Development of a scaleable synthesis of a geminal dimethyl tertiary amine as an inhaled muscarinic antagonist for the treatment of COPD

Article abstract of DOI:10.1021/op200233r

An efficient and scalable process for the synthesis of muscarinic antagonist, PF-3635659 1, is described, illustrating redesign of an analogue-targeted synthesis which contained a scale-limiting rhodium-activated C-H amination step. The final route includes a reproducible modified Bouveault reaction which has not previously been reported on a substrate of this complexity, or on such a scale with over 5 kg of the requisite gem-dimethylamine prepared via this methodology.

Full text of DOI:10.1021/op200233r

Article
pubs.acs.org/OPRD
Development of a Scaleable Synthesis of a Geminal Dimethyl
Tertiary Amine as an Inhaled Muscarinic Antagonist for the
Treatment of COPD
Barry R. Dillon,*,†,§ Dannielle F. Roberts,*,‡,⊥ David A. Entwistle,† Paul A. Glossop,‡ Craig J. Knight,†
Daniel A. Laity,† Kim James,‡ Celine F. Praquin,† Ross S. Strang,‡ and Christine A. L. Watson‡
‡Department of Worldwide Medicinal Chemistry, †Department of Pharmaceutical Sciences, Pfizer Global Research and Development,
Sandwich Laboratories, Ramsgate Road, Kent CT13 9NJ, U.K.
S
* Supporting Information
ABSTRACT: An efficient and scalable process for the synthesis of muscarinic antagonist, PF-3635659 1, is described, illustrating
redesign of an analogue-targeted synthesis which contained a scale-limiting rhodium-activated C−H amination step. The final
route includes a reproducible modified Bouveault reaction which has not previously been reported on a substrate of this
complexity, or on such a scale with over 5 kg of the requisite gem-dimethylamine prepared via this methodology.
Retrosynthesis of target 1 was first performed with analogue
generation in mind; as such, a late-stage intermediate such as
alcohol 2 (Figure 1) was desirable.
INTRODUCTION
■
Chronic obstructive pulmonary disease (COPD) is charac-
terised by airflow limitation that is both progressive and not
fully reversible, resulting in chronic deterioration of lung
function.1 COPD is currently the fourth largest cause of death
in the U.S. and is projected to become the third leading cause
of death worldwide by 2020.1 Bronchodilator drugs are com-
monly used to manage the symptoms of COPD, and one class of
these prescribed for symptom control are inhaled muscarinic
receptor antagonists. Five distinct muscarinic receptors (M1−M5)
are known in humans, three of which are present in the
human lung (M1−M3). The M3 receptor is located on airway
smooth muscle cells where it mediates contractile responses.2
PF-3635659 1 is a potent antagonist of the M3 receptor and,
as such, is a suitable, once-daily, inhaled treatment for COPD
that has advanced to phase II clinical studies. Herein we de-
scribe the process development which led to a synthetic route
amenable for large-scale preparation.
Literature precedents suggested that azetidinols could be
conveniently formed by reaction of primary amines with epi-
halohydrins,3 and hence, it was introduction of the sterically
encumbered gem-dimethyl moiety that was envisaged to present
the greatest challenge. Under basic conditions phenylacetoni-
triles are known to be excellent nucleophiles which can react
with Michael acceptors such as acrylonitrile or alkyl acrylates.4
This knowledge might lead to supposition that Ritter chem-
istry,5 similar to that performed in our laboratories previously,6
would be a suitable way to access the primary gem-dimethylamine
(Figure 2); however, for this particular series such chemistry
Figure 2. Suggested Ritter chemistry.
proved to be far less reliable with complex mixtures produced in
the key amination step. Instead, an alternative electrophile was
sought and identified in the form of protected cyclic sulphamidate
6 (Scheme 1).7
RESULTS AND DISCUSSION
Synthesis began with Boc protection of cyclic sulphamidate 5
under standard conditions. Diphenylacetonitrile 4 was then
■
Received: August 25, 2011
Published: January 6, 2012
Figure 1. Retrosynthetic analysis of PF-3635659 1.
© 2012 American Chemical Society
195
dx.doi.org/10.1021/op200233r | Org. ProcessRes. Dev. 2012, 16, 195−203

Organic Process Research & Development
Article
Scheme 1. First-generation discovery route
deprotonated and offered to protected compound 6 as a
nucleophile. This reaction proceeded with gratifyingly high
conversion and the subsequent ring-opened intermediate was
deprotected with acid in situ to afford key gem-dimethylamine 3
as its hydrochloride salt. Following a salt break, further reaction of
3 with epichlorohydrin provided late-stage intermediate 7, also as
its hydrochloride salt, which concurrently enabled an array of
analogue syntheses.
For the purposes of the lead target, alcohol 7 could be
mesylated under standard conditions followed by displacement
with allyl protected resorcinol 8.8 Allylation was the preferred
protection strategy over methylation as it would allow us to
avoid the use of toxic demethylating agents such as boron
tribromide. The original intent from this point was to unmask
the primary amide through nitrile hydrolysis and then to finally
unveil the phenol by palladium-mediated deallylation. On ex-
posure of protected compound 9 to the preferred basic hydro-
lysis conditions, however, enol ether 10 was unexpectedly
obtained. This isomerisation negated the need for late-stage use
of palladium and meant we were instead able to perform a mild
acidic cleavage which provided us with first access, after basic
workup, to target 1 in its parent form on 100-mg scale.
In general terms this strategy looked very promising for large
scale, with good yields and well-documented reagents through-
out. Our attention now shifted to the long-term provision of
cyclic sulphamidate 5 which had previously been obtained from
an in-house bulk supply by the method depicted in Scheme 2.
lyzing isocyanate 11; however, this could be managed by careful
cooling and controlled addition of formic acid. The second was
regrettably more detrimental when it was found that the rhodium-
activated C−H amination, as described by Du Bois et al.,9 demon-
strated poor reproducibility in our hands. Isolated yields
of 5 fluctuated, and this outcome appeared to worsen with
increasing scale with only a maximum of 50% achieved on
greater than 10 g. Various attempts at optimisation were made,
including a screen of alternative rhodium catalysts, but no reliable
improvement was secured. Hence, an alternative approach became
necessary.
This led us to reevaluate how we might utilise Michael addi-
tion as a means to construct our four-carbon skeleton; however,
we would need a robust alternative to the aforementioned Ritter
chemistry.5 Ultimately the option we settled upon was a modi-
fied Bouveault reaction. The Bouveault reaction10 was originally
reported in 1904 and involves reaction of an alkyl Grignard with
a tertiary amide to give an aldehyde or ketone, plus a secondary
amine. In addition, there is commonly a side product identified
in these reactions which corresponds to the geminal dialkyl-
amine, and it was noted by our colleagues some 17 years pre-
vious to our endeavours that addition of an oxophilic Lewis acid
species (such as titanium or zirconium tetrachloride) to these
reactions can promote this side product as the major out-
come (Figure 3).11 This approach showed huge potential for
Scheme 2. Synthesis of cyclic sulphamidate 5
Figure 3. Bouveault (i) and modified Bouveault (ii) reactions.
installation of the gem-dimethyl functionality in our lead
compound, and so we proceeded along this course to generate
the required amide starting material (Scheme 3).
Diphenylacetonitrile 4 was again employed as the nucleo-
phile, this time partnered with tert-butyl acrylate 13. Michael
addition proceeded smoothly with subsequent deprotection
affording crystalline acid 14 in good yield. This could then be
Problems unfortunately arose at this stage for two distinct
reasons. The first was the exothermic tendency when hydro-
196
dx.doi.org/10.1021/op200233r | Org. ProcessRes. Dev. 2012, 16, 195−203

Organic Process Research & Development
Article
Scheme 3. Second-generation discovery route
coupled with azetidinol 15, available commercially as the
hydrochloride salt, to provide key amide 16, which was found
to crystallise from ethyl acetate.
towards either solvation or solvent occlusion, depending on the
solvent. A solvated form would not be suitable to progress,12
in particular due to this being an inhalation candidate with a
requirement for lactose stability.13
We were now ready to attempt the decisive modified
Bouveault step. For our Lewis acid we chose to use zirconium
tetrachloride, which can be purchased as a weighable solid and
demonstrates superior handling properties over its fuming,
liquid titanium counterpart. Initial trials were promising, pro-
viding desired tertiary amine 7 in 30−40% yields when using a
single equivalent of zirconium tetrachloride and 4 equiv of
methylmagnesium chloride. It seemed, however, that reagent
stoichiometry was critical to optimising yield and that this needed
to be considered on a substrate-specific basis. It was found to be
optimal in this case to use 2 equiv of Lewis acid and 9 equiv of
Grignard reagent in order to achieve quantitative conversion and
reproducibly high isolated yields. Gratifyingly, common inter-
mediate 7 was also seen to be crystalline on milligram scale, pre-
senting another useful optimisation handle for scale-up.
With key amine 7 in hand, the route previously outlined in
Scheme 1 could be mirrored using allyl-protected resorcinol 8.
Alternatively the end game could be altered to incorporate 3-
methoxyphenol 17 (Scheme 3). This proved advantageous
mainly due to the commercial availability of 17 versus literature
precedented preparation of allyl ether 8.8 This did of course
bring back into play the issue of late-stage use of boron
tribromide for the final deprotection, and this would need to be
further addressed as we proceeded. The final compound 1 could be
cleanly isolated and converted to its hydrochloride salt; however,
thermal analysis and single-crystal X-ray data of preliminary samples
suggested that the material in this salt form had a propensity
As we now homed in on nomination of compound 1 as a
potential candidate, we began to undertake a thorough review
of the route, with a view to facilitating scale-up towards multi-
kilogram manufacture. Although the existing process had thus
far been robust for its purpose, further opportunities for im-
provement could be identified. These were triaged to highlight
the two highest priority transformations as being the de-
methylation (in which the avoidance of boron tribromide was
strongly desirable) and the modified Bouveault reaction. For
the latter, it was felt that the modified Bouveault reaction was a
suitable method to quickly scale to fund early-phase studies;
however, it was realized that the current process could be
difficult to scale in terms of both practical and safety con-
siderations. Additional areas for improvement involved the
avoidance of all column chromatography, identification of stable
isolable intermediates, and a final recrystallisation that avoided
solvation.
Whilst the current route offered convenient divergence for
analogue screening, it was felt that a more targeted dis-
connection strategy could be applied. Disconnection at the
amide bond (Figure 4) offered increased convergence. Addi-
tionally it allowed the replacement of unprotected azetidinol
15, which had proven difficult to source on scale, with the more
readily available and cost-effective benzhydryl-protected mesy-
lated azetidinol 24.
Figure 4. Amide disconnection strategy.
197
dx.doi.org/10.1021/op200233r | Org. ProcessRes. Dev. 2012, 16, 195−203

Organic Process Research & Development
Article
Ideally this route would be compatible with application of the
modified Bouveault reaction to amide 20, in which protection
of the phenol could be avoided (and thus save a deprotection
step). Thus, opening investigations into this alternative route
focused on the direct coupling of resorcinol 25 with mesylated
azetidinol 24 (Scheme 4).
crystallization. Attempts to exchange solvent to improve
isolation were untenable due to instability with even mild
warming. Direct drop of the amine product 23 as a salt even-
tually offered a way forward with amine 23 being isolated as a
hemi-oxalate salt in 84% yield (5.4 kg over two batches) and
excellent purity, with palladium levels detected at only 1 ppm
(progressing to none detected after subsequent steps).
Acid 14 was prepared by Michael Addition, as previously
described (Scheme 3), followed by deprotection with lithium
hydroxide in 87% yield over two steps (12.1 kg). Coupling with
amine-oxalate salt 23 was carried out with 1-(3-dimethylamino-
propyl)-3-ethylcarbodiimide hydrochloride (WSCDI) and cata-
lytic 4-dimethylaminopyridine (DMAP) in propionitrile
(EtCN). Alternative coupling agents of carbonyldiimida-
zole (CDI) or boronic acid/trimethoxyborane under thermal
conditions failed to offer good conversion. Both DMAP and
1-hydroxy-benxotriazole (HOBt) proved beneficial additives
in combination with WSCDI, with DMAP chosen on scale in
consideration of sourcing and safety. Propionitrile offered
linearity between reaction and workup washes before solvent
exchange into methanol facilitated crystallisation of amide 21
in good yield 71% (7.1 kg). Interestingly salt break of amine-
oxalate salt 23 was not required prior to coupling. It was
found that addition of an extra 0.5 equiv of WSCDI allowed
for decomposition of the oxalic acid prior to reaction and
thus conferred a simplified process.
With 21 in hand our focus shifted to the modified Bouveault
reaction of this more extensively functionalised amide. Pleasingly,
following the previous procedure of treatment with 2 equiv of
zirconium tetrachloride and 9 equiv of methylmagnesium
chloride, amide 21 converted to desired gem-dimethylamine 18
in encouraging yields (50−65%). Limitations to the procedure
included cryogenic conditions, requirement to decant due to
significant inorganic precipitation, excessive volumes upon
workup (∼100 mL/g), and the requirement for column
chromatography.
Initially we looked at order of addition; we were interested in
previous investigations which suggested that combining methyl
Grignard and zirconium tetrachloride formed an in situ tetra-
methylzirconium species.14 Experimentation showed that, in
line with our current process, best results were obtained when
undertaking this pre-preparation, as opposed to adding all
reagents simultaneously; hence, we believe the active species for
this transformation is tetramethylzirconium. The report also
suggests instability of this species upon exposure to elevated
temperatures which again was reflected in poorer yields when
the reaction temperaure exceeded 0 °C and also when using a
preprepared solution which was stored for an extended period
of time at 0 °C (>8 h). Pleasingly, when tetramethylzirconium
was prepared fresh, we found that the reaction profile was
stable and consistent up to 0 °C, and thus, we could avoid
cryogenic conditions. Safety studies of this reagent reassured us
of its stability with a TSU of the zirconium tetrachloride/
methylmagnesium chloride mixture showing no significant
Scheme 4. Phenol displacement of mesyl-azetidinol
Initial attempts at ether formation proved unsuccessful with
complex reaction profiles including over-reaction to dimeric
byproduct. A range of bases, solvents, and reagent equivalents
were screened; however, in each instance a clean reaction profile
could not be obtained. To accelerate progress we thus moved at
this point to adopt a phenol protection strategy. As anisole de-
protections generally require harsh conditions and form
alkylating byproduct, it was felt that pursuing more labile
alternatives may offer some advantages. Hence, acetate-protected
analogue 26 was trialled in coupling with mesylated azetidine 24;
however, this proved unstable to various reaction conditions
producing reaction profiles analogous to that observed with
resorcinol 25. Benzoyl-protected resorcinol was also attempted,
but limited success led us to revert to 3-methoxyphenol 17 with
a view to enable the anisole deprotection at a later stage. The
mesylate displacement with this partner progressed cleanly with
potassium carbonate in propionitrile with the solvent also
facilitating a separative workup, with subsequent isolation of aryl
ether 29 directly by filtration from methanol in a yield of 71%
(11.2 kg). Acetonitrile also afforded clean conversion, but an
extensive workup procedure made this option less desirable.
With aryl ether 29 in hand, a method was sought for
benzhydryl deprotection to afford amine 23. A range of Lewis
acid and catalytic hydrogenation conditions were
attempted with the benzhydryl group proving relatively inert.
Our only positive result was obtained with 20% weight
Pearlman’s catalyst (Pd(OH)2/C) in ethanol at 60 psi and
25 °C which provided clean conversion (Scheme 5). The amine
product 23, however, proved unstable to heat which presented
some difficulties in isolation. Telescoping amine 23 into the
amide coupling directly after catalyst filtration was attempted
and found to give high conversion with clean reaction profiles.
Unfortunately, subsequent crystallization of amide 21 was not
robust, and even when crystal growth was achieved, the isolated
yields of 21 were poor (<40%). It is postulated that the
diphenylmethylene byproduct liberated by the benzhydryl pro-
tecting group under hydrogenation conditions inhibited
Scheme 5. Preparation of amide 21
198
dx.doi.org/10.1021/op200233r | Org. ProcessRes. Dev. 2012, 16, 195−203

Organic Process Research & Development
Article
Scheme 6. Modified Bouveault reaction
exotherm or gas evolution upon heating to 200 °C at a rate of
2 °C per minute. The addition of the Grignard reagent to zirco-
nium tetrachloride was highly exothermic with an RC1 mea-
surement of −942.7 kJ/mol and an associated adiabatic tem-
perature rise of 80 K. This would be sufficient to boil the
solvent if added as a single dose without additional controls;
hence, the process was made plausible via slow addition and full
reactor cooling. The gas flow was also measured during this
process with the flow of evolved gas below the limit of detec-
tion during the Grignard addition.
Our attentions were next drawn to the reaction stochiometry.
Previous investigations on less complex substrates could be
carried out with 1 equiv zirconium tetrachloride and 4 equiv
methyl Grignard;11 however, a screen for our substrate showed
a steep decline in conversion to desired product below 1.5 equiv
of zirconium tetrachloride and 7 equiv of methyl Grignard with
2.1 and 9 equiv, respectively, offering the best conversion.
Key byproducts for this step were methyl ketones 30 and 31
(Scheme 6) arising respectively from breakdown of the complex
after monomethyl addition or from nucleophilic attack at the
nitrile position of the desired product.
With respect to reaction workup it was noted that a water
quench was highly exothermic and off-gassed. Head-space an-
alysis showed the gas to be methane, and gas flow monitoring
showed that its liberation could be dose controlled. In addition,
the water quench resulted in excessive precipitation which
led to the initial decanting procedure. Acetic acid diluted in
2-methyltetrahydrofuran (MeTHF) was found to give control of
the exotherm and afforded a much more manageable suspension.
The associated gas evolution during this addition was measured
as −1018.9 kJ/mol of amide 21 over a 45 min addition at
−5 °C and the associated adiabatic temp rise as 45 K. Gas
evolution was detected during this addition with the rate mea-
sured as 3.9 L/min/mol of amide 21, and the total amount of
gas evolved measured as 90 L/mol. The safety studies showed a
controlled quench offered a manageable exotherm and gas
evolution within the acceptable limits of the pilot plant for
safety and environment. Alternative quenches were attempted
to minimize gas evolution and exothermic events, notably by
the use of acetone or ethyl acetate; however, the profile and iso-
lated yield through the original acetic acid quench were far
superior in both respects. Subsequent addition of aqueous am-
monium chloride resulted in a biphasic solution that could now
be manipulated by a standard workup. Whilst the maximum
workup volume was still high at ∼41 volumes, this approach
offered significant improvement to throughput over the original
procedure. Ultimately, solvent exchange into ethanol provided
a suspension with product obtained by filtration in good yield
and purity (average yield of 66% on 5.1-kg scale).
gem-Dimethylamine intermediate 18 offered a common
intermediate to the early-phase route (Scheme 3) from which
we could directly seek to enable the remaining demethylation
and hydrolysis steps. The two procedures could potentially be
undertaken in either order (Scheme 7).
Whilst the original sequence in early phase had been hydro-
lysis followed by demethylation, we now noted that the hydro-
lysis was generally a cleaner transformation. In addition, most
common demethylation procedures result in alkylating by-
product with strict control required. Hence, our preference was
to first undertake the demethylation of 18 to afford phenol 32
and then nitrile hydrolysis to primary amide 1.
Screening a range of demethylation conditions found no
alternative hits for conversion of ether 18 to phenol 32 other
than boron tribromide which was undesired due to safety and
toxicity issues. We were encouraged, however, by a report from
Merck15 in which DL-methionine had been employed as a de-
methylating agent with the advantage that the methylated-
methionine byproduct (colloquially referred to as vitamin U) is
not highlighted as an alkylating agent/genotoxic alert.16 Whilst
the reaction proved very specific to variations in solvent, tem-
perature, and acid, conditions of 30 °C for 2−3 days in neat
methanesulfonic acid (MsOH) and 3 equiv of DL-methionine
afforded clean demethylation of 18. To avoid an extensive basic
quench of the neat acid upon workup, which would require
Scheme 7. Alternative end game approaches to target molecule 1
199
dx.doi.org/10.1021/op200233r | Org. ProcessRes. Dev. 2012, 16, 195−203

Organic Process Research & Development
Article
strict control of exothermic events, the reaction was first diluted
with isopropyl acetate/MeTHF. The methanesulfonic acid
could then be largely washed out with water before quenching
any residual acid and exchanging solvent to enable crystalliza-
tion from toluene of phenol 32 in 68% yield (3.4 kg). It was
noted at this stage that the level of zirconium in isolated phenol
32 was 13 ppm. This was significantly lower than the level of
228 ppm identified in the previous gem-dimethyl intermediate
18 and now complied with all the requisite specifications.
The final step involving hydrolysis of nitrile 32 to primary
amide 1 was undertaken using potassium hydroxide in tert-amyl
alcohol. Optimal conditions were found to be 18 equiv of
potassium hydroxide at 100 °C from which only two impurities
were observed, carboxylic acid 33 and decarboxylated by-
product 34 (Figure 5).
In summary we have presented a concise and scalable
synthesis of PF-3635659 1 as its hydrochloride salt which can
support the required clinical studies toward the advancement of
this phase II candidate (Scheme 8).17 We have demonstrated
two diverse alternatives for incorporation of a nontrivial, ste-
rically encumbered geminal dimethyl amine moiety which
encompass a broad functional group compatibility. The modi-
fied Bouveault reaction in particular has been performed on a
synthetically advanced intermediate and on a scale which, to
our knowledge, has never previously been attempted. We have
also implemented a mild and technically uncomplicated method
of unmasking a methylated phenol at a late stage in the synthesis,
avoiding the notorious genetic toxicty problems of the more
commonly used boron tribromide. The final route is now
chromatography free, aided in no small part by continuous
reappraisal of potential crystalline intermediates throughout the
entire process from first synthesis to clinical preparation, with all
steps developed to be suitable for safe and standard processing.
EXPERIMENTAL SECTION
■
1H NMR spectra were recorded on a Bruker 400 MHz NMR
spectrometer. HPLC analyses were performed using a reverse
phase technique. LC/MS analysis was performed using the follow-
ing system; Hewlett-Packard 1100 with SB C18 3.0 mm ×50 mm,
1.8 μm particles; mobile phase consisting of solvent A, 0.05% TFA
in water, solvent B, 0.05% TFA in acetonitrile. 0 min =
5% solvent B; 3.5 min = 100% solvent B; 4.5 min = 100%
solvent B; 4.6 min = 5% solvent B; run time 5 min; column
temperature 50 °C; 225 nm; with Waters Micromass ZQ
2000/4000 mass detector. Combustion analyses were
performed by Warwick Analytical Service, University of
Warwick Science Park, The Venture Centre, Sir William
Lyons Road, Coventry CV4 7EZ, U.K. Experimental data is
provided for the largest batch produced.
Figure 5. Impurities from nitrile hydrolysis of 32.
Both impurities could be minimised by controlling reaction
time and temperature. It was found that the level of de-
carboxylation increased significantly above 110 °C; therefore,
the boiling point of tert-amyl alcohol (bp 102 °C) offered
convenient control when compared to the initially utilised
3-methyl-3-pentanol (bp 123 °C). With the level of impurities
minimised, both 33 and 34 could be purged upon isolation of
target molecule 1 as the crude hydrochloride salt in 83% yield
(2.85 kg).
Whilst this crude hydrochloride salt met HPLC purity
specification, it was found to have occluded tert-amyl alcohol
up to 6 wt %, which, given its non-ICH classification, presented
complications and a desire to limit. A reslurry in aqueous
methylethyl ketone (MEK) purged the alcohol to acceptable
levels of <1 wt % in 76% yield over the two steps, providing
2.6 kg of target molecule 1 as a hydrochloride salt for toxico-
logical and clinical trials.
Preparation 1: 1-Benzhydryl-3-(3-methoxyphenoxy)-
azetidine 29. To propionitrile (130.5 L) at 20 °C ( 5 °C)
was added methanesulfonic acid 1-benzhydrylazetidin-3-yl ester
24 (14.5 kg, 45.7 mol, 1 equiv) resulting in a solution.
Potassium carbonate (7.6 kg, 54.8 mol, 1.2 equiv) was added in
single portion and to the resulting slurry was added a
Scheme 8. Enabled route to deliver initial toxicity and clinical batches of PF-3635659·HCl salt
200
dx.doi.org/10.1021/op200233r | Org. ProcessRes. Dev. 2012, 16, 195−203

Organic Process Research & Development
Article
preformed solution of 3-methoxyphenol (6.8 kg, 54.8 mol, 1.2
equiv) in propionitrile (14.5 L) maintaining a temperature of
20 °C ( 5 °C). The mixture was heated to 80 °C ( 5 °C) for
18 h, whereupon the reaction was deemed to have reached
completion with <5% area starting material by HPLC.
The reaction was cooled to 20 °C ( 5 °C) then 1 M NaOH
(72.5 L) was added, maintaining a temperature of 20 °C
( 5 °C) by controlled addition. The resulting solution was
stirred for 15 min before being allowed to separate. The layers
were partitioned, and the organic layer was washed with 1 M
NaOH (2 × 43.5 L) and aqueous brine (24.0 kg NaCl in
72.5 L water) at 20 °C ( 5 °C). The organic layer was placed
under atmospheric distillation conditions and concentrated to low
volume with a distillate temperature of 97 °C before solvent
exchange into methanol (2 × 362.5 L), resulting in a final volume
of ∼60 L and a distillate temperature of 65 °C. The resulting
suspension is cooled to 0 °C ( 5 °C) for 5 h and then filtered
through a pressure filter. The cake was washed with methanol
(87.0 L) and then dried under vacuum at 50 °C ( 5 °C) for 10 h
to give 1-benzhydryl-3-(3-methoxyphenoxy)azetidine 29 as a
white solid (11.2 kg, 71%).
reaction was then placed under vacuum distillation at 50 °C
( 5 °C) and the tert-butanol removed, leaving an aqueous
concentrate of ∼22 L. Water (128 L) was added and the
reaction warmed to 40 °C ( 5 °C) to achieve dissolution. The
solution was then cooled to 30 °C ( 5 °C) before extracting
with dichloromethane (3 × 20 L). The dichloromethane layers
were discarded, and the aqueous layer was distilled under vacuum
at 45 °C ( 5 °C) to remove ∼5 L of distillate (all traces of
dichloromethane). The mixture was then cooled to 25 °C
( 5 °C) and acidified to pH 2 by addition of 37% conc. HCl
(6.9 L), resulting in a suspension. After stirring at 25 °C ( 5 °C)
for 1 h the suspension was filtered under vacuum and the cake
washed with water (10 L) and dried under vacuum at 40 °C
( 5 °C) to give 4-cyano-4,4-diphenylbutyric acid 14 as a white
solid (6.1 kg, 87%).
1HNMR (400 MHz, DMSO-d6) δ: 2.21−2.32 (m, 2H),
2.71−2.83 (m, 2H), 7.29−7.38 (m, 2H), 7.38−7.50 (m, 8H),
12.30−12.51 (br s, 1H); 13CNMR (100 MHz, DMSO-d6) δ:
30.6, 33.5, 50.8, 121.8, 126.5, 128.0, 129.1, 139.4, 172.8.
Preparation 4: 5-[3-(3-Methoxyphenoxy)azetidin-1-yl]-
5-oxo-2,2-diphenylpentanenitrile 21. To a suspension of
4-Cyano-4,4-diphenyl-butyric acid 14 (3.15 kg, 11.9 mol,
1 equiv) in propionitrile (31.5 L) at 20 °C ( 5 °C) were
added 3-(3-methoxyphenoxy)azetidine hemioxalate 23 (2.66 kg,
11.9 mol, 1 equiv) and 4-dimethylaminopyridine (145 g, 1.2 mol,
0.1 equiv). The reaction was stirred for 10 min before adding
WSCDI (3.42 kg, 17.8 mol, 1.5 equiv) resulting in slight
effervescence, a 10 °C exotherm and dissolution to a solution. The
reaction was recooled to 20 °C ( 5 °C) and after 3 h the reaction
was deemed complete by HPLC with <3% area of starting
material 23. Aqueous 2 M HCl (13.0 L) was added, maintaining a
temperature of 25 °C ( 5 °C); then the biphasic mixture was
stirred for 10 min at 20 °C ( 5 °C) before separating and washing
the organic layer with aqueous 2 M NaOH (15.8 L) and water
(2 × 15.8 L). The organic layer was placed under atmospheric
distillation conditions, initially collecting the distillate at 84 °C
and ending at 99 °C and low volume before solvent exchange
into methanol (2 × 50 L), resulting in a final volume of ∼25 L and
a distillate temperature of 65 °C. The resulting suspension is
cooled to 20 °C ( 5 °C) and stirred for 5 h and then 0 °C
( 5 °C) for 2 h before filtering in a pressure filter, washing the
cake with methanol (12.6 L), and then drying under vacuum at
45 °C ( 5 °C) for 12 h to afford 5-[3-(3-methoxyphenoxy)-
azetidin-1-yl]-5-oxo-2,2-diphenylpentanenitrile 21 as a white solid
(3.6 kg, 71%).
1H NMR (400 MHz, DMSO-d6) δ: 2.94−3.02 (m, 2H),
3.58−3.66 (m, 2H), 3.70 (s, 3H), 4.51 (s, 1H), 4.79−4.87
(m, 1H), 6.36−6.42 (m, 2H), 6.49−6.55 (m, 1H), 7.16−7.22
(m, 3H), 7.29 (t, J = 7.52 Hz, 4H), 7.40−7.49 (m, 4H);
13CNMR (100 MHz, DMSO-d6) δ: 55.0, 59.6, 65.6, 76.9, 100.8,
106.5, 106.8, 127.0, 127.1, 128.4, 130.1, 142.3, 157.9, 160.6.
Anal. Calcd for C23H23NO2: C 79.97; H 6.71; N 4.05. Found:
C 80.15; H 6.76; N 4.02.
Preparation 2: 3-(3-Methoxyphenoxy)azetidine Hemi-
oxalate 23. To ethanol (102 L) were added 1-benzhydryl-3-
(3-methoxyphenoxy)azetidine 29 (5.1 kg, 14.7 mol, 1 equiv)
and Pd(OH)2/C (20 wt % on carbon) (1.02 kg, 20 wt %), and
the vessel was placed under hydrogenation conditions of 60 psi
H2 and 20 °C ( 5 °C) for 72 h whereupon the hydrogen up-
take had significantly decreased and HPLC reported <5%
starting material. The reaction mixture was filtered through filter
aid and concentrated to ∼22 L under vacuum distillation using
full vacuum whilst maintaining the temperature below 35 °C. To
the resulting solution at 20 °C ( 5 °C) was added oxalic acid
(0.66 kg, 7.4 mol, 0.5 equiv) resulting in a thick suspension and a
temperature rise of 4 °C. The mixture was left to stir for 5 h at
20 °C ( 5 °C) then 3 h at 0 °C ( 5 °C) before filtering through
a pressure filter and washing the cake with ethanol (20 L) and
then drying under vacuum at 50 °C ( 5 °C) for 16 h to give 3-
(3-methoxyphenoxy)azetidine hemioxalate 23 as a white solid
(2.68 kg, 81%).
1HNMR (400 MHz, DMSO-d6) δ: 2.04−2.23 (m, 2H),
2.65−2.83 (m, 2H), 3.73 (s, 3H), 3.74−3.79 (m, 1H), 3.92−
4.01 (m, 1H), 4.21−4.32 (m, 1H), 4.38−4.48 (m, 1H), 4.90−
5.01 (m, 1H), 6.34−6.44 (m, 2H), 6.52−6.62 (m, 1H), 7.16−
7.24 (m, 1H), 7.31−7.37 (m, 2H), 7.38−7.49 (m, 8H);
13CNMR (100 MHz, DMSO-d6) δ: 27.6, 33.4, 51.0, 54.5, 55.1,
56.5, 65.2, 101.0, 106.6, 107.1, 121.9, 126.4, 128.0, 128.4, 128.6,
129.1, 130.2, 139.5, 139.6, 157.4, 160.6, 170.4. Anal. Calcd for
C27H26N2O3: C 76.03; H 6.14; N 6.57. Found: C 76.12; H
6.19; N 6.38.
Preparation 5: 5-[3-(3-Methoxyphenoxy)azetidin-1-yl]-5-
methyl-2,2-diphenylhexanenitrile 18. All vessels were rinsed
and boiled out with tetrahydrofuran and then analysed to
ensure <0.1 wt % water before commencing with the reaction.
The ZrCl4 was freshly weighed into dry CTC immediately
before use, due to its hygroscopic nature.
1H NMR (400 MHz, DMSO-d6) δ: 3.73 (s, 3H), 3.76−3.85
(m, 2H), 4.11−4.23 (m, 2H), 4.97−5.09 (m, 1H), 6.34−6.44
(m, 2H), 6.52−6.61 (m, 1H), 7.15−7.23 (m, 1H); 13CNMR
(100 MHz, DMSO-d6) δ: 52.6, 55.1, 66.4, 101.0, 106.7, 107.1,
130.3, 157.3, 160.6.
Preparation 3: 4-Cyano-4,4-diphenylbutyric Acid 14.
Diphenylacetonitrile (5.1 kg, 26.4 mol, 1 equiv) and tert-
butanol (26 L) were heated to 45 °C ( 5 °C). To the resulting
solution were added a suspension of KOH (145 g, 2.64 mol,
0.1 equiv) in methanol (0.55 L) and then tert-butyl acrylate
(4.16 kg, 31.7 mol, 1.2 equiv) over 30 min. The reaction was
heated at 60 °C ( 5 °C) for 5 h, whereupon the reaction was
complete with HPLC indicating <5% starting material. Upon
cooling to 20 °C ( 5 °C) a solution of lithium hydroxide
hydrate (2.88 kg, 68.6 mol, 2.6 equiv) in water (26 L) was
added and the batch heated at reflux (89 °C) for 5 h. The
To tetrahydrofuran (35.0 L) at −15 °C to −10 °C under
a N2 atmosphere was carefully added ZrCl4 (4.4 kg, 18.9 mol,
201
dx.doi.org/10.1021/op200233r | Org. ProcessRes. Dev. 2012, 16, 195−203

Organic Process Research & Development
Article
2.3 equiv) over three equal portions, maintaining a temperature
below 0 °C. The potential exotherm from a single addition of
the full 4.4 kg of ZrCl4 was estimated to be 44 °C, and the
observed exotherm for each portion of 1.5 kg of ZrCl4 was
8.2 °C. The resulting brown suspension was cooled to −5 °C
( 5 °C) before the addition of a solution of methylmagnesium
chloride (3 M in tetrahydrofuran, 24.6 L, 73.9 mol, 9 equiv) over
1 h, maintaining the temperature below 0 °C. The potential heat
rise from the Grignard addition was estimated as 88 °C; however,
this was readily controlled by addition rate. A line wash of
tetrahydrofuran (20 L) was applied. The resulting black slurry was
stirred at −5 °C ( 5 °C) for 30 min before adding a pre-
formed solution of 5-[3-(3-methoxyphenoxy)azetidin-1-yl]-
5-oxo-2,2-diphenylpentanenitrile 21 (3.5 kg, 8.21 mol, 1 equiv)
in tetrahydrofuran (10.5 L) maintaining the temperature below
0 °C ( 5 °C) (potential for 20 °C rise controlled by addi-
tion rate). An additional line wash with tetrahydrofuran (17.0 L)
was carried out before the mixture was stirred for 4 h at −5 °C
( 5 °C) resulting in reaction completion which was noted by
>80% area of desired product by HPLC and <8% methylketone
30. The reaction was quenched with a preformed solution of
2-methyltetrahydrofuran (35.0 L) and glacial acetic acid (4.2 L,
73.9 mol, 9.0 equiv) maintaining the temperature below −5 °C
and venting the methane off-gas through a hydrogen vent. The
reaction was set to 10 °C and water (50 L) was added to the
reaction followed by a solution of aqueous ammonium chloride
(3.5 kg) in water (35.0 L) to ensure full quench maintaining the
temperature below 10 °C. The mixture was then stirred for 16 h at
15 °C ( 5 °C) then allowed to separate and the bottom layer
discarded. At 15 °C ( 5 °C) the organic layer was washed with a
solution of dilute aqueous sodium hydroxide prepared from 40%
NaOH (2.4 L) and water (32.6 L) and then with a solution of
brine (3.5 kg NaCl in 35.0 L water). The resulting organic layer
was concentrated to low volume under atmospheric distillation
conditions with a final distillate temperature of 70 °C and then
was solvent exchanged into ethanol (3 × 50.0 L portions),
resulting in a final suspension of ∼18 L in volume and a con-
sistent boiling point of 80 °C. The mixture was cooled to 20 °C
( 5 °C) for 5 h and then to 0 °C ( 5 °C) for 2 h before the
resulting thick suspension was filtered through a pressure filter and
the cake washed with ethanol (14 L). The solid was dried in a
vacuum oven at 50 °C ( 5 °C) under vacuum for 12 h to give
5-[3-(3-methoxyphenoxy)azetidin-1-yl]-5-methyl-2,2-diphenylhex-
anenitrile 18 as a white, crystalline solid (2.65 kg, 74%).
completion identified by <3% area starting material by HPLC.
The mixture was cooled to 10 °C ( 5 °C) then isopropyl acetate
(26.5 L) added followed by the slow addition of water (26.5 L).
(Note − there is a theoretical 27 °C exotherm, though the
solution temperature is controlled to <25 °C by addition rate).
2-Methyltetrahydrofuran (5.3 L) was then added and the layers
were mixed for 15 min at 20 °C ( 5 °C) then separated. The
aqueous layer is back-washed with isopropyl acetate (13.3 L).
The organic layers were combined and washed at 20 °C ( 5 °C)
with 1 M NaOH (26.5 L), water (13.3 L), and brine (2.65 kg
NaCl in 13.3 L water). The organic layer was placed under atmo-
spheric distillation conditions (reflux = 86 °C) and solvent ex-
change conducted into toluene. Completion was noted by boiling
point (target distillate temperature of >105 °C was targeted from
small scale, although a distillate temperature of 112 °C was
achieved on scale using a total of 155.8 kg of toluene over two
portions) and a final reaction volume of ∼4 L/kg. The mixture
was cooled to 20 °C ( 5 °C) at a rate of 1 °C/min and granulated
for 5 h before cooling to 5 °C ( 5 °C) and granulating for a
further 2 h. The suspension was filtered through a pressure filter
and the cake washed with cold toluene (10.6 L) and dried at
50 °C ( 5 °C) under reduced pressure to afford 5-[3-(3-hydro-
xyphenoxy)azetidin-1-yl]-5-methyl-2,2-diphenylhexanenitrile 32 as
a white crystalline solid (1.65 kg, 64%).
1HNMR (400 MHz, DMSO-d6) δ: 0.98 (s, 6H), 1.22 (dt, J =
4.2, 8.3 Hz, 2H), 2.46−2.54 (m, 2H), 2.99−3.05 (m, 2H),
3.38−3.45 (m, 2H), 4.62 (t, J = 5.7 Hz, 1H), 6.20−6.25 (m,
2H), 6.34−6.38 (m, 1H), 7.03 (t, J = 8.1 Hz, 1H), 7.29−7.35
(m, 2H), 7.38−7.47 (m, 8H), 9.4 (s, 1H); 13CNMR (100 MHz,
DMSO-d6) δ: 20.6, 33.2, 34.9, 51.4, 52.9, 53.3, 65.0, 101.8,
105.1, 108.1, 122.2, 126.5, 127.8, 129.0, 130.0, 140.1, 158.0,
158.6. Anal. Calcd for C28H30N2O2: C 78.84; H 7.09; N 6.57.
Found: C 78.72; H 7.12; N 6.51.
Preparation 7: 5-[3-(3-Hydroxyphenoxy)azetidin-1-yl]-5-
methyl-2,2-diphenylhexanamide 1 (Hydrochloride Salt).
To a suspension of 5-[3-(3-hydroxyphenoxy)azetidin-1-yl]-
5-methyl-2,2-diphenylhexanenitrile 32 (3.05 kg, 7.15 mol,
1 equiv) and tert-amyl alcohol (30.5 L) at 22 °C was added
flake-KOH (7.42 kg, 132.3 mol, 18.5 equiv) and the mixture
heated to 80 °C ( 5 °C) for 2.5 h (note a change of
morphology occurs, resulting in a thick suspension). The
suspension was then heated to 99 °C ( 5 °C) at a rate of
0.5 °C/min and then held at this temperature for 24 h,
whereupon the reaction was deemed complete with <5% starting
material remaining by HPLC. The reaction was cooled to 20 °C
( 5 °C) and quenched slowly with water (30.5 L), maintaining
the temperature below 25 °C. The layers were mixed for 1 h
and separated; the upper organic layer was washed with water
(30.5 L). The organic layer was adjusted to a pH range of 6.8−8
with 37% hydrochloric acid (0.66 L, 1.1 equiv), maintaining a
temperature below 25 °C, before the layers were separated. The
upper layer was passed through a 2 μm filter to spec-free the
solution, and all additional solvent and reagents from this point were
passed through 2 μm filters (as were reactors and equipment
cleaned and verified to spec-free prior to use). To the clarified upper
organic layer was added tert-butylmethyl ether (9.3 L) and 37%
hydrochloric acid (0.63 L, 1.05 equiv) maintaining a temperature
below 25 °C ( 5 °C) and the mixture stirred for 19 h at 20 °C
( 5 °C) resulting in a thick suspension. The suspension was filtered
through a pressure filter and the cake washed with tert-butylmethyl
ether (22.3 L) and dried at 70 °C ( 5 °C) under reduced pressure
in a tray drier for 12 h to afford crude the hydrochloride salt of 5-[3-
(3-hydroxyphenoxy)azetidin-1-yl]-5-methyl-2,2-diphenylhexanamide
1HNMR (400 MHz, DMSO-d6) δ: 0.89 (s, 6H), 1.23 (dt, J =
8.21, 4.10 Hz, 2H), 2.49−2.53 (m, 2H), 2.98−3.08 (m, 2H),
3.38−3.47 (m, 2H), 3.72 (s, 3H), 4.69 (t, J = 5.57 Hz, 1H),
6.31−6.44 (m, 2H), 6.48−6.59 (m, 1H), 7.09−7.22 (m, 1H),
7.28−7.36 (m, 2H), 7.37−7.50 (m, 8H); 13CNMR (100 MHz,
DMSO-d6) δ: 20.5, 33.1, 34.8, 51.4, 52.8, 53.3, 55.0, 65.1, 100.7,
106.6, 106.6, 122.2, 126.5, 127.6, 129.0, 130.1, 140.1, 158.0, 160.5;
LRMS ESI m/z 441 [M + H]+. Anal. Calcd for C29H32N2O2
(contains 0.3 mol water): C 79.06; H 7.32; N 6.36. Found: C
78.13; H 7.25; N 6.11.
Preparation 6: 5-[3-(3-Hydroxyphenoxy)azetidin-
1-yl]-5-methyl-2,2-diphenylhexanenitrile 32. 5-[3-(3-
Methoxyphenoxy)azetidin-1-yl]-5-methyl-2,2-diphenylhexane-
nitrile 18 (2.65 kg, 6.01 mol, 1 equiv) was added to me-
thanesulfonic acid (13.3 L), resulting in an 18 °C exotherm. The
mixture was stirred at 20 °C ( 5 °C) for 30 min, resulting in a
viscous solution. DL-methionine (2.69 kg, 18.03 mol, 3 equiv)
was added resulting in an 8 °C exotherm, the mixture was heated
to 30 °C ( 5 °C) for 72 h whereupon the reaction had reached
202
dx.doi.org/10.1021/op200233r | Org. ProcessRes. Dev. 2012, 16, 195−203

Organic Process Research & Development
Article
1 (PF-3635659·HCl) as a white crystalline solid with 6 wt %
occluded tert-amyl alcohol (2.85 kg, 83%).
REFERENCES
■
(1) Rabe, K. F.; Hurd, S.; Anzueto, A.; Barnes, P. J.; Buist, S. A.;
Calverley, P.; Fukuchi, Y.; Jenkins, C.; Rodriguez-Roisin, R.; van Weel,
C.; Zielinski, J. Am. J. Respir. Crit. Care Med. 2007, 176 (6), 532−555.
(2) Coulson, F. R.; Fryer, A. D. Pharmacol. Ther. 2003, 98 (1),
59−69.
(3) (a) Okutani, T.; Kaneko, T.; Masuda, K. Chem. Pharm. Bull 1974,
22 (7), 1490−1497. (b) Denmark, S. E.; Forbes, D. C.; Hays, D. S.;
DePue, J. S.; Wilde, R. G. J. Org. Chem. 1995, 60 (5), 1391−1407.
(c) Krishna Reddy, V. V. R. M.; Kishore Babu, K.; Ganesh, A.;
Srinivasulu, P.; Madhusudhan, G.; Mukkanti, K. Org. Proc. Res. Dev.
2010, 14 (4), 931−935.
(4) (a) Dei, S.; Budriesi, R.; Sudwan, P.; Ferraroni, M.; Chiarini, A.;
Garnier-Suillerot, A.; Manetti, D.; Martelli, C.; Scapecchi, S.; Teodori, E.
Bioorg. Med. Chem. 2005, 13 (4), 985−998. (b) Bongartz, J.-P.; Buntinx,
M.; Coesemans, E.; Hermans, B.; Van Lommen, G.; Van Wauwe, J.
Bioorg. Med. Chem. Lett. 2008, 18 (21), 5819−5823. (c) Hartmann, R. W.;
Batzl, C.; Pongratz, T. M.; Mannschreck, A. J. Med. Chem. 1992, 35 (12),
2210−2214.
A stirred suspension of the hydrochloride salt of 1 (2.85 kg)
in methylethylketone (27.0 L) and water (1.5 L) was heated to
reflux (73 °C) and the resulting suspension stirred for 16 h
before cooling to 20 °C ( 5 °C) at a rate of 1 °C per min and
then stirring for 5 h. The suspension was filtered on a pressure
filter and the cake washed with methylethylketone (19.9 L)
before drying at 70 °C ( 5 °C) in a tray drier under reduced
pressure for 12 h to afford the purified hydrochloride salt of 5-
[3-(3-hydroxyphenoxy)azetidin-1-yl]-5-methyl-2,2-diphenyl-
hexanamide 1 (PF-3635659·HCl) as a white crystalline solid
(2.60 kg, 91%). Combined yield over both steps is 76%.
1HNMR (400 MHz, DMSO-d6) δ: 1.10 (s, 6H), 1.22−1.34
(m, 2H), 2.42−2.55 (m, 2H), 3.28−3.40 (m, 2H), 3.65−3.88
(m, 2H), 4.70−4.80 (m, 1H), 5.55−5.70 (br s, 2H), 6.23−6.36
(m, 2H), 6.45−6.53 (m, 1H), 7.03−7.12 (m, 1H), 7.19−7.39
(m, 10H); 13CNMR (100 MHz, DMSO-d6) δ: 19.8, 19.9, 31.7,
53.0, 53.9, 59.5, 61.0, 62.5, 64.1, 102.2, 105.1, 109.2, 126.4,
127.8, 128.8, 128.9, 130.1, 130.2, 143.6, 156.7, 156.9, 158.9,
174.7; LRMS ESI m/z 445 [M + H]+. Anal. Calcd for
C28H32N2O3·HCl: C 69.91; H 6.91; N 5.82; Cl 7.37. Found: C
69.75; H 7.05; N 5.66; Cl 7.16.
(5) (a) Ritter, J. J.; Minieri, P. P. J. Am. Chem. Soc. 1948, 70, 4045−
4048. (b) Ritter, J. J.; Kalish, J. J. Am. Chem. Soc. 1948, 70, 4048−4050.
(6) Glossop, P. A.; Lane, C. A. L.; Price, D. A.; Bunnage, M. E.;
Lewthwaite, R. A.; James, K.; Brown, A. D.; Yeadon, M.; Perros-Huguet,
C.; Trevethick, M. A.; Clarke, N. P.; Webster, R.; Jones, R. M.;
Burrows, J.; Feeder, N.; Taylor, S. C. J.; Spence, F. J. J. Med. Chem. 2010,
53 (18), 6640−6652.
(7) Bower, J. F.; Rujirawanich, J.; Gallagher, T. Org. Biomol. Chem.
2010, 8 (7), 1505−1519.
ASSOCIATED CONTENT
■
(8) Gozzo, F. C.; Fernandes, S. A.; Rodrigues, D. C.; Eberlin, M. N.;
Marsaioli, A. J. J. Org. Chem. 2003, 68 (14), 5493−5499.
(9) (a) Du Bois, J.; Espino, C. G.; Fioro, K. W.; Kim, M. J. Am. Chem.
Soc. 2004, 126 (47), 15378−15379. (b) Chan, P. W. H.; Che, C.-M.;
Zhang, J. Tetrahedron Lett. 2005, 46 (32), 5403−5408. (c) Brodsky, B. H.;
Du Bois, J. Chem. Commun. 2006, 45, 4715−4717. (d) Brodsky, B. H.;
Du Bois, J.; Espino, C. G.; Fioro, K. W. Tetrahedron 2009, 65 (16),
3042−3051.
S
* Supporting Information
Experimental procedures and spectral data for all early-stage
synthetic routes. This material is available free of charge via the
Internet at http://pubs.acs.org.
AUTHOR INFORMATION
■
(10) Bouveault, L. Bull. Soc. Chim. Fr. 1904, 31 (3), 1306−1322.
(11) Denton, S. M.; Wood, A. Synth. Lett. 1999, 1, 55−56.
(12) (a) Hilfilker, R., Ed. Polymorphism in the Pharmaceutical
Industry; Wiley-VCH: Weinheim, Germany, 2006; Chapter 8.
(b) Vippagunta, S. R.; Brittain, H. G.; Grant, D. J. W. Adv. Drug
Delivery Rev. 2001, 48 (1), 3−26.
(13) (a) Baldrick, P.; Bamford, D. G. Food Chem. Toxicol. 1997,
35 (7), 719−733. (b) Traini, D.; Young, P. M.; Jones, M.; Edge, S.;
Price, R. Eur. J. Pharm. Sci. 2006, 27 (2−3), 243−251.
(14) Berthold, H. J.; Groh, G. Angew. Chem., Int. Ed. Engl. 1966,
5 (5), 516.
Corresponding Author
*Telephone: +44-1803-884325. E-mail: barry.dillon@
astrazeneca.com. Telephone: +44-1344-414205. E-mail:
dannielle.roberts@syngenta.com.
Present Addresses
§Astrazeneca - Brixham Environmental Laboratory, Freshwater
Quarry, Brixham, Devon TQ5 8BA, U.K.
⊥Syngenta, Jealott’s Hill International Research Centre,
Bracknell, Berkshire RG42 6EY, U.K.
(15) Melillo, D. G.; Larsen, R. D.; Mathre, D. J.; Shukis, W. F.;
Wood, A. W.; Colleluori, J. R. J. Org. Chem. 1987, 52 (53), 5143−
5150.
(16) Irie, H.; Fujii, N.; Ogawa, H.; Yajima, H.; Fujino, M.; Shinagawa, S.
Chem. Commun. 1976, 22, 922−923.
(17) Glossop, P. A.; Watson, C. A. L.; Price, D. A.; Bunnage, M. E.;
Middleton, D. S.; Wood, A.; James, K.; Roberts, D. F.; Strang, R. S.;
Yeadon, M.; Perros-Huguet, C.; Clarke, N. P.; Trevethick, M. A.;
Machin, I.; Stuart, E. F.; Evans, S. M.; Harrison, A. C.; Fairman, D. A.;
Agoram, B.; Burrows, J. L.; Feeder, N.; Fulton, C. K.; Dillon, B. R.;
Entwistle, D. A.; Spence, F. J. J. Med. Chem. 2011, 54 (19), 6888−
6904.
ACKNOWLEDGMENTS
■
With thanks to David Fengas, M. Jonathan Fray, and Keith
Reeves for their work on the optimisation of the modified
Bouveault chemistry; Katie Bainbridge, Trish Costello, David
Cox, Ben Greener, and Louise Marples for the analogue
synthesis which was crucial to advancing this series; the tech-
nology group, in particular Steven Fussell and Stuart Field,
for providing key data from reaction screens; and to members
of the safety group and hydrogenation laboratories, namely
Claire Crook, John Deering, and Trevor Newbury who pro-
vided project support and useful discussions. We also thank
everyone who provided technical input on the project,
including Dean Brick and Mark Taylor. Thanks also to
members of the Materials Science group, Chris Dallman,
Cheryl Doherty, Neil Feeder, and Toby Payne-Cook, for
their work on identifying and characterising the final hydro-
chloride salt form and to Rachel Osborne for her generous
contribution in writing experimentals.
203
dx.doi.org/10.1021/op200233r | Org. ProcessRes. Dev. 2012, 16, 195−203