Development of a scalable synthesis of GSK183390A, a PPAR α/γ agonist

Article abstract of DOI:10.1021/op700164t

A scalable synthesis of GSK183390A, a PPAR α/γ agonist, is described. This synthesis is highlighted by (1) a regioselective formal 1,3-dipolar cycloaddition reaction between an enamine and a nitrile inline dipole to form a 1,3,5-trisubstmited pyrazole and (2) a regioselective amidomethylation of an o-cresol derivative using 2-chloro-JV- hydroxymethylacetamide.

Full text of DOI:10.1021/op700164t

Organic Process Research & Development 2007, 11, 1032–1042
Development of a Scalable Synthesis of GSK183390A, a PPAR r/γ Agonist
Lynette M. Oh, Huan Wang,* Susan C. Shilcrat, Robert E. Herrmann, Daniel B. Patience, P. Grant Spoors, and Joseph Sisko
Chemical DeVelopment, GlaxoSmithKline, 709 Swedeland Road, King of Prussia, PennsylVania 19406, U.S.A.
Abstract:
A scalable synthesis of GSK183390A, a PPAR r/γ agonist, is
described. This synthesis is highlighted by (1) a regioselective
formal 1,3-dipolar cycloaddition reaction between an enamine and
a nitrile imine dipole to form a 1,3,5-trisubstituted pyrazole and
(2) a regioselective amidomethylation of an ꢀ-cresol derivative
using 2-chloro-N-hydroxymethylacetamide.
Figure 1. GSK183390.
dipole 13, generated by base-promoted dehydrohaloge-
nation of hydrazonoyl halides 12,2 and alkyne 14 or its
Introduction
Peroxisome proliferator-activated receptors (PPARs) are a
equivalent (Scheme 2). Since the appropriate pairing of
the reacting partners would be crucial for ensuring both
regioselectivity and reactivity,3 we started our investigation
by screening different dipole precursors and dipolarophiles.
1,3-Dipolar Cycloaddition Reaction. The preparation of
the coupling partners is outlined in Scheme 3. Hydra-
zonoyl halides 18 and 19 were prepared by condensation
of methylhydrazine with ethyl glyoxylate, followed by
treatment of the resulting hydrazone 17 with NBS or NCS.
Four electron-rich olefins were synthesized5 for use as
potential alkyne equivalents to be screened in the cy-
cloaddition reaction.
group of ligand-activated transcription factors of the nuclear
receptor family, including three distinct subtypes commonly
designated as PPARR, PPARγ and PPARδ. These PPARs bind
to DNA sequences and regulate gene expression, thus influenc-
ing cellular functions such as lipid metabolism, cell proliferation
and differentiation. Consequently, the activation of PPARs
provides promising therapeutic opportunities for the manage-
ment of diseases related to these cellular activities. GSK183390A
(1, Figure 1) has recently emerged as a potent dual agonist of
PPARR/γ and a candidate for treatment of dyslipidemia.1 Its
progression into clinical studies has necessitated a synthetic route
amenable to large-scale synthesis.
Shown in Scheme 1 is the original synthetic route to 1 via
the coupling of benzylamine 10 and pyrazole 11. Two main
drawbacks exist for this route: (1) The condensation between
methylhydrazine and diketone 4 produced a mixture of the
desired 1,3,5-trisubstituted pyrazole 5 and the undesired isomer
6 in a 1:2 ratio, with the desired isomer 5 as the minor product.
Not only did the poor regioselectivity render the yield of this
step unacceptably low, but it also entailed chromatographic
purification of the product. (2) The synthesis of benzylamine
10 required six linear steps, which seemed lengthy for a
molecule of its size and relative simplicity.
With the coupling partners in hand, the cycloaddition reaction
was evaluated to identify the ideal pairing. The results are
summarized in Table 1.
Chloride 19 proved to be an unreactive partner with all the
dipolarophiles screened, providing minor conversion to the
desired product even after extended reaction times. The reactions
of bromide 18 with 4-tert-butylphenylacetylene 14,TMS enol
ether 20 or enol phosphate 21 were sluggish and gave complex
mixtures. Among several unidentified components, the major
product proved to be tetrazine 25, the head-to-tail dimer of the
dipole. However, the pairing of hydrazonoyl bromide 18 and
the more electron-rich morpholine enamines (22, 23) provided
good reactivity and readily gave clean pyrazole formation.
Our search for a new route, therefore, focused on a
regioselective method for the pyrazole formation to deliver
isomer 5 exclusively, as well as a more efficient synthesis of
benzylamine 10. This report discloses a new and scalable route
to 1, which was employed in a pilot plant campaign to produce
40 kg of active pharmaceutical ingredient (API).
(2) (a) Oh, L. M. Tetrahedron Lett. 2006, 47, 7943. (b) Tanaka, K.; Maeno,
S.; Mitsuhashi, K. J. Heterocycl. Chem. 1985, 22, 565. (c) Tanaka,
K.; Maeno, S.; Mitsuhashi, K. Chem. Lett. 1982, 4, 543. (d) Shawali,
A. S.; Párkányi, C. J. Heterocycl. Chem. 1980, 17, 833.
(3) For general reviews, see: (a) Bianchi, G.; DeMicheli, C.; Gandolfi,
R. 1,3-Dipolar Cycloadditions involving X)Y Groups. In The
Chemistry of Double Bonded Functional Groups; Patai, S., Ed.;
Interscience: London, 1977: pp 369–532. (b) Carruthers, W. Cycload-
dition Reactions in Organic Synthesis; Pergamon Press, 1990, New
York.
Results and Discussion
(4) It is known that electron-rich olefins promote reactivity since this is
a LUMO-dipole, HOMO-dipolarophile controlled reaction.3b
(5) (a) TMS enol ether 20: Martin, V. A.; Murray, D. H.; Pratt, N. E.;
Zhao, Y.-B.; Albizati, K. F. J. Am. Chem. Soc. 1990, 112, 6965. (b)
Enol phosphate 21: Whitehead, A.; Moore, J. D.; Hanson, P. R.
Tetrahedron Lett. 2003, 44, 4275. (c) Enamine 22: Modified procedure
of Carlson, R.; Nilsson, A.; Stromquist, M. Acta Chem. Scand. 1983,
B37, 7. Hünig’s base was added to ensure complete reaction. (d)
Enamine 23: Schiehser, G. A.; White, J. D. J. Org. Chem. 1980, 45,
1864.
Synthesis of Pyrazole 11. In order to address the serious
regiochemical issue in the pyrazole formation step, we
proposed an alternative route employing a formal 1,3-
dipolar cycloaddition reaction between a nitrile imine
* To whom correspondence should be addressed. Tel: 610-270-5362. Fax:
610-270-4022 . Email: huan.2.wang@gsk.com.
(1) Faucher, N. E.; Martres, P. PCT Int. Appl. WO 2005049578 A1, 2005.
1032
•
Vol. 11, No. 6, 2007 / Organic Process Research & Development
10.1021/op700164t CCC: $37.00
2007 American Chemical Society
Published on Web 11/16/2007

Scheme 1. Original synthesis of GSK183390a
a (a) NaOEt, 0 to 80 °C, 99%; (b) CH3NHNH2, EtOH, 90 °C, 5, 22%, 6, 45%; (c) NH2OH· HCl, NaOAc, EtOH, rt, 93%; (d) ammonium formate, Pd/C, reflux
MeOH, 50%; (e) aq HBr, 97%; (f) Boc2O, CH2Cl2, Et3N, 96%; (g) K2CO3, DMF, ethyl bromoisobutyrate, 69%; (h) TFA, CH2Cl2, 82%; (i) NaOH, rt, 98%; (j) SOCl2,
toluene, 80 °C, then Et3N, 10, rt, 96%; (k) NaOH, 80 °C, 73%.
Scheme 2. Proposed 1,3-dipolar cycloaddition for
regioselective pyrazole formation.
Scheme 3. Preparation of hydrazonoyl halides and alkyne
equivalentsa
Furthermore, the regioselectivity of the cycloaddition was
completely controlled by the substitution pattern of the enamine:
when 1,1-disubstituted morpholine enamine 22 was used, only
the desired 1,3,5-pyrazole isomer 5 was produced, whereas the
1,2-disubstituted morpholine enamine 23 led exclusively to the
1,3,4-pyrazole isomer 24.6 Since the pairing of 1,1-disubstituted
enamine 22 and hydrazonoyl bromide 18 in the 1,3-dipolar
cycloaddition fulfilled both requirements of reactivity and
regioselectivity, scale-up and optimization activities on each
were initiated.7
a (a) MeNHNH2 (16), THF; (b) NBS or NCS, THF; (c) LiHMDS, TMSCl,
THF, –25 °C; (d) LDA, ClPO(OEt)2, THF, –25 °C; (e) TiCl4, morpholine,
toluene, 80 °C; (f) SO2 · pyridine, DMSO, CH2Cl2; (g) morpholine.
to a THF solution of methylhydrazine at 0 °C and subsequent
heating to 50 °C. Examination of this procedure revealed a
number of serious process safety problems.
Scale-Up Studies of Hydrazonoyl Bromide 18 Synthesis. The
laboratory preparation of hydrazone 17 called for the addition
of ethyl glyoxylate (purchased and used as a solution in toluene)
First, the product showed limited thermal stability. The
adiabatic calorimetry on hydrazone 17 detected decomposition
initiating at 75 °C, and such early decomposition put significant
constraints on the permissible operating temperature of the
process. There was also significant sample-to-sample variability
in the onset of decomposition depending on the purity,
especially in highly colored samples.
(6) The reaction involving the 1,2-disubstituted enamine 23 also appeared
to be more sluggish than the corresponding 1,1-disubstituted enamine
22. For studies on the relative reactivity of olefins as dipolarophiles,
including the effects of 1,1- versus 1,2-substitution, see: Jäger, V.;
Colinas, P. A. Nitrile Oxides. In 1,3-Dipolar Cycloadditions; Padwa,
A., Pearson,W. H., Eds.; Interscience: New York, 2002; pp 376–378.
(7) A brief base screen revealed that triethylamine was the best base,
whereas Hünig’s base and N-methylmorpholine gave incomplete
conversions for the cycloaddition reaction. A solvent screen also
indicated that a number of solvents could be employed, including
EtOAc, toluene, THF and CH2Cl2.
Second, calorimetry on the original procedure showed that
the addition of ethyl glyoxylate resulted in a large but addition
Vol. 11, No. 6, 2007 / Organic Process Research & Development
•
1033

Table 1. Preliminary screen of the 1,3-dipolar cycloadditiona
developed in which the toluene solution of ethyl glyoxylate
was added to a methanol solution of methylhydrazine at
40–50 °C. The heat output (-75 kJ/mol of methylhydra-
zine; patr 53 °C, Figure 3) was now addition rate-limited,
while the boiling point of the MeOH cosolvent introduced
an upper temperature barrier that eliminated the possibility
of runaway decomposition of hydrazone 17.
The revised procedure also produced a reaction mixture
that was homogeneous throughout and allowed for online
spectroscopic monitoring by ReactIR. Figure 4 illustrates
the ReactIR monitoring of the pilot plant run. The toluene
solution of ethyl glyoxylate was added over 2.25 h.9 The
addition of ethyl glyoxylate (approximated by the growth
of toluene at 773 cm-1), the consumption of methylhy-
drazine (1621 cm-1) and the formation of hydrazone 17
(1710 cm-1) all occurred at a steady rate10 and reached a
plateau upon the completion of addition. A process sample
taken approximately 45 min after the end of addition
confirmed that the reaction was complete. The success of
ReactIR monitoring not only cut the cycle time by over
2 h but also reduced the plant personnel’s exposure to
toxic methylhydrazine by requiring a single sample.
This new procedure was executed successfully in our pilot
plant, producing 39.4 kg of hydrazone 17 (59–64% yield).
Moreover, hydrazone 17 produced by the revised procedure was
only slightly colored and showed a markedly improved thermal
stability profile, with the onset of decomposition at 124 °C
shown by adiabatic calorimetry.
The bromination of hydrazone 17 was then achieved
by slow addition of a methylene chloride (CH2Cl2) solution
of hydrazone 17 to a slurry of N-bromosuccinimide (NBS)
in ethyl acetate (EtOAc).11 After the reaction was over,
the slurry was filtered to remove succinimide, and the
product was stored as a cold solution in EtOAc/CH2Cl2.
Bromide 18 had only limited thermal stability when
isolated as an oil, exhibiting early and severe decomposi-
tion with a large exotherm (-722 J/g) during differential
scanning calorimetry (94–131 °C). Adiabatic calorimetry
also indicated the onset of decomposition at a relatively
low temperature (∼69 °C), prompting us to closely
monitor its behavior during the 1,3-dipolar cycloaddition
step (vide infra).
Scale-Up Studies of Enamine 22 Synthesis. Enamine 22 was
prepared from commercially available 4-tert-butyla-
cetophenone and morpholine with TiCl4 as a water
scavenger, Hünig’s base as an HCl scavenger and toluene
as the solvent. One difficulty encountered during scale-
up was that the preformed TiCl4–morpholine complex and
side products (TiO2 and the HCl salt of Hünig’s base)
formed sticky agglomerates that resulted in a very slow
a Chloride 19 was unreactive with all dipolarophiles screened.
rate-controlled exotherm (-63 kJ/mol of methylhydrazine).
However, the heat-up to 50 °C was accompanied by additional
heat output (-28 kJ/mol; patr8 25 °C) which started from 23
°C, increased at ∼35 °C, reached a maximum rate at 50 °C,
and then continued for another 15 min (Figure 2). The
significant heat accumulation, combined with the low thermal
stability of the product hydrazone 17, raised serious problems
of possible runaway decomposition.
A third issue also arose during the safety test run. It was
observed that uncontrolled precipitation of hydrazone 17
occurred during the heat-up stage, resulting in fouling on the
ReactIR probe and disabling the process monitoring. Consider-
ing the batch variation in the thermal stability profile of 17, it
was speculated that a better controlled crystallization might yield
product of more consistent quality.
It became clear that the key to an inherently safer
process was the temperature control, with a higher starting
temperature in order to accelerate the reaction and
minimize the thermal accumulation. On the other hand,
an upper boundary should also be placed on the system
temperature so as not to exceed the thermally stable range
of hydrazone 17. Therefore, a revised process was
(9) A Dynochem model was built to predict the addition time needed to
maintain reactor temperature at or below the 50 °C threshold. The
actual addition time in the plant correlated very well to the recom-
mended addition time of 2.5 h.
(10) The species at 1745 cm-1 is presumably the tetrahedral intermediate
(addition of methylhydrazine to ethyl glyoxylate) prior to the loss of
water. It initiated upon the addition of ethyl glyoxylate and reached
maximum at approximately 50% addition.
(11) This procedure resulted in an addition rate-controlled heat output.
Dynochem model predictions of the heat curve over the addition time
correlated well with plant data.
(8) Predicted adiabatic temperature rise.
1034
•
Vol. 11, No. 6, 2007 / Organic Process Research & Development

Figure 2. Heat output during hydrazone formation using the original procedure.
Figure 3. Heat output during hydrazone formation using the revised procedure.
Figure 4. ReactIR monitoring of hydrazone 17 formation in the pilot plant.
filtration. This was circumvented by adding solid sodium
sulfate at the beginning of the reaction (before the
formation of the TiCl4–morpholine complex) to give a
suspendable slurry of the reaction mixture which was
Vol. 11, No. 6, 2007 / Organic Process Research & Development
•
1035

easily filtered.12 The resulting filtrate, containing enamine
22, was distilled to remove excess morpholine and excess
Hünig’s base and stored with triethylamine (base for the
cycloaddition step).
Scheme 5. Formation of morpholine adduct 28
Scale-Up Studies of the 1,3-Dipolar Cycloaddition. During
our initial investigation, it was revealed by LC/MS that, upon
addition of triethylamine, bromide 18 was transformed im-
mediately into triethylamine adduct 26. Therefore, we reasoned
that this adduct was the immediate precursor to nitrile imine
dipole 13, which participated in the regioselective 1,3-dipolar
cycloaddition.
Scheme 4. Formation of triethylamine adduct 26
Scheme 6. New route to pyrazole 11
This speculation was supported by calorimetry data obtained
for the reaction. The cycloaddition reaction was carried out in
a step-wise fashion where an EtOAc solution of bromide 18
was treated with triethylamine at 40 °C. After holding the
reaction mixture overnight, a solution of enamine 22 was added
as a bolus, and the reaction mixture was held at 40 °C for
another 3 h. The reaction between triethylamine and bromide
18 produced an addition rate-limited exotherm (-76 kJ/mol of
enamine 22) with a patr of 10 °C, indicating a fast, moderately
exothermic transformation of bromide 18 to triethylamine
adduct 26. The bolus addition of enamine 22 produced a heat
spike, with heat output (-96 kJ/mol of enamine 22; patr 12
°C) continuing at a low rate for ∼1.5 h post-addition. The
overall enthalpy for the two processes, -172 kJ/mol of enamine
22, compared reasonably well to the results obtained from a
typical procedure (-211 kJ/mol of enamine 22; patr 27 °C).13
Although the fast consumption of bromide 18 alleviated our
concerns with its thermal instability (vide supra), its chemical
instability in the presence of nucleophilic organic bases caused
a different issue. It was found that the displacement of bromide
also occurred with morpholine.14 Unlike the triethylamine
adduct 26, the morpholine adduct 28 does not re-eliminate to
generate the dipole. In order to control its formation, the residual
morpholine from the enamine formation has to be removed by
thorough distillation. However, since morpholine is also gener-
ated in situ during the spontaneous elimination of pyrazoline
27 to pyrazole 5 (Scheme 5), the reaction conditions need to
be carefully controlled to minimize the formation of morpholine
adduct 28.
It was found that temperature control was again critical in
this process. The rate of the in situ elimination of morpholine
from 27 is faster at 70 °C, generating as much as 10–20% of
28. On the other hand, when the reaction was conducted at 40
°C, the formation of morpholine adduct 28 was minimized
(3–5%) while an acceptable rate of the cycloaddition reaction
was still maintained. Therefore, a solution of bromide 18 in
EtOAc/CH2Cl2 was added to a solution of enamine 22 and
triethylamine in toluene at 40 °C. After the reaction was
complete, hydrochloric acid was added to ensure complete
elimination of morpholine from pyrazoline 27 to form pyrazole
5 (Scheme 6). Morpholine adduct 28 was removed in the
aqueous layer during the workup. The organic layer containing
pyrazole 5 was then concentrated and diluted with THF, and
the hydrolysis to pyrazole acid 11 was effected by aqueous
lithium hydroxide. This procedure was scaled up in the pilot
plant to deliver a total of 28.1 kg of pyrazole acid 11 (53–56%
yield based on enamine 22).
(12) Glass coupon abrasion tests in the laboratory showed that the addition
of solid sodium sulfate would not damage glass-lined vessels, which
was confirmed by visual inspection after the run in the plant.
(13) The typical procedure consisted of adding a solution of bromide 18
to a solution of enamine 22 and triethylamine at 40 °C. The yield in
the step-wise calorimetry experiment was ∼10% lower, and the HPLC
profile was also messier than the typical runs.
(14) It is interesting to note that the displacement of bromide was not
observed with Hünig’s base, presumably due to its greater steric
bulk.
1036
•
Vol. 11, No. 6, 2007 / Organic Process Research & Development

Scheme 7. Proposed synthesis of benzylamine 10
Table 2. Regioselectivity in the amidomethylation of ꢀ-cresol
derivatives
Synthesis of Benzylamine 10. It was envisioned that a more
efficient synthesis of benzylamine 10 could start from inex-
pensive ꢀ-cresol, as shown in Scheme 7. Three chemical
transformations would be needed to install the functionalities
of 10: (1) amidomethylation of the arene, (2) alkylation of the
phenoxide, and (3) deprotection of the amide, in order to deliver
amine 10 for coupling with the pyrazole fragment.
Amidomethylation of o-Cresol DeriVatiVes. It was identified
at the outset that the feasibility of this new route hinged on the
para/ortho selectivity of the amidomethylation reaction.15
Therefore, a brief study was conducted with ꢀ-cresol derivatives
30–32 to investigate the regioselectivity of the amidomethylation
reaction (Table 2).
Initial amidomethylation of ꢀ-cresol with phthalimide
33 in toluene with catalytic amount of pTSA gave a ∼1:1
mixture of p-35 and o-35 at elevated temperatures (entry
1). However, strongly acidic conditions (H2SO4/HOAc)
allowed the reaction to proceed at a lower temperature
and improved the para:ortho ratio for ꢀ-cresol to ∼3:1
(entry 3). Anisole 31 gave a higher para:ortho ratio (14:
1) than ꢀ-cresol (entry 4), whereas reaction with less
electron-rich 32 required a higher temperature which
resulted in a lower selectivity (entry 5). Switching to the
more reactive chloroacetamide 34, however, successfully
installed the amide functionality on 32 with virtually
complete selectivity for the desired para-product (entry
7).
regioselectivity
p-35:o-35
entry reagent arene
conditions
1
2
3
4
5
6
7
33
33
33
33
33
34
34
30 pTSA (cat.), tol, 70–90 °C 0.8:1
31 pTSA (cat.), tol, 70–90 °C 4:1
30 H2SO4/HOAc (1/10), rt
31 H2SO4/HOAc (1/10), rt
3:1
14:1
32 H2SO4/HOAc (1/10), 60 °C 7:1
31 H2SO4/HOAc (1/10), rt
32 H2SO4/HOAc (1/10), rt
60:1
para only
Scheme 8. Alkylation of ꢀ-cresol with r-bromoisobutyrate
Not only did this study establish 2-chloro-N-hydroxym-
ethylacetamide 34 as a suitable amidomethylation reagent,
it is also worth noting that alkylated derivatives 31 and
32 gave higher para selectivity than the parent ꢀ-cresol.
This finding led to the conclusion that the alkylation of
ꢀ-cresol should precede amidomethylation in the synthetic
sequence.
Alkylation of o-Cresol. The original alkylation of ꢀ-cresol
in the synthesis of 10 employed R-bromoisobutyrate 36. This
procedure produced a mixture of desired ester product 32 and
methacrylate 37. A quick screen of base/solvent combinations
revealed that this reaction was sluggish and often stalled
(Scheme 8).
As an alternative, the more reactive chloretone (38) was
investigated as the alkylator in this reaction.16 The reaction
(15) For general reviews on amidomethylation, see: (a) Zaugg, H.; Martin,
W. Org. React. 1965, 14, 52. (b) Zaugg, H. E. Synthesis 1970, 49. (c)
Zaugg, H. E. Synthesis 1984, 85. (d) Zaugg, H. E. Synthesis 1984,
181.
(16) Corey, E. J.; Barcza, S.; Klotmann, G. J. Am. Chem. Soc. 1969, 91,
4782.
Vol. 11, No. 6, 2007 / Organic Process Research & Development
•
1037

Scheme 9. Alkylation of ꢀ-cresol with chloretone
Scheme 11. Alkylation of ꢀ-cresol with r-bromobutyric
acid
Scheme 10. Mechanism of chloretone alkylation
Scheme 12. New route to benzylamine 10
was near quantitative. The major side product was methacrylic
acid (40), which was difficult to separate from the desired
product 39 by simple extraction methods due to the similarity
in their pKa’s. However, it was found that methacrylic acid could
be removed by the addition of NaHSO3, presumably in a
Michael fashion to form water-soluble adduct 47, which allowed
the extraction of pure 39 into an organic solvent. The addition
of HSO3- to methacrylic acid was found to be very pH-
sensitive, with the optimal pH value between 6 and 7 (Scheme
11).
between ꢀ-cresol and chloretone proceeded much faster than
that with bromoester 36, producing the desired product 39, as
well as methacrylic acid (40) as the main side product (Scheme
9).
Deprotection of the Amide. With amide-acid 48 in hand from
the alkylation/amidomethylation sequence, the deprotection of
the amide functionality was accomplished in refluxing ethanol
using H2SO4.17 The free acid was concomitantly esterified to
give the desired free amine 10 in >90% yield (Scheme 12).
This new route provided a much more efficient synthesis of
the benzylamine fragment, and scale-up was conducted.
Scale-Up Studies of Benzylamine 10 Synthesis. Most of the
investigation during the scale-up focused on the ꢀ-cresol
alkylation stage and the subsequent workup. It was observed
during the calorimetry evaluation that the addition of R-bro-
moisobutyric acid caused a large (-300 kJ/mol), addition rate-
moderated exotherm. The reaction mixture was very thick,
which occasionally led to loss of agitation and, consequently,
However, ꢀ-cresol could not be completely consumed in this
reaction (Scheme 10). Upon treatment with base, chloretone is
converted into epoxide 41, which opens to give acid chlorides
42 and 43. Each acid chloride then either hydrolyzes to give
the corresponding acid (40, 39) directly or reacts with ꢀ-cresol
to give an ester (44 or 45), which then hydrolyzes slowly to
regenerate ꢀ-cresol. The intermediacy of esters 44 and 45 was
established by LC/MS analysis.
Although ꢀ-cresol was inexpensive, its presence in the
product mixture complicated purification of the product. Fur-
thermore, the isolated yield obtained from this reaction was only
65%. Therefore, focus was shifted to finding another alkylator
that would consume the starting material and provide a higher
yield.
Fortunately, it was found that R-bromoisobutyric acid
alkylated ꢀ-cresol smoothly in the presence of NaOH. The
phenol was consumed in 2 h at 50 °C in MEK, and the yield
(17) It is important to use chloroacetamide as the protecting group for ease
of removal: alcoholysis of the corresponding acetamide was ∼5 times
slower than that of 48.
1038
•
Vol. 11, No. 6, 2007 / Organic Process Research & Development

Figure 5. Heat output of the ꢀ-cresol alkylation reaction.
Scheme 13. Removal of side products from the ꢀ-cresol
alkylation
large heat spikes. The degradation of R-bromoisobutyric acid
to methacrylic acid was also observed post-addition and could
be monitored spectroscopically (Figure 5).
To avoid agglomeration and loss of agitation, alkylation of
ꢀ-cresol was carried out by slowly adding R-bromoisobutyric
acid to the sodium salt of ꢀ-cresol. The addition was completed
over 8 h in the plant. After the reaction was complete, the
reaction mixture was held at 50 °C to consume the excess
R-bromoisobutyric acid.
The resultant reaction mixture contained several components
besides the desired product: aldol condensation products from
methyl ethyl ketone, methacrylic acid, and occasionally excess
ꢀ-cresol. It was found that accurate control of the pH was crucial
for the separation of these components. After the reaction was
complete, water was added to the mixture and methyl ethyl
ketone was removed by distillation, resulting in an aqueous
solution of pH ) 14. A TBME wash was performed to remove
the MEK self-aldol condensation products. The pH was then
lowered from 14 to 6.5,18 NaHSO3 was added, and the mixture
was stirred at room temperature overnight. After the complete
consumption of methacrylic acid, the pH of the mixture was
adjusted further to 1.5 to fully protonate acid 39. Extraction
with TBME separated it from the methacrylic acid adduct,
which was left in the aqueous layer (Scheme 13).
In transitioning from the alkylation step to the amidomethy-
lation step, concerns arose with the potential interaction between
residual TBME, the extraction solvent in the alkylation stage,
and sulfuric acid, the cosolvent in the amidomethylation stage.
It was observed in the laboratory that the amidomethylation
was sluggish when 39 was contaminated with residual TBME.
Moreover, two side products were observed and assigned
structures 50 and 51 based on LC/MS information (Scheme
14). The apparent presence of isobutylene raised concerns about
possible uncontrolled decomposition of TBME under strongly
acidic conditions. Therefore, after TBME extraction of the
alkylation product, we elected to switch solvent from TBME
Scheme 14. TBME participation in the amidomethylation
reaction
(18) In the case of excess ꢀ-cresol being present in the reaction mixture,
the pH could first be lowered to 8.5 and the aqueous layer washed
with TBME to remove ꢀ-cresol. However, ꢀ-cresol was usually
consumed during our plant runs.
to HOAc, allowing for complete removal of TBME before the
addition of sulphuric acid and 2-chloro-N-hydroxymethylac-
etamide.
Vol. 11, No. 6, 2007 / Organic Process Research & Development
•
1039

Scheme 15. Formation of GSK183390A
evaporation, the crude mixtures were subjected to column
chromatogaphy using ethyl acetate/heptane as eluent to provide
the following. 1-Methyl-5-(4-tert-butylphenyl)-2H-pyrazole-3-
(carboxylic acid ethyl ester) (5): 1H NMR (400 MHz, CDCl3)
δ 7.42 (2H, d, J ) 8.0 Hz), 7.28 (2H, d, J ) 8.0 Hz), 6.79
(1H, s), 4.36 (2H, q, J ) 7.2 Hz), 4.06 (s, 3H), 1.36 (3H, t, J
) 7.2 Hz), 1.29 (9H, s). 1-Methyl-4-(4-tert-butylphenyl)-2H-
1
pyrazole-3-(carboxylic acid ethyl ester) (24): H NMR (400
MHz, CDCl3) δ 7.32 (5H, m), 4.31 (2H, q, J ) 7.2 Hz), 3.94
(3H, s), 1.29 (12H, m). Diethyl 1,4-dimethyl-1,4-dihydro-
1
1,2,4,5-tetrazine-3,6-dicarboxylate (25): H NMR (300 MHz,
CDCl3) δ 4.29 (4H, q, J ) 7.2 Hz), 3.18 (6H, s), 1.30 (6H, t,
J ) 7.2 Hz).
[Methylhydrazono] Acetic Acid Ethyl Ester (17). Meth-
ylhydrazine (12.5 kg, 271 mol, 1.0 equiv) was dissolved in
MeOH (90.7 L) and heated to 40–43 °C. Ethyl glyoxylate (50.4
kg, 45–50% solution in toluene, 247 mol, 0.93 equiv) was added
while maintaining the temperature below 50 °C. MeOH (3.3
L) was added to the reaction mixture through the addition line.
The mixture was heated to 50 °C for 3 h. Volatiles (MeOH,
methylhydrazine and toluene) were distilled off, and TBME
(20.2 L)/heptane (20.2 L) was added to the reaction mixture.
The mixture was cooled to -5–0 °C during which time solids
began to precipitate out. After a 1-h hold at 0 °C, the slurry
was filtered, and the cake was washed with TBME (20.2 L)/
heptane (20.2 L). Yield ) 20.6 kg (64%). 1H NMR (400 MHz,
DMSO-d6) δ 8.81 (1H, s), 6.5 (1H, s), 4.11 (2H, q, J ) 7.2
Hz), 2.79 (3H, d, J ) 4.0 Hz), 1.21 (3H, t, J ) 7.2 Hz).
Bromo[methylhydrazono] Acetic Acid Ethyl Ester (18).
A slurry of NBS (28.2 kg, 158 mol, 1.0 equiv) in EtOAc (82.4
L) was cooled to 0–5 °C. In another vessel, hydrazone 17 (20.6
kg, 158 mol, 1.0 equiv) was dissolved in methylene chloride
(82.4 L) and added to the NBS/EtOAc slurry while maintaining
the temperature below 25 °C. The mixture was stirred at 5–10
°C for 1 h. Once the reaction was complete, the mixture was
filtered, and the unwanted cake was washed with EtOAc (41.2
L). The combined filtrate and wash were stored at 5–10 °C for
use in the next step.19 Quantitative yield was assumed. An
analytically pure sample can be obtained by removing the
solvent: 1H NMR (300 MHz, CDCl3) δ 6.51 (1H, s), 4.35 (2H,
q, J ) 7.2 Hz), 3.31 (3h, d, J ) 2.7 Hz), 1.36 (3H, t, J ) 7.2
Hz).
With residual TBME completely removed, the amido-
methylation proceeded smoothly to give exclusively the para
product 48, which was crystallized from toluene to produce a
total of 47.9 kg of acid-amide 48 (77–79% yield over two steps,
based on ꢀ-cresol). The deprotection of the amide was ac-
complished in refluxing EtOH with H2SO4, and extraction with
toluene provided free amine 10 as a toluene solution.
Coupling and Final Particle Formation. The coupling of
benzylamine 10 and pyrazole 11 was achieved by converting
pyrazole acid 11 to its acid chloride using SOCl2, followed by
exposure to benzylamine 10 in the presence of triethylamine
(Scheme 15). The reaction proceeded smoothly in toluene at 0
°C to room temperature. The ester functionality was saponified
using NaOH in aqueous ethanol to afford the desired product
1. The wet crude API, obtained after acidification and filtration,
was dissolved in n-propyl acetate and washed with 0.5 N HCl
at 65 °C. Subsequent distillation followed by a cooled, seeded
crystallization produced API grade GSK183390A. The yield
of the process was determined to be 86% on a 40-kg scale
(based on pyrazole acid 11).
Conclusion
A scalable synthesis of GSK183390A, a PPAR R/γ agonist,
was developed. This synthesis is highlighted by (1) a regiose-
lective, formal 1,3-dipolar cycloaddition reaction to form 1,3,5-
trisubstituted pyrazole 5 and (2) a regioselective amidometh-
ylation of ꢀ-cresol derivatives to allow for an efficient synthesis
of benzylamine 10. This route has been scaled up to produce
40 kg of API in the pilot plant.
4-[1-(4-tert-Butyl-phenyl)-vinyl]-morpholine (22). Mor-
pholine (52.0 kg, 597 mol, 6.0 equiv) was added to a slurry of
toluene (175.0 L) and sodium sulfate (87.5 kg) at 5–8 °C.
Titanium tetrachloride (18.8 kg, 99 mol, 1.0 equiv) was added
while maintaining the reaction temperature below 15 °C. A light
green slurry (titanium tetrachloride–morpholine complex) formed,
and Hünig’s base (64.2 kg, 497 mol, 5.1 equiv) was added
followed by tert-butylacetophenone (17.5 kg, 99 mol, 1.0 equiv).
The reaction mixture was heated to 70–73 °C for 3 h. Once
the reaction was complete, the mixture was cooled to 20–23
°C and centrifuged. The cake was washed with toluene (35.0
L), and the wash was added to the filtrate. The filtrate,
containing enamine 22, was subjected to vacuum distillation
until morpholine was <10% and Hünig’s base was <10% by
Experimental Section
General. Unless otherwise indicated, all reactions were
conducted in glass-lined reactors under a nitrogen atmosphere.
Solvents and reagents were obtained from commercial sources
and used without further purification.
General Procedure for 1,3-Dipolar Cycloaddition Screen-
ing Study in Table 1. In a reaction vial, hydrazonoyl bromide
18 or chloride 19 (500–700 mg, 1.0 equiv) and dipolarophile
(0.8 equiv) were dissolved in toluene (5 mL). Triethylamine
(3.0 equiv) was added, and the reaction mixture was heated
from 25 to 70 °C for 2–3 h. After the reaction was over, the
mixture was cooled to 5–10 °C, 4 N HCl (5 mL) was added,
and the mixture was stirred for approximately 1 h. The acidic
aqueous layer was separated to waste, and the organic layer
was washed with water (5 mL). After concentration via rotary
(19) Hydrazonyl bromide 18 is a skin sensitizer, and extra precautions were
taken in the plant during its handling.
1040
•
Vol. 11, No. 6, 2007 / Organic Process Research & Development

GC analysis. Ethyl acetate (73.5 L) and triethylamine (58.3 kg,
576 mol) were added in preparation for the next step. Quantita-
tive yield was assumed. An analytically pure sample can be
obtained by removing the solvent: 1H NMR (400 MHz,
pyridine-d5) δ 7.56 (2H, d, J ) 8.0 Hz), 7.44 (2H, d, J ) 8.0
Hz), 4.46 (1H, s), 4.25 (1H, s), 3.71 (2H, t, J ) 4.8 Hz), 2.79
(2H, t, J ) 4.8 Hz), 1.33 (9h, s).
resultant suspension over ∼8 h. The reaction mixture was stirred
at 50 °C for an additional 3 h. Water (145 L) was added, and
the reaction mixture was held at room temperature overnight.
MEK was removed by vacuum distillation. The resultant
solution was washed with TBME (145 L). The pH of the
aqueous phase was adjusted to 6.5 with 3 N HCl (59.9 kg)
before NaHSO3 (27.9 kg, 268 mol) was added to the reaction
mixture. The resultant mixture was stirred at room temperature
overnight. The mixture was acidified with 3 N HCl (124 kg) to
pH 1.6 and extracted with TBME (290 L × 2). The product
was isolated as a TBME solution (quantitative yield assumed).
An analytically pure sample can be obtained by removing
1-Methyl-5-(4-tert-butylphenyl)-2H-pyrazole-3-(carboxy-
lic acid ethyl ester) (5). The mixture of enamine 22 (24.4 kg,
99 mol, 1.0 equiv), triethylamine (58.3 kg, 576 mol, 5.8 equiv)
and EtOAc (73.5 L) from the previous step was heated to 40–43
°C, at which time an EtOAc (116.5 L)/CH2Cl2 (77.5 L) solution
of bromide 18 (31.2 kg, 149 mol, 1.5 equiv) was added over
30–45 min while maintaining the temperature between 40 and
50 °C. The reaction mixture was held at 40–43 °C for an
additional 3.5 h. After the reaction was complete, the mixture
was cooled to 0–5 °C, and 4 N HCl (213.2 kg, 8.0 equiv) was
added while maintaining the temperature below 30 °C. After
acidification was complete, the mixture was stirred for an
additional 1 h at 20–23 °C. The phases were allowed to separate,
and the aqueous layer was discarded. The organic layer was
sequentially washed with deionized water (104.9 L) and aqueous
lithium hydroxide solution (8.4 kg, 200 mol) in deionized water
(83.2 L) to ensure that the pH was basic. The organic layer
was distilled under reduced pressure to remove ethyl acetate
(approximately 373 L) in preparation for the hydrolysis step.20
1-Methyl-5-(4-tert-butylphenyl)-2H-pyrazole-3-(carboxy-
lic acid) (11). To the remaining solution of pyrazole ester in
EtOAc/toluene (∼63 L) was added THF (122.0 L) followed
by aqueous solution of lithium hydroxide (29.2 kg, 696 mol)
in deionized water (300 L). The mixture was heated to 60–63
°C for 4–6 h. After the reaction was complete, THF/toluene
(∼170 L) was distilled off under vacuum while maintaining
the temperature below 60 °C. TBME (87.8 L) was added, and
the reactor contents stirred and allowed to settle. The organic
layer was removed, and the aqueous layer (containing the
lithium salt of acid 11) was washed with TBME (87.8 L). The
aqueous layer was heated to 35 °C, and isopropanol (104.9 L)
was added. The pH was adjusted to 4 by addition of 4 N HCl
(156.3 kg) while maintaining the process temperature at 40–45
°C. Seeds of acid 11 (200 g, 0.77 mol) were added, and the
mixture stirred at pH ) 4 for approximately 1 h, during which
time a light precipitate was observed. The acidification was
continued with 4 N HCl (37.0 kg) until pH ) 0 was achieved.
The mixture was cooled to 0 °C over 1 h and filtered via
centrifuge. After filtration, the cake was washed with water (70.8
L) followed by TBME/heptane (7.1 L/63.7 L). Yield ) 14.4
kg (56% from enamine 22). 1H NMR (300 MHz, DMSO-d6)
δ 12.69 (1H, s), 7.52 (4H, m, A2B2), 6.80 (1H, s), 3.91 (3H, s),
1.32 (9H, s).
1
TBME: H NMR δ (CDCl3, 400 MHz) 7.20 (d, 1H, J ) 7.3
Hz), 7.13 (t, 1H, J ) 8.3 Hz), 6.97 (t, 1H, J ) 7.5 Hz), 6.84
(d, 1H, J ) 8.2 Hz), 2.29 (s, 3H), 1.67 (s, 6H); 13C NMR δ
(CDCl3, 100 MHz) 16.70, 25.17, 78.93, 117.43, 122.26, 126.29,
129.82, 131.09, 153.16, 180.32.
2-[(4-{[(Chloroacetyl)amino]methyl}-2-methylphenyl)oxy]-
2-methylpropanoic Acid (48). The TBME solution of 39 from
the previous stage (19.7 kg assuming 100% yield, 101 mol)
was distilled to minimum stir volume and HOAc (49.4 L) was
charged. Vacuum distillation was continued until the content
of residual TBME was 0.5%. Additional HOAc (18.2 kg) and
a premixed solution of HOAc and H2SO4 (49.4 L/9.9 L) were
added to achieve a 10:1 ratio of HOAc and H2SO4. 2-Chloro-
N-(hydroxymethyl)acetamide (13.8 kg, 112 mol, 1.1 equiv) was
added. The reaction mixture was stirred at room temperature
overnight. The reaction mixture was quenched with aqueous
NaOAc (18.4 kg, 224 mol, in 197.5 L of water) and extracted
with TBME (99 L × 2). The combined organic layer was
washed with water (99 L) and then distilled to minimum stir
volume. Toluene (198 L) was added, and the reactor contents
were distilled to minimum stir volume. The toluene charge-
distillation process was repeated a second time to ensure
complete azeotropic removal of residual water and HOAc.
Toluene (198 L) was added, and GC analysis showed HOAc
content in the reaction mixture to be 1.7%.21 Additional toluene
(40 L) was added, the mixture was heated to 90 °C, and
complete dissolution was achieved. The reactor content was
cooled to 50 °C over 30 min. Crystallization of the product
was observed. The mixture was held at 50 °C for 1 h, cooled
to room temperature over 20 min, and held at room temperature
for 1 h. The product was collected by centrifuge filtration, rinsed
with cyclohexane (40 L × 2), and tray-dried under vacuum at
45 °C: 24.1 kg, 79% from ꢀ-cresol. LC purity: 99.8%. 1H NMR
δ (CDCl3, 400 MHz) 7.11 (d, 1H, J ) 2 Hz), 7.00 (dd, 1H, J
) 8.0, 2.0 Hz), 6.87 (br s, 1H), 6.78 (d, 1H, J ) 8.0 Hz), 4.40
(d, 2H, J ) 5.6 Hz), 4.13 (s, 2H), 2.26 (s, 3H), 1.65 (s, 6H);
13C NMR δ (CDCl3, 100 MHz) 16.72, 25.22, 42.48, 43.40,
78.95, 116.91, 125.86, 130.06, 130.23, 130.75, 153.02, 166.39,
178.40.
2-Methyl-2-[(2-methylphenyl)oxy]propanoic acid (39).
ꢀ-Cresol (14.5 kg, 134 mol) was dissolved in methyl ethyl
ketone (290 L), and NaOH pellets (26.8 kg, 670 mol, 5 equiv)
were added. The mixture was heated to 50 °C and stirred at 50
°C for 2 h. A solution of R-bromoisobutyric acid (40.3 kg, 241
mol, 1.8 equiv) in methyl ethyl ketone (87 L) was added to the
Ethyl 2-{[4-(Aminomethyl)-2-methylphenyl]oxy}-2-me-
thylpropanoate (10). Amide 48 (23.6 kg, 78.7 mol) was
dissolved in EtOH/H2SO4 (94.4 L/11.8 L), and the resultant
solution was heated to reflux for 19 h. The reaction mixture
was distilled to minimum stir volume, toluene (118 L) was
(20) It is important to remove EtOAc completely before charging LiOH.
Residual EtOAc is hydrolyzed by LiOH to acetate which buffers the
reaction at a lower pH, leading to slow or incomplete hydrolysis.
(21) The crystallization of the product is inhibited by large amounts of
residual HOAc.
Vol. 11, No. 6, 2007 / Organic Process Research & Development
•
1041

added, and the mixture was concentrated to minimum stir
volume. The azeotropic distillation was repeated a second time
with a fresh portion of toluene (118 L), and the product was
extracted into water (118 L). The aqueous layer was diluted
with a fresh portion of toluene (118 L) and basified with NaOH
(18.9 kg, in 112 L of water) at 0 °C. The free amine was
extracted into toluene and used as a solution in the next stage
(quantitative yield assumed). An analytically pure sample of
the amine could be isolated as the oxalate: 1H NMR δ (DMSO-
d6, 400 MHz) 7.28 (d, 1H, J ) 2.2 Hz), 7.17 (dd, 1H, J ) 8.4,
2.2 Hz), 6.61 (d, 1H, J ) 8.4 Hz), 4.18 (q, 2H, J ) 7.1 Hz),
3.91 (s, 2H), 2.17 (s, 3H), 1.54 (s, 6H), 1.18 (t, 3H, J ) 7.1
Hz); 13C NMR δ (DMSO-d6, 100 MHz) 14.36, 16.82, 25.51,
42.21, 61.56, 79.11, 116.39, 127.45, 127.64, 129.04, 132.14,
153.87, 164.84, 173.78.
2-[(4-{[({5-[4-(1,1-Dimethylethyl)phenyl]-1-methyl-1H-
pyrazol-3-yl}carbonyl)amino]methyl}-2-methylphenyl)oxy]-
2-methylpropanoic Acid (1, GSK183390). A toluene solution
of SOCl2 (9.3 kg, 78 mol, 1.3 equiv, in 6.8 L of toluene) was
added to acid 5 (15.5 kg, 60 mol) in toluene (155 L) at 90 °C
over 27 min. The resultant mixture was heated at 90 °C for
1 h, and the conversion to the acid chloride was complete. The
reaction mixture was distilled to minimum stir volume to
remove excess SOCl2, and the residue was diluted with toluene
(77.5 L). The acid chloride solution was added to a mixture of
benzylamine 10 (121.9 kg of toluene solution, containing 19.7
kg of benzylamine 10, 78 mol, 1.3 equiv) and Et3N (23.6 kg,
234 mol) at 0 °C over 30 min. The reaction mixture was then
stirred at 20 °C for 30 min. The reaction mixture was washed
with HCl (2 N, 160 kg) and NaOH (1 N, 159 kg). Water (109
L) was added to the resultant toluene solution, and toluene was
removed by azeotrope. EtOH (201 L) and NaOH (1 N, 143
kg) were added to the resultant suspension, and the mixture
was heated at 60 °C for 2 h to effect hydrolysis of the ester.
EtOH was removed from the mixture by vacuum distillation.
The aqueous solution was cooled to room temperature and
acidified with 1 N HCl (158 kg). The product acid GSK183990A
precipitated and was collected by centrifuge filtration.
GSK183390A (1). Crude GSK183390 (137.4 kg wet crude
cake containing 47.0 kg of GSK183390, 101 mol) was slurried
in water-saturated n-propyl acetate (376 L) and heated to 45–50
°C followed by the addition of 0.5 N HCl solution (141 L) to
guarantee dissolution of crude 1. The reactor contents were then
stirred and settled, and the aqueous acidic layer was removed
to waste. Deionized water (141 L) was added, the mixture was
stirred and settled, and the aqueous layer was removed to waste.
The organic phase was heated to 65–70 °C, and a clarification
filtration was performed. The organic phase was distilled down
to 5–6 vol under reduced pressure while maintaining temper-
ature below 60 °C and ensuring the KF specification was met
(<5000 ppm for maximum recovery). The temperature was
heated up to 70 °C to ensure complete dissolution. The contents
of the reactor were cooled to between 58 and 62 °C, and seeds
of 1 (470 g, 1.0 mol, 0.001 equiv) were added as a slurry in
n-propyl acetate (800 mL). The contents of the reactor were
held at 60 °C for approximately 30 min, then cooled down to
0 °C over 2 h and held at 0 °C for 1 h before filtration through
a centrifuge. The cake was washed with heptane (188.0 L) and
dried in the oven at 60–65 °C under vacuum. Yield ) 43.6 kg
(86% based on acid 11). NMR (300 MHz, DMSO-d6) δ 8.59
(1H, t, J ) 6.3 Hz), 7.51 (4H, m, A2B2), 7.10 (s, 1H), 7.01
(1H, d, J ) 9.0 Hz), 6.74 (s, 1H), 6.65 (1H, d, J ) 9.0 Hz),
4.31 (2H, d, J ) 6.0 Hz), 3.90 (3H, s), 2.13 (3H, s), 1.48 (6H,
s), 1.32 (9H, s).
Acknowledgment
We gratefully acknowledge L. Wernersbach for her initial
effort in the amidomethylation chemistry, Dr. D. Bhanushali
for his effort in the pilot campaign and Dr. A. Freyer for his
support in NMR spectroscopy.
Received for review July 18, 2007.
OP700164T
1042
•
Vol. 11, No. 6, 2007 / Organic Process Research & Development