 (d, J ) 8.4 Hz, 1H), 3.62 (d, J ) 8.4 Hz,
1H), 1.85-1.60 (m, 8H). 13C NMR (100 MHz; d6-DMSO) δ
173.4, 139.8, 130.7, 129.4, 120.1, 81.6, 61.1, 40.3, 38.7, 22.5,
22.4. HRMS (ES +ve) calcd for C13H16N2OBr 295.0446, found
295.0452.
Conclusions
A facile and scalable four-stage synthesis of the glycine
transporter inhibitor GSK2137305 (1) has been developed. The
new route considerably lowers the cost and delivery time
required to prepare multikilogram quantities of 1, through the
rapid access to the imidazolidinone intermediate 7 and control
3-(4-Bromophenyl)-1,4-diazaspiro[4.4]non-3-en-2-one 8.
To a slurry of 7 (14.0 kg, 1 equiv) in EtOAc (280 L) at 20 (
5 °C was charged DDQ (11.9 kg, 1.1 equiv). The resultant
reaction mixture was heated to 60 ( 5 °C until the reaction
was deemed complete by HPLC analysis. The reaction was
washed twice at 60-65 °C with aqueous sodium sulfite solution
(10%w/w; 140 and 70 L) and then with water (70 L). The
organic phase was concentrated to ca. 140 L by distillation at
atmospheric pressure, and the resulting slurry cooled to 0 ( 3
°C. The solids were isolated by filtration, washed with cold
EtOAc (0 ( 3 °C; 2 × 28 L), and dried at 50 ( 5 °C under
vacuum to afford off-white crystals of 8 (11.7 kg, 84% yield)
(8) Ullmann coupling screen conditions: Aryl bromide 8 (0.34 mmol),
4-methylimidazole (0.68 mmol), CuI (0.068 mmol), Cs2CO3 (0.68
mmol), ligand (0.14 mmol), and solvent (1.2 mL) heated to 110 °C
for 30 h under an atmosphere of nitrogen, monitoring by HPLC.
Ligands screened: L-proline; ethylene glycol; 1,10-phenanthroline;
ninhydrin; (()-trans-1,2-diaminocyclohexane; 2-dimethylaminoetha-
nol; N,N′-dimethylethylenediamine; trans-N,N′- dimethylcyclohexane-
1,2-diamine; 8-hydroxyquinoline. Solvent screened: DMSO; o-xylene;
toluene; 1,4-dioxane; proprionitrile; NMP; n-butyl acetate; DMPU;
1-butanol; 1-pentanol.
(9) Residual copper levels of e10 ppm in drug substance 1 were required.
To achieve this specification, copper levels of <100 ppm were targeted
in 9.
1
with 99% area purity by HPLC. H NMR (400 MHz; d6-
(10) Regioisomeric ratio determined by 1H NMR. See Supporting Informa-
tion for further details.
DMSO) δ 10.10 (s, 1H), 8.28 (d, J ) 8.8 Hz, 2H), 7.72 (d, J
) 8.8 Hz, 2H), 1.99-1.81 (m, 8H). 13C NMR (100 MHz; d6-
DMSO) δ 163.9, 159.0, 131.6, 129.8, 125.2, 89.7, 37.0, 23.8.
HRMS (ES +ve) calcd for C13H14N2OBr 293.0289, found
293.0298.
3-[4-(4-Methyl-1H-imidazol-1-yl)phenyl]-1,4-diazaspiro-
[4.4]non-3-en-2-one 9. A slurry of 8 (10.5 kg, 1 equiv),
4-methyl imidazole (10.5 kg, 3.6 equiv), potassium carbonate
(10.5 kg, 2.1 equiv), and copper(I) iodide (1.5 kg, 0.15 equiv)
in DMSO (52 L) was heated to 130 ( 3 °C until the reaction
was deemed complete by HPLC (∼36 h). The reaction mixture
(11) Only DMSO was investigated as solvent for the ‘ligandless’ reaction,
since it is a preferred dipolar aprotic solvent for use on scale and some
success in isolating 9 from a DMSO solution had previously been
achieved. For other examples of copper-catalysed coupling of imida-
zoles with aryl halides that do not require a separate ligand, see refs
6b,e, and 7.
(12) 3-Aminobenzotrifluoride, chloroacetyl chloride, and 2-chloro-N-[3-
(trifluoromethyl)phenyl]acetamide were all flagged as potential geno-
toxic impurities. Chloroacetyl chloride will be hydrolysed to chloro-
acetic acid under the reaction conditions, and this, together with
3-aminobenzotrifluoride and 2-chloro-N-[3-(trifluoromethyl)pheny-
l]acetamide, was nonmutagenic when tested in a bacterial mutation
screening assay (Ames test).
Vol. 15, No. 1, 2011 / Organic Process Research & Development
•
47

1
was cooled to 60-70 °C and treated with isopropyl acetate (73
L) and 3% w/w aqueous L-cysteine solution (73 L). The layers
were separated, and the lower aqueous phase was back-extracted
with isopropyl acetate (73 L) at 60 ( 5 °C. The organic phases
were combined, diluted with ethanol (42 L), and then washed
twice with a mixture of 3% w/w aqueous L-cysteine solution
(31 L) and 10% w/w K2CO3 solution (31 L), followed by water
washes (2 × 52 L). Isopropyl acetate (73 L) was charged, and
the reaction was concentrated by atmospheric distillation to ca.
105 L, during which time the product crystallised. The slurry
was cooled to 20 ( 3 °C, and the solids were isolated by
filtration, washed with isopropyl acetate (52 L then 26 L), and
dried under vacuum at 50 ( 5 °C to provide an off-white solid
of 9 (5.65 kg, 54%) with 97% area purity by HPLC and ca.
5% regioisomer 10 by 1H NMR. 1H NMR (400 MHz;
d6-DMSO) δ 10.05 (s, 1H), 8.43 (d, J ) 9.2 Hz, 2H), 8.26 (d,
J ) 1.2 Hz, 1H), 7.76 (d, J ) 9.2 Hz, 2H), 7.54 (s, 1H), 2.18
(s, 3H), 2.00-1.82 (m, 8H). 13C NMR (100 MHz; d6-DMSO)
δ 164.1, 158.8, 138.9, 138.8, 134.7, 129.4, 128.5, 119.3, 113.8,
89.6, 37.1, 23.8, 13.6. HRMS (ES +ve) calcd for C17H19N4O
295.1559, found 295.1561.
99% area purity by HPLC. Mp 258-260 °C. H NMR (500
MHz; d6-DMSO) δ 10.57 (s, 1H), 8.46 (d, J ) 8.5 Hz, 2H),
8.28 (d, J ) 1.0 Hz, 1H), 8.11 (s, 1H), 7.80 (d, J ) 9.0 Hz,
2H), 7.77 (d, J ) 8.5 Hz, 1H), 7.59 (t, J ) 8.0 Hz, 1H), 7.56
(s, 1H), 7.44 (d, J ) 8.0 Hz, 1H), 4.38 (s, 2H), 2.19 (s, 3H),
2.16-1.68 (m, 8H). 13C NMR (100 MHz; d6-DMSO) δ 166.3,
162.6, 157.5, 139.4, 139.1, 138.9, 134.8, 130.1, 129.5 (q, JCF
) 31.6 Hz), 129.4, 128.3, 124.1 (q, JCF ) 272 Hz), 122.8, 119.9
(q, JCF ) 3.5 Hz), 119.4, 115.2 (q, JCF ) 3.8 Hz), 113.8, 93.3,
43.4, 34.4, 23.6, 13.6. HRMS (ES +ve) calcd for C26H25F3N5O2
496.1955, found 496.1937.
Recrystallisation of 1. Crude 1 (9.65 kg) was dissolved in
DMSO (48 L) at 80 ( 5 °C, filtered, and charged with
1-propanol (96 L). The solution was cooled to 65 ( 3 °C and
seeded with a slurry of 1 (9.6 g) in 1-propanol (100 mL). The
slurry was cooled to 20 ( 3 °C, and the solids were isolated
by filtration, washed with 1-propanol (2 × 29 L), and dried
under vacuum at 50 ( 5 °C to provide a white, crystalline solid
of 1 (7.74 kg, 80%) with >99% area purity by HPLC and <0.3%
of the imidazole regioisomer 2.
2-{3-[4-(4-Methyl-1H-imidazol-1-yl)phenyl]-2-oxo-1,4-di-
azaspiro[4.4]non-3-en-1-yl}-N-[3-(trifluoromethyl)phenyl]ac-
etamide 1. Chloroacetyl chloride (3.15 kg, 1.1 equiv) was
charged to a solution of 3-aminobenzotrifluoride (4.73 kg, 1.15
equiv) in NMP (30 L), maintaining a temperature in the range
25 ( 5 °C. The reaction solution was stirred at 25 ( 5 °C until
the reaction was deemed complete by HPLC (ca. 10 min). This
solution was then charged to a stirred solution of 9 (7.5 kg, 1
equiv) and 45% wt. aqueous KOH solution (5.5 L, 2.5 equiv)
in NMP (30 L), rinsing in with NMP (7.5 L). The reaction was
stirred at 25 ( 5 °C until deemed complete by HPLC (ca. 6 h).
Product was crystallised by addition of water (37 L), and the
slurry temperature cycled to 70 ( 3 °C before isolation by
filtration at 20 ( 3 °C. The solids were washed sequentially
with 2:1 v/v NMP/water (22 L vol), water (2 × 22 L), and
1-propanol (22 L). The cake was reslurried in 1-propanol (75
L), then filtered, and dried under vacuum at 50 ( 5 °C to
provide white crystals of GSK2137305 (1) (9.8 kg, 78%) with
Acknowledgment
We thank our R&D Chemical Development colleagues
Claire Crawford from Synthetic Chemistry for automated
reaction screening support; Jamie Russell, Alistair Greenlaw,
and Steve Gooding from Process Technologies for enabling the
realisation of the chemistry in the pilot plant; Andy Smith for
key analytical support; and Felicity Shaw from Technical
Sourcing.
Supporting Information Available
1H and 13C NMR, LC/MS, and HPLC data of the final
product (1), isolated intermediates (7, 8, and 9), and the
regioisomeric impurity 2. This material is available free of
charge via the Internet at http://pubs.acs.org.
Received for review July 30, 2010.
OP100210S
48
•
Vol. 15, No. 1, 2011 / Organic Process Research & Development