Evaluation of novel synthetic methods for the preparation of the sodium channel inhibitor, GW273225X

Article abstract of DOI:10.1021/op4001753

The evaluation of efficient synthetic methods for the preparation of (R)-2,4-diamino-5-(2,3-dichlorophenyl)-6- fluoromethylpyrimidine, GW273225X (1), is described. The initial synthesis using ethylfluoroacetate was evaluated against three alternative rout

Full text of DOI:10.1021/op4001753

Article
pubs.acs.org/OPRD
Evaluation of Novel Synthetic Methods for the Preparation of the
Sodium Channel Inhibitor, GW273225X
Matthew D. Walker,* F. David Albinson, Hugh F. Clark, Stacy Clark, Nicholas P. Henley,
Richard A. J. Horan, Chris W. Jones, Charles E. Wade, and Richard A. Ward
Product Development, GlaxoSmithKline Pharmaceuticals, Medicines Research Centre, Gunnels Wood Road, Stevenage, SG1 2NY,
U.K.
S
* Supporting Information
ABSTRACT: The evaluation of efficient synthetic methods for the preparation of (R)-2,4-diamino-5-(2,3-dichlorophenyl)-6-
fluoromethylpyrimidine, GW273225X (1), is described. The initial synthesis using ethylfluoroacetate was evaluated against three
alternative routes using either nucleophilic fluorination, electrophilic fluorination, or sodium fluoroacetate.
ether 5 was treated with guanidine hydrochloride and NaOMe
in methanol to give the racemic 2,4-diaminopyrimidine 6 in
56% overall yield from nitrile 3. Resolution of the racemate 6
was then effected by forming the DBTA salt.3 At this point a
recrystallisation was required in order to obtain the salt with the
required de, giving the DBTA salt 7 in 35% yield from the
racemate 6. The DBTA salt 7 was free-based by treatment with
Et3N before recrystallisation of intermediate grade material to
give the desired product GW273225X 1 in 83% yield and >99%
ee from the DBTA salt 7.4 The high toxicity of the materials
being used in this synthesis required the adoption of stringent
safety control measures, causing the cycle time of this process
to become protracted. We therefore chose to investigate a
number of alternative routes to the racemate of GW273225X 1
that avoided or minimised the use of ethyl fluoroacetate and
ethyl iodide.
Development of Alternative Routes to GW273225X
1Use of Nucleophilic Fluorination. Previous investiga-
tions, which focused on the deoxyfluorination of alcohol 11,
had identified (diethylamino)sulfur trifluoride (DAST) as the
only successful reagent for completing this transformation (see
Scheme 2).1a,5
Synthesis of the alcohol 11 was carried out by Claisen
condensation of nitrile 3 with ethyl diethoxyacetate, followed
by O-alkylation and guanidination to give acetal 9. The acetal 9
was hydrolysed and reduced with sodium borohydride to give
the alcohol 11 in 43% yield from nitrile 3.
INTRODUCTION
GW273225X 1 (Figure 1) was initially candidate-selected in
1995 for neuropathic pain, but development was stopped in
■
Figure 1. GW273225X 1 and lamotrigine (Lamictal) 2.
1997 due to a lack of efficacy. The program was restarted in
2005 as a follow-up to lamotrigine 2 (Lamictal) after evaluation
for bipolar disorder and epilepsy.1
Initial toxicology and clinical studies were supplied via a
route that started from a Claisen condensation of 2,3-
dichlorophenylacetonitrile 3 with highly toxic ethylfluoroace-
tate.2 The material was provided as a single atropisomer by
usage of a classical resolution procedure. Before preparing
further supplies of material, we were keen to assess the viability
of alternative routes. We report here how three new routes to
GW273225X were identified and their suitability for large-scale
manufacture evaluated.
We then focused on optimisation of the direct deoxyfluori-
nation of alcohol 11 using DAST.6 A limited range of
compatible solvents were screened with no success. A screen
of reaction conditions, including concentration, temperature,
and DAST stoichiometry, was carried out but failed to improve
on the previously identified reaction conditions. Alcohol 11 was
then deoxyfluorinated using 5 equiv of DAST in DCM, which
gave the racemate 6 in 87% yield. The requirement for large
excesses of potentially explosive DAST and cryogenic reaction
temperatures meant work in this area was terminated.
RESULTS AND DISCUSSION
■
Initial Supply Route to GW273225X 1. The synthesis of
GW273225X 1 (Scheme 1) was carried out by first preparing
the racemic 2,4-diaminopyrimidine 6 in three telescoped stages
followed by resolution using dibenzoyl-L-tartaric acid hydrate
(DBTA·H2O).
Claisen condensation of nitrile 3 with 1.5 equiv of
ethylfluoroacetate promoted by NaOMe in methanol gave
enol 4 after acidification of the reaction mixture with aqueous
HCl. Care was required at this point due to the instability of the
enol product towards the retro-Claisen reaction. The wet cake
obtained from the isolation of 4 was alkylated with EtI and
K2CO3 in DMF after azeotropic drying, and the resultant enol
Received: July 1, 2013
© XXXX American Chemical Society
A
dx.doi.org/10.1021/op4001753 | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
Scheme 1. Route A to GW273225X 1 via Ethylfluoroacetate
Scheme 2. Synthesis of Racemic Diaminopyrimidine 6 via Deoxyfluorination
Scheme 3. Synthesis of Racemic Diaminopyrimidine 6 via Nucleophilic Fluorination
An alternative stepwise strategy was subsequently inves-
tigated, whereby alcohol 11 would be converted into an
intermediate mesylate 12 and then transformed into the desired
fluoride 6.7 Although the mesylate 12 could be formed in situ,
we were initially unable to transform the mesylate 12 into the
fluoride 6 using CsF. Pleasingly, the use of the ionic liquid
[bmim][BF4] as a solvent cleanly afforded the required fluoride
6, and subsequent factor screening identified that the ionic
liquid could be employed as a cosolvent.8 We found that only
1.25 equiv of the ionic liquid and 1.25 equiv of CsF with
respect to the alcohol 11 were required to produce the fluoride
6 in aqueous acetonitrile (Scheme 3).
water to control the precipitation of the product and cleanly
afford the fluoride 6 in 58% yield from the alcohol 11.
Development of Alternative Routes to GW273225X
1Use of Electrophilic Fluorination. Electrophilic fluori-
nation was an attractive option due to the simpler synthesis of
the required substrates and the ready access to a range of easily
handled electrophilic fluorinating agents.9 It was felt that the
analogous des-fluoro enol ether 14 would be the most viable
substrate for this approach due to concerns about reagent
compatibility with other substrates, such as the analogous des-
fluoro 2,4-diaminopyrimidines.
Initially an adaptation of the existing route was used with
methyl acetate replacing the toxic ethyl fluoroacetate.10 The
initial Claisen reaction was significantly slower than with ethyl
fluoroacetate even with higher stoichiometries of both base and
methyl acetate. The enol 13 was isolated as an oil and alkylated
Unfortunately, owing to the low solubility of the fluoride 6 in
many organic solvents, isolation of the product from the ionic
liquid proved difficult. This was circumvented by the addition
of DMSO to the reaction mixture, followed by slow addition of
B
dx.doi.org/10.1021/op4001753 | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
under the ethyl iodide conditions used previously to give the
enol ether 14 in order to investigate the fluorination. The enol
ether 14 was extracted into toluene, washed with aqueous
NaCl, and distilled in order to remove the DMF and ensure
azeotropic drying of the product solution ready for use in the
fluorination.
Development of Alternative Routes to GW273225X
1Use of Sodium Fluoroacetate. We decided to
investigate the use of alternative fluoroacetate analogues in
order to minimise the fluoroacetate stiochiometry required in
Stage 1 and to simplify and reduce both the process cycle time
and the waste streams generated. It was found that sodium
fluoroacetate (SFA) was the only readily commercially available
fluoroacetate synthon, and it was anticipated that by use of an
appropriate activating agent the condensation reaction would
only require a stoichiometric amount of SFA.13,14
Screening of activating agents demonstrated 1,1′-carbon-
yldiimidazole (CDI) to be a viable reagent for this activation,
and addition of 1.2 equiv of CDI to a slurry of nitrile 3 and 1
equiv of SFA in DMF at 80 °C promoted the formation of the
sodium enolate 15 with good conversion. A number of solvents
were screened in an attempt to replace DMF, but due to the
low solubility of CDI, DMF was found to be optimal in keeping
the process volume efficient.
A method for conversion of the sodium enolate 15 to the
2,4-diaminopyrimidine 6 was then required that avoided the
highly inefficient isolation of the enol 4. Direct O-alkylation of
the reaction mixture generated from the CDI mediated Claisen
condensation failed to go to completion using the original
conditions, and the alkylation was found to require 6 equiv of
ethyl iodide due to competitive alkylation of imidazole
generated from the CDI reaction.15 Alternative methods for
forming the enol ether 5 were investigated, using further
various activating agents, followed by quenching with ethanol,
but only the starting enol 4 was observed in any reaction.
We did find that treatment of the sodium enolate 15 with
thionyl chloride cleanly gave a new product which was
eventually identified as the imidazolide 16 rather than the
analogous vinyl chloride as expected. Due to the potential
formation of toxic dimethylcarbamoyl chloride (DMCC) from
DMF and thionyl chloride, we elected to screen a number of
alternative activating agents. After screening, it was established
that the use of propylphosphonic anhydride (T3P) also readily
generated the same imidazolide 16 with good conversion.
Optimisation showed that the addition of 1.1 equiv of T3P
(50% w/w in ethyl acetate) to a solution of the sodium enolate
15 at 30 °C initially generated the corresponding activated
phosphonic ester and only increasing the temperature to 80 °C
promoted complete conversion to the imidazolide 16.
Following work up with aqueous potassium carbonate and a
solvent swap into methanol, the crude imidazolide 16 was
condensed with guanidine hydrochloride using sodium
methoxide in methanol to give the desired 2,4-diaminopyr-
imidine 6. This efficient one-pot telescoped process was carried
out using these reactions to yield 2,4-diaminopyrimidine 6 in
46% overall yield over the three stages (Scheme 5).
The direct fluorination of the enol ether 14 was initially
investigated by treatment with NaHMDS at −20 °C followed
by the addition of fluorinating agent N-fluorobenzenesulpho-
nimide (NFSI), which was shown to give the desired
fluorinated enol ether 5.11 Optimisation of the reaction
conditions was carried out between −20 and 0 °C and
included a range of bases (strong and weak, organic and
inorganic), solvents (ethers, toluene, and heptane), reagent
equivalents, concentrations, and NFSI addition rates.
Only the strong bases (LDA and MHMDS where M = Li,
Na, K) gave any significant amount of product, and the use of
LiHMDS in THF at lower temperatures was found to be
optimal (although 2-MeTHF and toluene also gave good
results). The other critical factor was found to be the rate of
addition of the NFSI which needed to be as fast as possible,
presumably in order to prevent equilibration of the anionic
species in the reaction (hence leading to under- and
overfluorinated products). However, it was also found that
the reaction of the anion of the enol ether 14 with NFSI was
very exothermic, preventing us from carrying out the NFSI
addition rapidly.
This problem was overcome by a reverse addition protocol,
whereby a solution of the anion was added to a solution of
NFSI. This allowed the exotherm to be controlled by the rate of
addition and also reduced the potential for anion exchange,
thereby reducing the levels of under- and overfluorinated
impurities.
A basic workup of this reaction was also desired to avoid the
potential of generating hydrofluoric acid, and after a study of
different workup conditions, it was found that quenching the
THF/toluene reaction mixture with dilute aqueous sodium
bicarbonate allowed the complete removal of the NFSI derived
byproducts after two washes with minimal loss of the desired
enol ether 5 to the aqueous phases.
Following removal of THF by distillation, the toluene
solution of enol ether 5 was reacted with guanidine
hydrochloride under the reaction conditions used in the
original route to give 2,4-diaminopyrimidine 6 in 40% overall
yield from the nitrile 3 (Scheme 4).12
Scheme 4. Synthesis of Racemic Diaminopyrimidine 6 via
Electrophilic Fluorination
Evaluation of Alternative Routes to GW273225X 1.
The evaluation of the synthetic routes towards GW273225X 1
was carried out using decision analysis methods based on a
weighting of the desired attributes for the synthesis and an
evaluation of each route against these attributes (see Table 1).16
The evaluation shows clearly that the SFA route performs the
best in this analysis, primarily due to the efficiency of the one-
pot telescoped process. The cycle time for the SFA process was
anticipated to be much lower than the current process, and the
application of the novel Claisen condensation reaction
significantly reduced the toxicity of the process by cutting the
amount of fluoroacetate used to 1 equiv. The development of a
new activation strategy, using T3P as a significantly safer
C
dx.doi.org/10.1021/op4001753 | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
give a white crystalline solid (104.1 g). This solid (92.0 g) was
slurried in a mixture of propan-1-ol (538 mL) and water (290
Scheme 5. Synthesis of Racemic Diaminopyrimidine 6 via
Claisen Condensation with SFA
mL) and heated to 80
3 °C for 30 min to give a clear
solution. The resulting solution was filtered to a clean vessel
followed by a mixture of propan-1-ol (59.8 mL) and water
(32.2 mL) at 80 3 °C and cooled to 71 3 °C. The mixture
was seeded with (R)-2,4-diamino-5-(2,3-dichlorophenyl)-6-
fluoromethylpyrimidine (1) (92 mg) in propan-1-ol (1.84
mL) and the mixture aged for 2 h at 71 3 °C. The slurry was
cooled to 0 3 °C over 4.5 h. Water (1077 mL) was added
over 8 h, stirred for 1 h, and filtered. The cake was washed with
a mixture of propan-1-ol (37.5 mL) and water (87.5 mL) and
with water (2 × 125 mL) and pulled dry. The solid was dried
under vacuum at 70 5 °C to give a white solid (81.9 g, 83%
overall yield): [α]2D0 −56.75 (c = 0.53, EtOH). mp 215−216 °C.
1H NMR (400 MHz, d6-DMSO) δ 7.63 (dd, J = 8.1, 1.5 Hz,
1H), 7.39 (t, J = 7.8 Hz, 1H), 7.24 (dd, J = 7.8, 1.5 Hz, 1H),
6.18 (br s, 2H), 6.03 (br s, 2H), 4.82 (dd, J = 22.5, 11.0 Hz,
2H), 4.70 (dd, J = 22.5, 11.0 Hz, 2H). 13C NMR (100 MHz, d6-
DMSO) δ 162.7 (C), 162.0 (C), 158.1 (d, J = 15.0 Hz, C),
135.7 (C), 132.7 (C), 132.1 (C), 131.6 (CH), 130.2 (CH),
128.4 (CH), 104.7 (d, J = 3.1 Hz, C), 82.5 (d, J = 167.5 Hz,
CH2). 19F NMR (376 MHz, d6-DMSO) δ −216.7 (t, J = 47.8
Hz). MS (ES+): m/z 287 [M + H]+.
alternative to ethyl iodide, also reduced the number of waste
streams generated from 8 to 3 by telescoping the activation and
guanidine condensation. Consequently, in developing a one-pot
process we have eliminated the transfer of highly toxic reaction
mixtures and have therefore made the process operationally
safer.17
2,4-Diamino-5-(2,3-dichlorophenyl)-6-fluoromethylpyri-
midine (6) from 2,3-Dichlorophenylacetonitrile (3)Method
A (Ethylfluoroacetate). 2,3-Dichlorophenylacetonitrile (3)
(100 g, 538 mmol) was suspended in methanol (200 mL)
and cooled to 10 5 °C. 30% (w/w) Sodium methoxide in
methanol solution (400 mL, 2.16 mol) was added followed by
methanol (20 mL), maintaining the temperature of the reaction
SUMMARY
■
In conclusion, three new routes have been developed suitable
for the synthesis of GW273225X 1 in up to 15% overall yield,
including an unusual CDI and T3P mediated heterocycle
synthesis.18 These new routes were evaluated against the
current route using decision analysis methods to provide a clear
candidate for further process development.
mixture below 30 °C. The temperature was adjusted to 20
5
°C. Ethylfluoroacetate (CAUTION: highly toxic) (78 mL, 807
mmol) was added followed by methanol (35 mL). The reaction
mixture was stirred at 20 5 °C for 18 h. The reaction mixture
was cooled to 5 5 °C and added into a mixture of 32% (w/w)
aqueous hydrochloric acid (320 mL) and water (1150 mL)
over 40 min, followed by methanol (45 mL), maintaining the
temperature below 20 °C. The reaction mixture was warmed to
20 5 °C over 60 min. The resultant slurry was filtered and the
cake washed with a mixture of water (330 mL), methanol (150
mL), and 32% (w/w) aqueous hydrochloric acid (22.5 mL)
before pulling dry. The cake was dissolved in ethyl acetate (800
mL), the solution transferred to the reaction vessel followed by
ethyl acetate (50 mL), and the organic solution was then
washed with a solution of 20% (w/w) aqueous sodium chloride
(45 mL) at 20 5 °C. The mixture was then washed with a
solution of 20% (w/w) aqueous sodium chloride (90 mL) at 20
EXPERIMENTAL SECTION
■
General. (R)-2,4-Diamino-5-(2,3-dichlorophenyl)-6-fluoro-
methylpyrimidine (1) from (R)-2,4-Diamino-5-(2,3-dichlor-
ophenyl)-6-fluoromethylpyrimidine·DBTA Salt (7). (R)-2,4-
Diamino-5-(2,3-dichlorophenyl)-6-fluoromethylpyrimidine·
DBTA salt (7) (250 g, 387 mmol) was dissolved in a mixture of
DMSO (675 mL) and water (75 mL) at 20
5 °C.
Triethylamine (113 mL, 811 mmol) was added and the mixture
stirred at 20 5 °C for 15 min. The solution was seeded with
(R)-2,4-diamino-5-(2,3-dichlorophenyl)-6-fluoromethylpyrimi-
dine (1) (200 mg) and aged for 90 min. Water (875 mL) was
added over 75 min; the resultant slurry was aged at 20 5 °C
for 60 min and filtered. The cake was washed with water (2 ×
375 mL), pulled dry, and dried under vacuum at 50 5 °C to
Table 1. Evaluation of Routes to Diaminopyrimidine 6
a
b
b
b
b
criteria
weight
EFA route
F− route
F+ route
SFA route
steps
2
4
5
3
4
3
5
1
2
4
4
3
5
1
health and safety
purity
5
3
4
5
yield
4
1
3
3
environmental
throughput
1
3
3
4
1
4
4
5
c
total
58
62
73
79
a
b
Scores for each weighting were given from 1 = least important to 5 = most important. Scores for each category were given from 1 = meets the
c
criteria least successfully to 5 = meets the criteria most successfully. The total score was given by the sum of the products of the weighting and the
scoring for each category.
D
dx.doi.org/10.1021/op4001753 | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
5 °C and distilled down to 380 mL in volume, under vacuum
at 200 50 mbar at with the jacket set at 55 °C. DMF (130
mL) was added and the organic solution distilled down to 250
mL in volume, under vacuum at 200 50 mbar with the jacket
set at 55 °C. The reaction mixture was cooled to 5 5 °C and
added over 20 min to a suspension of potassium carbonate
(93.3 g, 675 mmol) in DMF (300 mL) at 5 5 °C followed by
DMF (50 mL), maintaining the temperature below 20 °C. The
reaction mixture was stirred at 15 5 °C for 60 min, and ethyl
iodide (86 mL, 1.08 mol) was added in a single portion
followed by DMF (20 mL). The reaction mixture was heated to
65 5 °C over 1 h and stirred at 65 5 °C for 4 h. The
mixture was cooled to 25 5 °C, toluene was added (600 mL),
and the mixture was washed with water (800 mL) at 20 5 °C.
The organic phase was washed with a solution of 5% (w/w)
aqueous sodium chloride (350 mL). The organic solution was
distilled under vacuum at 125 50 mbar with the jacket set at
65 °C down to 130 mL and cooled to 20 5 °C. The resultant
mixture was added over 40 min to a mixture of guanidine
hydrochloride (56.4 g, 590 mmol) and 30% (w/w) sodium
methoxide in methanol (110 mL, 594 mmol) in methanol (155
mL), followed by methanol (35 mL) maintaining the
temperature below 25 °C. The reaction mixture was heated
to 30 5 °C over 30 min, stirred at 30 5 °C for 3 h, and
cooled to 20 5 °C over 30 min. A mixture of methanol (300
mL) and water (300 mL) was added over 30 min, maintaining
5
5 °C followed by DMF (50 mL), maintaining the
temperature below 20 °C. The reaction mixture was stirred at
15 5 °C for 60 min, and ethyl iodide (86 mL, 1.08 mol) was
added in a single portion followed by DMF (20 mL). The
reaction mixture was heated to 65 5 °C over 1 h and stirred
at 65 5 °C for 4 h. The mixture was cooled to 25 5 °C;
toluene was added (600 mL) and the mixture washed with
water (800 mL) at 20 5 °C. The organic phase was washed
with a solution of 5% (w/w) aqueous sodium chloride (350
mL), and the organic solution was distilled under vacuum at
125 50 mbar with the jacket set at 65 °C down to 130 mL
and cooled to 20 5 °C. The organic solution was distilled
down to 154 mL in volume, under vacuum at 125 50 mbar
with the jacket set at 60 5 °C and cooled to 20 5 °C. To a
stirred mixture of 72 mL of this solution and toluene (125 mL)
at −25 5 °C was added a 1 M in THF LiHMDS solution
(216 mL, 216 mmol), maintaining the temperature below −15
°C. This solution was added over 2 h to a solution of NFSI (68
g, 216 mmol) in THF (250 mL) at 0 5 °C via cannula. The
reaction mixture was stirred at 0 5 °C for a further 1 h and
quenched by the addition of 8% (w/w) aqueous sodium
bicarbonate (100 mL), followed by water (150 mL). The
reaction mixture was warmed to 20
5 °C and allowed to
separate. The organic phase was washed with a mixture of 8%
(w/w) aqueous sodium bicarbonate (100 mL) and water (150
mL). The organic phase was distilled down to 100 mL in
volume, under vacuum at 100 50 mbar with the jacket set at
45 °C. Toluene (100 mL) was added, and the reaction mixture
was distilled down to 70 mL in volume, under vacuum at 100
50 mbar with the jacket set at 65 °C. The resulting mixture was
added to a stirred mixture of guanidine hydrochloride (12.2 g,
128 mmol) and 30% (w/w) sodium methoxide in methanol
solution (24 mL, 126 mmol) in methanol (30 mL), followed by
methanol (20 mL) maintaining the temperature below 25 °C.
The reaction mixture was heated to 30 5 °C over 30 min,
stirred at 30 5 °C for 3 h, and cooled to 20 5 °C over 30
min. The reaction mixture was stirred at 20 5 °C for 40 min,
and the resultant slurry was filtered. The cake was washed with
methanol (3 × 20 mL), pulled dry, and dried under vacuum at
50 5 °C to give a cream coloured crystalline solid (13.07 g,
40% yield).
the temperature of the reaction mixture at 20
5 °C. The
reaction mixture was stirred at 20 5 °C for 40 min, and the
resultant slurry was filtered. The cake was washed with
methanol (4 × 200 mL), pulled dry, and dried under vacuum
at 50 5 °C to give a cream coloured crystalline solid (86.6 g,
56.1% yield).
2,4-Diamino-5-(2,3-dichlorophenyl)-6-fluoromethylpyri-
midine (6) from 2,3-Dichlorophenylacetonitrile (3)Method
B (Electrophilic Fluorination). 2,3-Dichlorophenylacetonitrile
(3) (100 g, 538 mmol) was suspended in methanol (200 mL)
and cooled to 10
5 °C. 30% (w/w) Sodium methoxide
solution (410 mL, 2.15 mol) was added followed by methanol
(20 mL), maintaining the temperature of the reaction mixture
below 30 °C. The temperature was adjusted to 20
Methyl acetate (65 mL, 818 mmol) was added followed by
methanol (35 mL). The reaction mixture was stirred at 20
5 °C.
5
2,4-Diamino-5-(2,3-dichlorophenyl)-6-fluoromethylpyri-
midine (6) from 2,3-Dichlorophenylacetonitrile (3)Method
C (Sodium Fluoroacetate). 2,3-Dichlorophenylacetonitrile (3)
(100 g, 538 mmol) and sodium fluoroacetate (CAUTION:
highly toxic) (54 g, 538 mmol) were suspended in ethyl acetate
°C for 18 h. Methyl acetate (22 mL, 278 mmol) was added, and
the reaction mixture was stirred at 20 5 °C for 24 h. Methyl
acetate (44 mL, 554 mmol) was added, and the reaction
mixture was stirred at 20
reaction mixture was cooled to 5
5 °C for a further 24 h. The
5 °C and added into a
(200 mL) and the temperature adjusted to 75
5 °C. The
mixture of 32% (w/w) aqueous hydrochloric acid (274 mL)
and water (1200 mL) over 60 min. The reaction mixture was
extracted with ethyl acetate (800 mL) and then ethyl acetate
(400 mL) at 20 5 °C. The combined organic extracts were
washed with 20% (w/w) aqueous sodium chloride (50 mL),
followed by 20% (w/w) aqueous sodium chloride (100 mL) at
20 5 °C. The organic solution was distilled down to 150 mL
in volume, under vacuum at 200 50 mbar with the jacket set
slurry was aged for 1 h, and a solution of 1,1′-carbon-
yldiimidazole (105 g, 648 mmol) in dimethyl formamide (400
mL) was added over 2 h, followed by dimethyl formamide (20
mL). The reaction mixture was stirred at 75 5 °C for 30 min.
50% (w/w) Propanephosphonic cyclic anhydride in ethyl
acetate (352 mL, 591 mmol) was added and the mixture aged
for 3 h at 75 5 °C. The mixture was cooled to 20 5 °C;
toluene (400 mL) was added and then washed with 18.8% (w/
w) aqueous potassium carbonate solution (400 g). The organic
solution was then washed with 5% (w/w) aqueous LiCl
solution (400 g) at 20 5 °C. Methanol (400 mL) was added,
and the organic phase was distilled down to 200 mL in volume,
under vacuum at 120 50 mbar with the jacket set at 65 °C.
at 40
acetate (500 mL) and distilled down to 350 mL in volume,
under vacuum at 200 50 mbar with the jacket set at 40
5 °C. The organic solution was diluted with ethyl
5
°C. DMF (130 mL) was added and the organic solution
distilled down to 300 mL in volume, under vacuum at 200 50
mbar with the jacket set at 55 °C. The resultant mixture was
cooled to 5 5 °C and added over 20 min to a suspension of
potassium carbonate (93.3 g, 675 mmol) in DMF (300 mL) at
The reaction mixture was cooled to 20
5 °C, and then
guanidine hydrochloride (51.4 g, 538 mmol) and methanol
(200 mL) were added. 30% (w/w) Sodium methoxide in
E
dx.doi.org/10.1021/op4001753 | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
methanol (102 mL, 551 mmol) was charged to the reaction
71.7 (CH). 19F NMR (376 MHz, d6-DMSO) δ −217.8 (t, J =
47.8 Hz). MS (ES+): m/z 287 [M + H]+.
maintaining the temperature at 20
5 °C, followed by
methanol (20 mL). The reaction mixture was warmed to 45
5 °C and stirred for 3 h. Water (400 mL) was added; the
mixture was cooled to 20 5 °C and stirred for 30 min. The
resultant slurry was filtered, and the cake was washed with a
mixture of water (100 mL) and methanol (100 mL). The cake
was washed with methanol (2 × 200 mL), and the solid was
dried under vacuum at 50 5 °C to give a cream coloured
crystalline solid (71.0 g, 46% yield): mp 224−226 °C.
2,4-Diamino-5-(2,3-dichlorophenyl)-6-fluoromethylpyri-
midine (6) from 2,4-Diamino-5-(2,3-dichlorophenyl)-6-hy-
droxymethylpyrimidine (11).1 To a stirred suspension of 2,4-
diamino-5-(2,3-dichlorophenyl)-6-hydroxymethylpyrimidine
(11) (2.0 g, 7.01 mmol) in dichloromethane (20 mL) was
added triethylamine (1.37 mL, 8.48 mmol) and DMAP (5 mg,
0.04 mmol) followed by methanesulphonic anhydride (1.47 g,
8.42 mmol), and the reaction mixture was stirred at 20 5 °C
for 1 h. Further triethylamine (0.49 mL, 2.83 mmol) and
methanesulphonic anhydride (0.61 g, 3.49 mmol) were added,
and the reaction was stirred at 20 5 °C for a further 15 min.
The reaction mixture was filtered and washed with DCM (2 ×
4 mL). The crude solid was suspended in a mixture of
acetonitrile (3 mL) and water (0.6 mL), and then [bmim][BF4]
(1.65 mL, 8.84 mmol) and CsF (1.33 g, 8.84 mmol) were
ASSOCIATED CONTENT
* Supporting Information
■
S
General methods description and spectral data for (R)-2,4-
diamino-5-(2,3-dichlorophenyl)-6-fluoromethylpyrimidine (1),
2,4-diamino-5-(2,3-dichlorophenyl)-6-fluoromethylpyrimidine
(6), and (R)-2,4-diamino-5-(2,3-dichlorophenyl)-6-fluorome-
thylpyrimidine·DBTA salt (7). This material is available free of
charge via the Internet at http://pubs.acs.org.
AUTHOR INFORMATION
Corresponding Author
*Telephone: 01438 551179. E-mail: Matthew.D.Walker@GSK.
com.
■
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
We thank our R&D Product Development colleagues Keith
Allworth and David Stevens from Synthetic Chemistry,
Marcello Murru from Process Design, and Julien Patoor from
Analytical Sciences. We also thank our colleagues Paul Evans
and Jamie Russell from Investigational Material Supply.
added. The reaction was heated to 90
5 °C for 6 h and
cooled to 20 5 °C. The reaction mixture was diluted with
DMSO (15 mL) and water (7 mL) before water (20 mL) was
added over 10 h. The resultant solid was filtered and washed
with water (2 × 4 mL) to give a white solid (0.49 g, 24% yield).
(R)-2,4-Diamino-5-(2,3-dichlorophenyl)-6-fluoromethyl-
pyrimidine·DBTA Salt (7) from 2,4-Diamino-5-(2,3-dichlor-
ophenyl)-6-fluoromethylpyrimidine (6). 2,4-Diamino-5-(2,3-
dichlorophenyl)-6-fluoromethylpyrimidine (6) (80 g, 279
mmol) was added to a solution of dibenzoyl-L-tartaric acid
hydrate (105 g, 279 mmol) in a mixture of water (125 mL) and
REFERENCES
■
(1) (a) Nobbs, M. S.; Rodgers, S. J. Preparation of (R)-2,4-diamino-5-
(2,3-dichlorophenyl)-6-fluoromethylpyrimidine as a drug. World Patent
WO 9709317 A2, 1997. (b) Miller, A. A.; Nobbs, M. S.; Hyde, R. M.;
Leach, M. J. Preparation and formulation of phenylpyrimidines for
treating central nervous system disorders. European Patent EP 372934,
1996. (c) Alvaro, G.; Large, C. Pharmaceutical compositions comprising
3,5-diamino-6-(2,3-dichlorophenyl)-1,2,4-triazine or R-(−)-2,4-diamino-
5-(2,3-dichlorophenyl)-6-(fluoromethyl)pyrimidine and an NK1 antago-
nist. World Patent WO 2008090117 A1, 2008.
(2) The UK HSE has assigned a SLOD DTL (Specified Likelihood of
Death Dangerous Toxic Load) for ethyl fluoroacetate as 1840 ppm·
min. For example, it is estimated that, for exposure of a population to
80 ppm of ethyl fluoroacetate for 23 min (1840 ppm·min), 50% of this
population would be killed [Assessment of the Dangerous Toxic Load
(DTL) for Specified Level of Toxicity (SLOT) and Significant
Likelihood of Death (SLOD)]. http://www.hse.gov.uk/chemicals/
haztox.html (accessed April 19, 2013).
industrial methylated spirits (IMS) (590 mL) at 20
5 °C
followed by IMS (160 mL). The resultant suspension was
heated to 65 5 °C for 16 h, cooled to 20 5 °C over 2 h, and
aged for 3 h. The solid was filtered and the cake washed with a
mixture of IMS (272 mL) and water (48 mL), with IMS (2 ×
320 mL), and pulled dry. The solid was dried under vacuum at
45 5 °C to give a white crystalline solid (84.3 g). This solid
(80.0 g) was slurried in a mixture of DMSO (80 mL) and IMS
(136 mL) and heated to 65 5 °C to give a clear solution that
was cooled to 55 5 °C and seeded with (R)-2,4-diamino-5-
(2,3-dichlorophenyl)-6-fluoromethylpyrimidine·DBTA salt (7)
(80 mg) in IMS (0.5 mL). The mixture was aged for 15 min
,and IMS (584 mL) was added over 1 h. The slurry was cooled
to 20 5 °C over 70 min, aged for 2 h, and filtered. The cake
was washed with a mixture of DMSO (16 mL) and IMS (144
mL), with IMS (2 × 160 mL), and pulled dry. The solid was
dried under vacuum at 50 5 °C to give a white solid (59.6 g,
34.9% yield): 1H NMR (400 MHz, d6-DMSO) δ 8.01 (m, 8H),
7.71 (m, 4H), 7.67 (dd, J = 8.1, 1.5 Hz, 1H), 7.41 (t, J = 7.8 Hz,
1H), 7.25 (dd, J = 7.6, 1.5 Hz, 1H), 6.76 (br s, 2H), 6.52 (br s,
2H), 5.83 (s, 2H), 4.87 (dd, J = 15.9, 11.3 Hz, 1H), 4.75 (dd, J
= 15.9, 11.3 Hz, 1H). 13C NMR (100 MHz, d6-DMSO) δ 167.5
(C), 164.7 (C), 162.2 (C), 160.6 (C), 154.8 (d, J = 15.3 Hz,
C), 134.2 (CH), 133.8 (CH), 132.6 (C), 132.2 (C), 131.5
(CH), 130.5 (CH), 129.3 (CH), 128.9 (CH), 128.7 (C), 128.5
(CH), 105.2 (d, J = 3.5 Hz, C), 81.9 (d, J = 167.9 Hz, CH2),
(3) Other chiral acids were evaluated in the salt formation, but only
DBTA was shown to give any diastereoselectivity in the crystallisation
of the respective salt.
(4) The recrystallisation of enatiomerically enriched GW273225X 1
was performed under carefully temperature controlled conditions in
order to avoid excessive epimerisation of the chiral atropisomeric axis.
A graph of the effect of temperature and time on the enantiomeric
excess of the product was generated and used to define the processing
conditions.
̌
́
(5) Silhar, P.; Pohl, R.; Votruba, I.; Hocek, M. Org. Biomol. Chem.
2005, 3001−3007.
(6) Alternative deoxyfluorinating agents such as Deoxo-fluor (bis-(2-
methoxyethyl)aminosulphur trifluoride) were also investigated without
success.
(7) Treatment of the mesylate 12 with NaI in acetone cleanly
afforded the corresponding iodide in good yield. Unfortunately,
attempts to form the racemate 6 from the iodide by treatment with
CsF failed to yield any of the desired product.
(8) Kim, D. W.; Song, C. E.; Chi, D. Y. J. Org. Chem. 2003, 68,
4281−4285.
(9) (a) Kirk, K. L. Org. Process Res. Dev. 2008, 12, 305−321. (b) Lal,
G. S.; Pez, G. P.; Syvret, R. G. Chem. Rev. 1996, 96, 1737−1755.
F
dx.doi.org/10.1021/op4001753 | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
(10) Efforts were made to develop alternative routes to the enol ether
14; however use of more reactive acetate sources (acetic anhydride or
acetyl chloride) did not give the desired enol 13 and instead gave the
corresponding enol acetate, which did not undergo the desired
reaction with guanidine hydrochloride. Attempts to synthesise the enol
ether 14 directly using triethylorthoacetate under Brønsted or Lewis
acidic conditions also failed to show any reaction.
(11) Alternative commercial electrophilic fluorinating agents such as
Selectfluor and N-fluoropyridinium tetrafluoroborate salts were also
investigated without success.
(12) The racemic 2,4-diaminopyrimidine 6 prepared by this new
route shows excellent purity, containing only small amounts of the
difluorinated analogue and none of the des-fluorinated analogue.
Although the largest impurity going into the guanidination was
residual enol ether 14, this reacted comparatively slowly in the
guanidination reaction and was removed in the subsequent isolation.
(13) Attempts at generation of fluoroacetic acid or fluoroacetyl
chloride in organic solvent were unsuccessful due to either water
solubility or instability of the desired compound.
(14) Saunders, B. C.; Stacey, G. J. J. Chem. Soc. 1948, 1773−1779.
(15) Initial studies showed it was possible to convert the isolated enol
4 into the enol ether 5 using triethyl orthopropionate and acetic acid.
However, telescoping this methodology with the new CDI/sodium
fluoroacetate synthesis of the enol 4 resulted in significant retro-
Claisen reaction, which was attributed to the presence of imidazole in
the reaction mixture.
(16) (a) Parker, J. S.; Moseley, J. D. Org. Process Res. Dev. 2008, 12,
1041−1043. (b) Moseley, J. D.; Brown, D.; Firkin, C. R.; Jenkin, S. L.;
Patel, B.; Snape, E. W. Org. Process Res. Dev. 2008, 12, 1044−1059.
(c) Parker, J. S.; Bower, J. F.; Murray, P. M.; Patel, B.; Talavera, P. Org.
Process Res. Dev. 2008, 12, 1060−1077.
(17) The hazard of sodium fluoroacetate is comparable to that of
ethyl fluoroacetate, both being highly toxic. The level of risk from the
two compounds was therefore judged to be largely affected by the
ability to control exposure to the materials. A more detailed risk
assessment of both processes would be required in order to judge this
accurately.
(18) The quoted yields for synthesis of GW273225X 1 do not
include recycling of the unwanted enantiomer, which could be
recovered by freebasing the unwanted diastereoisomer of the salt and
thermally racemising (S)-2,4-diamino-5-(2,3-dichlorophenyl)-6-fluoro-
methylpyrimidine by heating as a slurry in toluene.
G
dx.doi.org/10.1021/op4001753 | Org. Process Res. Dev. XXXX, XXX, XXX−XXX