A large-scale synthesis of potent glucokinase activator MK-0941 via selective o -arylation and o -alkylation

Article abstract of DOI:10.1021/op200068c

An efficient, practical preparation of MK-0941, a potent glucokinase activator, is described. Keys to the success of the synthesis are a highly selective mono-O-arylation of methyl 3,5-dihydroxybenzoate with 2-ethanesulfonyl-5-chloropyridine and the choic

Full text of DOI:10.1021/op200068c

ARTICLE
pubs.acs.org/OPRD
A Large-Scale Synthesis of Potent Glucokinase Activator MK-0941 via
Selective O-Arylation and O-Alkylation
Naoki Yoshikawa,* Feng Xu,* Juan D. Arredondo, and Takahiro Itoh
Department of Process Research, Merck Research Laboratories, Rahway, New Jersey 07065, United States
S Supporting Information
b
ABSTRACT: An efficient, practical preparation of MK-0941, a potent glucokinase activator, is described. Keys to the success of the
synthesis are a highly selective mono-O-arylation of methyl 3,5-dihydroxybenzoate with 2-ethanesulfonyl-5-chloropyridine and the
choice of a proper protective group for the subsequent SN2 O-alkylation. With the thorough understanding of the origins and fate of
in-process impurities, the second-generation robust synthesis with a minimum number of operations reproducibly prepares MK-
0941 in 56% overall yield with >99% purity.
’ INTRODUCTION
’ RESULTS AND DISCUSSION
Glucokinase (GK) is a member of the hexokinase enzymes
and catalyzes the phosphorylation of glucose to glucose-6-
phosphate.1 This enzyme is predominantly expressed in liver
and pancreatic β-cells.1 Acting as a glucose sensor of the insulin-
producing pancreatic islet β-cells, glucokinase plays a key role in
glucose homeostasis by controlling the conversion of glucose to
glycogen and regulating hepatic glucose production.1,2 To date, a
number of GK activators have advanced into clinical trials for the
treatment of type II diabetes.2
Selective Mono-O-arylation. Without involving extra manip-
ulation steps, such as sequential protection/deprotection, selec-
tive and direct mono-O-arylation of resorcinol derivatives,5ꢀ8
such as 3, is challenging due to the competing formation of bis-
arylated byproducts. Indeed, treatment of chloropyridine 23 with
1.1 equiv of 3 and 2.2 equiv of t-BuOK in 1,3-dimethyl-2-
imidazolidinone (DMI) at 100 °C resulted in a modest mono/
bis selectivity (4:9 = 3ꢀ4:1) along with the formation of
numerous byproducts including impurity 10, while the yield of
the desired product 4 varied from 37% to 67%.
In order to probe the mono/bis selectivity, we first studied the
formation of the bis-O-arylated byproduct 9 by treating 4 with
chloropyridine 2 in the presence of t-BuOK (Scheme 2). The O-
arylation of 4 was much slower than that of 3 under the same
conditions9 and generated numerous byproducts. The anion of
dihydroxybenzoate 3 appeared to be more reactive than that of
the mono O-arylated 4,9 which led us to believe that the reaction
selectivity could be competitively improved by increasing the
equivalents/concentration of benzoate 3.
Thus, studies on selective mono-O-arylation of 3 with 2 were
initiated in attempts to utilize the O-arylation reactivity differ-
ences between 3 and 4. Interestingly, as the ratio of 3 vs 2 varied,
the selectivity of the mono O-arylated 4 vs the bis O-arylayted
byproduct 9 was significantly changed (Table 1). In addition, the
use of excess of 3 also helped to suppress the formation of
impurities and therefore enhanced the yield. We were pleased to
find that the selectivity could be dramatically improved from 3:1
(4:9) to 28:1 in the presence of 2 equiv of t-BuOK, if the
equivalents of 3 were increased from 1.1 equiv to 1.5 equiv. The
use of 1.5 equiv of 3 was therefore determined to be optimal for
the best balance of yield, selectivity, and overall cost.10
MK-0941 (1)3 is a potent GK activator possessing a differentially
substituted 3,5-dihydroxybenzamide structure. In order to support
the drug development, an efficient synthesis suitable for large-scale
preparation was required. The chemistry utilized at early stages of
drug development is outlined in Scheme 1.4 Although the overall
synthetic strategy was straightforward, this synthesis had several
major drawbacks. The O-arylation of dihydroxybenzoate 3 with
chloropyridine 23 had only moderate mono/bis selectivity and also
suffered from poor reproducibility when scaled up. Attempts to
isolate 4 as a crystalline solid were not successful due to the
numerous in-process impurities generated in this step, which thus
were carried through to the subsequent steps. As such, multiple
recrystallizations of 7 were required to reject numerous impurities at
the expense of a significant downgrade of its enantiomeric purity.
Therefore, a subsequent recrystallization of 1 became necessary in
order to improve its enantiomeric purity. In addition, the overall
yield was highly variable, averaging around only ∼20% upon scale
up, as compared to 36% on a lab scale. Most importantly, significant
variability observed from batch to batch resulted in the lack of
control of the impurity profile of the final product 1, which raised a
serious concern about the robustness of the process.
In line with the current synthetic strategy, we envisioned that
robust control of the impurity profile of the final API could be still
attained, if the origin and fate of the process impurities could be
understood fundamentally. This article describes our efforts to
overcome all the issues encountered in the original synthesis,
leading to the successful development of a highly efficient and
practical-second generation synthesis of MK-0941 (1) in three
pots in >99% purity with full control of the impurity profile.
Among the various bases screened, the use of t-BuONa or t-
BuOK was most effective in terms of the reaction yield. Solvent
screening confirmed that the use of polar aprotic solvents was
Received: March 17, 2011
Published: May 03, 2011
r
2011 American Chemical Society
824
dx.doi.org/10.1021/op200068c Org. Process Res. Dev. 2011, 15, 824–830
|

Organic Process Research & Development
ARTICLE
Scheme 1. Initial synthesis of MK-0941 (1)
Scheme 2. O-Arylation of 3
Figure 1. Effects of t-BuOK on O-arylation profile.
steps. Therefore, further development of this step focused on the
enhancement of the mono/bis selectivity.
After several experiments, we found that the impurity profile of
the reaction was sensitive to the amount of t-BuOK charged.
Figure 1 depicts the correlation between the levels of byproducts
9 and 10 (Scheme 2) and the amount of t-BuOK. A significant
decrease of the formation of 9 was achieved when the amount of
t-BuOK was increased to >2.3 equiv. In contrast, the formation of
10, which could be rejected easily through crystallization, was
only subtly influenced by the increased charge of t-BuOK.
It is worthwhile to point out that the monoanion of 3 was very
soluble in AcNMe2, while the corresponding dianion potassium salt
formed a slurry even at 80 °C.11 Furthermore, if t-BuOK was
charged as a solid, the unreacted/undigested t-BuOK would lead to
variable results in terms of the reaction yield as well as the impurity
profile. To circumvent these issues and achieve reproducible O-
arylation, t-BuOK was dissolved in AcNMe2 before it was charged
into a solution of 2 at ambient temperature. The resulting slurry of
the potassium salt of 3 was further agitated at 50 °C followed by
95 °C for several hours to ensure a complete digestion of t-BuOK
(Scheme 3) and only then was chloropyridine 2 added. Applying
this protocol instead of charging solid t-BuOK into the reaction
mixture significantly improved the reproducibility and the reaction
profile. Thus, the use of 2.5 ( 0.2 equiv of t-BuOK and 1.5 equiv of
benzoate 3 in AcNMe2 was determined to be optimal12,13 to
minimize the formation of 9 to <0.2%, while the assay yield of 4
Table 1. Selected data for the effects of the amounts of 3 on
O-arylation selectivitya
entry
3 (equiv)
yield of 4 (%)b
selectivity (4:9)
1
2
3
1.1
1.3
1.5
55
70
83
3:1
6:1
28:1
a Reaction conditions: AcNMe2, 95 °C. Solid t-BuOK (2.0 equiv) was
charged in portions. b Assay yield by HPLC.
crucial to obtain high yields. Thus, AcNMe2 became our choice
of solvent for further optimization. The reaction temperature
also had a significant effect on the reactivity of O-arylation.
Incomplete conversion was observed at <80 °C.
Although the mono/bis selectivity was substantially improved to
a decent level (28:1) by increasing the charge of 3, it was found that
byproduct 9 was poorly rejected in the downstream chemistry. At
this point, it was determined that the mono/bis selectivity (4:9) had
to be further improved to >200:1 (i.e., <0.5% 9), in order to
practically prepare MK-0941 without introducing extra purification
825
dx.doi.org/10.1021/op200068c |Org. Process Res. Dev. 2011, 15, 824–830

Organic Process Research & Development
ARTICLE
was maintained at 85% (Figure 1). The mono/bis selectivity was
finally improved up to >500:1, which was significantly above the
targeted selectivity (>200:1).
Upon the completion of the reaction, the majority of the
unconsumed benzoate 3 was effectively removed through
i-PrOAc-aqueous workup.10 With the improved selectivity of O-
arylation, the desired 4 could be directly isolated as a crystalline
solid from i-PrOAc/heptane/2-propanol14 in 75% yield with
>94% HPLC purity (Scheme 3). In contrast to the original
process, the isolation of 4 at this stage rejected numerous
impurities and improved the robustness of the synthesis.
SN2 O-Alkylation of 4. With the preparation of 4established, we
turned our attention to the subsequent SN2 O-alkylation with
mesylate 5, which had also suffered from poor reproducibility along
with the formation of several troublesome impurities. To under-
stand the reaction pathways for the formation of these impurities,
which are rejected poorly in the downstream chemistry, and to gain
a control of their formation, their structures (Scheme 4) were
unambiguously elucidated by NMR and LC/MS.
Scheme 3. Optimized process for selective mono-O-
arylation
It was believed that the formation of these impurities was
caused by the labile nature of the TBS-protection as depicted in
Scheme 4. The cleavage of the TBS-ether in product 6 by a small
amount of H2O,15 which could be generated under the alkylation
conditions (Cs2CO3, 80 °C), could result in the formation of
ester dimer 12 as well as des-TBS 11. Similarly, (R)-OH-isomer
15a could arise from desilylation of the corresponding 15b,
which was formed via O-alkylation of 4 with regioisomer
mesylate 18, a byproduct of selective TBS-protection of 1,2-
propanediol during the preparation of 5.3 Indeed, treatment of 4
with 5 resulted in the formation of significant levels of des-TBS
11, ester-dimer 12, (S)-OH-isomer 14 and (R)-OH-isomer 15
(Table 2, entry 1). However, the observation of byproduct 14
intrigued us. In particular, the formation of (S)-OH-isomer 14
was also clearly observed upon exposure of the purified 6 to the
same SN2 O-alkylation conditions (Table 2, entry 2). Byproduct
Scheme 4. Plausible pathways for the formation of byproducts of SN2 O-alkylation
826
dx.doi.org/10.1021/op200068c |Org. Process Res. Dev. 2011, 15, 824–830

Organic Process Research & Development
ARTICLE
Table 2. Selected data for the effects of protective group on the O-alkylation of 4
entry
substrates
mesylates
time (h)
(S)-OH-isomer 14(%)a,b
(R)-OH-isomer 15 (%)a,b
desilylation 11 (%)a
ester-dimer 12 (%)a
1
2
3
4
5
4
5
7
12c
12c
7
0.24
0.73
6.30
0.04
0.13
0.97
n/d
11.0
ꢀ
5.3
6
ꢀ
ꢀ
19
ꢀ
ꢀ
ꢀ
11
4
n/d
ꢀ
n/d
0.31
ꢀ
unknown
ꢀ
20
12c
n/d
a HPLC area%. b Determined by HPLC after saponification and desilylation to free acid of 7. The percentages are reported as relative to the desired
product. c Purified products (6, 11, or 20) were exposed to the alkylation conditions for 12 h (Cs2CO3, AcNMe2, 80 °C) followed by aging for additional
12 h after addition of a small amount of H2O to mimic the presence of water in the actual reaction system.15
by GC analysis)16 was also completely removed. The crude stream
of 22 with 0.08% of 23 was subjected to the subsequent mesylation
(1.2 equiv MsCl, 1.4 equiv Et3N, 0 °C, 1 h) to afford the desired
TIPS-mesylate 19 in 94% assay yield (87% yield over two steps).
Therefore, the crude 19 in toluene with only <0.1% of 25 could be
directly used for the subsequent O-alkylation reaction without
further purification.
Scheme 5. Through-process to prepare TIPS-mesylate 19a
With the robust process for the preparation of 4 and TIPS-
mesylate 19 in hand, a streamlined through-process to prepare
DABCO salt 7 was then developed (Scheme 6). The use of TIPS
protected mesylate 19 dramatically suppressed the formation of
impurities such as desilylated byproducts, dimers, and OH-
isomers, as discussed in Scheme 4. With the benefit of the stable
TIPS protecting group, it became possible to lower the charge of
the mesylate 19 and Cs2CO3 significantly without sacrificing the
yield and purity profile. Since the SN2 O-alkylation was carried
out under heterogeneous conditions with dense Cs2CO3 salts,
the reaction tended to be slower at high moisture levels or with
slower agitation. A good mixing was essential to achieve >99%
conversion. Under the optimized conditions (1.3 equiv 19, 1.5
equiv Cs2CO3, 80 °C, 12 h), the alkylation reaction furnished the
desired product 20 in 93% assay yield.17
It is worthwhile to note that the unconsumed 4, ∼1% at the
end of the alkylation, would be converted in the subsequent steps
to the corresponding impurities, which were difficult to reject.18
After several experiments, we found that the unconsumed 4
could be selectively partitioned/rejected in the aqueous AcNMe2
layer during the workup, if the ratio of AcNMe2:H2O was
adjusted to approximately 3:2 (v/v). The product 20 could be
selectively extracted with MTBE.
In the initial synthesis, the deprotection of the silyl ether was
performed before the saponification of the methyl ester
(Scheme 1). However, the corresponding desilylated alcohol
11 from 20 could react19 with the unconsumed excess mesylate
19, which was carried through from the previous O-alkylation
step, to form byproducts in the subsequent saponification step.
To circumvent this issue, the order of this sequence was reversed.
In practice, the crude 20 was hydrolyzed first with aqueous
NaOH (5 N, 1.6 equiv) in THF/MeOH, while the residual
mesylate 19 was also quenched under this condition. In one pot
the resulting crude 29 was subsequently treated with aqueous
HCl at 35 °C for 8 h to afford the free acid 7. As such, the
hydrolysisꢀdeprotection proceeded in near quantitative yield.
With implementation of the process described above, free acid
7 was converted to the corresponding DABCO salt and effec-
tively isolated in 83% yield over three steps with >99% purity
(Scheme 6). Multiple recrystallizations of 7 were no longer
needed to meet the purity specifications. More importantly,
the new process performed with greatly improved robustness
a The ratio of 22:23:24 was determined by gas chromatography (GC)
analysis (RTX-200 column) and did not represent the actual mole ratio
without calibration.
14, in principle, could be generated from epoxide 17 via cleavage
of the TBS group of mesylate 5 (path a); however, the results of
the stress test (Table 2, entry 2) suggested that 14 was most likely
formed via intermediate 13 through intramolecular nucleophilic
attack to the aromatic ring (path b). This plausible ipso pathway
was further supported by the fact that treatment of alcohol 11
under the same conditions (Cs2CO3, AcNMe2, 80 °C, 12 h)
resulted in an elevated level (6.3%) of 14 (Table 2, entry 3).
With these results in hand, we envisioned that the formation of
these byproducts would be suppressed if the labile TBS group of
5 could be replaced with a more stable protective group such as
triisopropylsilyl group (TIPS). Indeed, the use of TIPS-pro-
tected mesylate 19 dramatically suppressed the formation of all
of these impurities as expected (Table 2, entry 4). The formation
of OH-isomers 14 and 15 was controlled to <0.05%. Further-
more, exposure of the TIPS-protected product 20 to the same
SN2 O-alkylation conditions resulted in the formation of only
negligible amount of 14 (Table 2, entry 5), which further
confirmed that the increased stability of TIPS group shut down
the degradation via pathway b (Scheme 4).
With the desired choice of protective group determined, we
turned our attention to develop a process to prepare mesylate 19.
Most of the impurities including 14 formed during the SN2 O-
alkylation reaction were successfully suppressed by employing a
more stable TIPS protective group. However, it was still crucial to
minimize the formation of the regioisomer 23 and to convert 23 to
the corresponding inactive bis-TIPS ether 24 (Scheme 5), because
23 could lead to the formation of (R)-OH-impurity 15 via the
corresponding mesylate 25 (Scheme 4). Under the optimized
conditions, treatment of 21 with 1.05 equiv of TIPSCl and 1.3
equiv of imidazole in MeCN at 0 °C followed by agingat 22°C for 2
h afforded 22 in 93% yield along with 0.09% of 23 and 11.8% of 24
by GC analysis (see, footnote of Scheme 5). Upon aqueous workup,
the unconsumed diol 21 (typically about 0.2% at the end of reaction,
827
dx.doi.org/10.1021/op200068c |Org. Process Res. Dev. 2011, 15, 824–830

Organic Process Research & Development
ARTICLE
Scheme 6. One-pot through-process to prepare DABCO salt 7 and endgame
and reliability to produce consistent results even on an
industrial scale.
’ EXPERIMENTAL SECTION
General. Degassing was conducted where specified by repeating
an evacuation/nitrogen refill cycle. HPLC assays were performed
using a reverse phase column eluted with 0.1% H3PO4 (aq)
and acetonitrile. The levels of OH-isomers 14 and 15 were
determined after saponification and desilylation to free acid of 7
by chiral HPLC analysis: ChiralPak IC; mobile phase: isocratic 0.1%
H3PO4 (aq)ꢀacetonitrile; column temperature: 40 °C; flow:
1.0 mL/min; 23% acetonitrile for 15 min, increased to 30%
acetonitrile in the next 15 min, and then hold at 30% acetonitrile
for the next 5 min. Retention times (Rt): minor enantiomer,
19.8 min; free acid 7, 21.3 min; (R)-OH-isomer (free acid of
15a), 24.1 min; (S)-OH-isomer (free acid of 14), 27.3 min. The
ratio of 22:23:24 was determined by gas chromatography (GC)
analysis: Column: RTX-200 30 m ꢁ 0.32 mm ꢁ 1.0 μm; constant
flow: 3.3 mL/min; carrier gas: He; inlet temperature: 240 °C;
detection: FID; oven program: 75 °C (3 min), ramp 20 °C/min to
190 °C, ramp 35 °C/min to 285 °C, and hold for 2 min. Retention
times (Rt): diol 21, 2.7 min; 22, 8.6 min; 23, 8.8 min; 24, 10.8 min.
3-(6-Ethanesulfonyl-pyridin-3-yloxy)-5-hydroxy-benzoic
Acid Methyl Ester (4). To a degassed solution of 3 (50.6 kg, 291.9
mol) in AcNMe2 (280 L) at 20ꢀ25 °C under N2 atmosphere was
added a degassed solution of t-BuOK (57.4 kg) in AcNMe2 (400 L)
dropwise over 2 h at <25 °C. The slurry was agitated at 20ꢀ25 °C
for 1 h, 50 °C for 1 h, and 95 °C for 0.5ꢀ1 h. A degassed solution of
chloropyridine 23 (40 kg, 194.5 mol) in AcNMe2 (120 L) was
added at 95ꢀ100 °C over 5 h. After additional 2 h age at 95 °C, the
batch was cooled to 20 °C and quenched into 1 M HCl (640 L) at
<30 °C. The pH of the quenched solution was adjusted to 2ꢀ3 with
1 M HCl and extracted twice with i-PrOAc (800 L þ 600 L). The
combined organic phase was washed with 5% NaCl (3 ꢁ 400 L).
The organic phase was azeotropically concentrated (jacket tem-
perature <50 °C) to ∼240 L followed by addition of 2-propanol
(24 L). The batch was seeded, and the slurry was aged at 25ꢀ30 °C
for 6 h. Heptane (420 L) was added over 10 h, and the batch was
cooled to 20 °C and agitated for 2ꢀ4 h before filtration. The wet
cake was washed with 30% i-PrOAc in heptane (2 ꢁ 200 L)
followed by 25% i-PrOAc in heptane (200 L). Vacuum oven dry at
45 °C afforded 4 (53 kg) as a white solid. 75% yield. Analytically
Endgame. In the initial synthesis, the free acid was liberated
from the corresponding DABCO salt through an acidic aqueous
extraction, prior to the final EDC coupling with 8. To streamline
the process, elimination of this salt break step without sacrificing
the efficiency and yield of the EDC coupling was desired. After
several experiments, we found that the addition mode was
important to achieve this goal. A slow addition of 1 equiv of
aqueous HCl to a solution of the DABCO salt 7, amine 8, and
EDC in aqueous MeCN in the presence of a catalytic amount of
pyridine gave the desired MK-0941 effectively and cleanly.
Without introducing HCl, the reaction stalled at ∼60% conver-
sion. Finally, MK-0941 was isolated as its methanesulfonate salt
in 90% yield and >99% purity.
The process was successfully scaled up on industrial scale and
produced >1 tonne API without sacrificing yield and purity.
’ CONCLUSION
In summary, we have developed an efficient and robust
process for the large-scale synthesis of MK-0941. The success
of the development is attributed to two key accomplishments, a
highly selective mono-O-arylation and the selection of TIPS-
protected mesylate 19 for SN2 O-alkylation. The selectivity of the
mono/bis O-arylation was improved from 3ꢀ4:1 to >500:1. The
direct isolation of 4 greatly improved the robustness of the
synthesis by facilitating the rejection of numerous impurities.
The use of TIPS-protected mesylate 19 in the O-alkylation step
was crucial to suppress the formation of OH-isomers, which were
difficult to reject in the downstream chemistry. By applying an
improved through process, DABCO salt 7 was efficiently
obtained with >99% purity in a reproducible manner. In contrast
to ∼20% overall yield in the first generation synthesis, the highly
efficient second generation synthesis produces MK-0941 in 56%
overall yield and >99% purity with minimal operations/isola-
tions. More importantly, based on fundamental understanding
the origins and fate of the impurities, a robust control of the
impurity profile of the final product was finally achieved without
any recrystallizations or chromatographic purification.
828
dx.doi.org/10.1021/op200068c |Org. Process Res. Dev. 2011, 15, 824–830

Organic Process Research & Development
ARTICLE
pure sample was prepared by recrystallization. 1H NMR (500 MHz,
CDCl3) δ 8.47 (d, J = 2.8 Hz, 1 H), 8.05 (d, J = 8.2 Hz, 1 H), 7.48
(dd, J=2.3, 1.4Hz, 1H), 7.43(dd, J=2.3, 1.4Hz, 1H), 7.28(dd,J=
2.3, 1.4 Hz, 1 H), 6.84 (t, J = 2.3 Hz, 1 H), 3.82 (s, 3 H), 3.42 (q, J =
7.4 Hz, 2 H), 1.33 (t, J = 7.4 Hz, 3 H). 13C NMR (100 MHz,
CDCl3) δ 166.14, 158.09, 156.82, 155.36, 149.79, 141.01, 133.07,
124.99, 124.26, 114.10, 112.59, 112.07, 52.63, 46.93, 6.87. Anal.
Calcd for C15H15NO6S: C, 53.40; H, 4.48; N, 4.15. Found: C,
53.42; H, 4.23; N, 4.01.
solution of the free acid of 29 was stirred at 35 °C for 6ꢀ8 h until
>99.5% of free acid 29 was converted to the corresponding free
acid 7. The batch was cooled to ambient temperature and
i-PrOAc (75 L) and 15% NaCl (40 L) were added. The aqueous
layer was separated and extracted with i-PrOAc (40 L). The
combined organic layer was washed with 15% NaCl (75 L).
The above solution was azeotropically dried (batch temperature
<25 °C) with i-PrOAc. The solution was filtered to remove a small
amount of inorganic salts and diluted with i-PrOAc to 97 L and
MeOH (16 L). The batch was heated to 50 °C. 6.4 L of a DABCO
solution, prepared by dissolving DABCO (2.03 kg, 18.1 mol) in
i-PrOAc (43 L), was charged. A well-dispersed slurry of 7 (200 g)
in i-PrOAc (2 L) was charged as seed, and the resulting slurry was
stirred at 50 °C for 2 h to form a seed bed. The remaining DABCO
solution was added at 50 °C over 6 h. The resulting slurry was aged
at 50 °C for 1 h and then cooled to 22 °C over 1 h. After aging at
22 °C for 5 h, thesolid was collectedbyfiltration. The wet cake was
washed with 5% MeOH/i-PrOAc (20 L) followed by i-PrOAc(70L).
Drying in vacuum under nitrogen at 40 °C afforded 7 (11.9 kg)
as an off-white solid. 98.5% purity. >98.8% ee. 89% yield from 4.
1H NMR (400 MHz, DMSO-d6) δ 8.60 (d, J = 2.8 Hz, 1 H), 8.04
(d, J = 8.7 Hz, 1 H), 7.61(dd, J = 8.7, 2.8 Hz, 1 H), 7.36(dd, J = 2.4,
1.2 Hz, 1 H), 7.16 (dd, J = 2.4, 1.2 Hz, 1 H), 6.98 (t, J = 2.4 Hz, 1
H), 4.53ꢀ4.45 (m, 1 H), 3.54 (dd, J = 11.5, 5.8 Hz, 1 H), 3.48 (dd,
J = 11.5, 4.7 Hz, 1 H), 3.41 (d, J = 7.4 Hz, 2 H), 2.97 (s, 6 H), 1.21
(d, J = 6.2 Hz, 3 H), 1.14 (t, J = 7.4 Hz, 3 H). 13C NMR (100 MHz,
DMSO-d6) δ 167.52, 159.38, 156.53, 155.16, 149.86, 141.10,
137.83, 125.50, 124.25, 113.11, 112.08, 110.53, 75.16, 64.13,
46.22, 44.21, 16.39, 6.81. Anal. Calcd for C20H25N2O7S: C,
54.91; H, 5.76; N, 6.40. Found: C, 54.86; H, 5.51; N, 6.36.
3-(6-Ethanesulfonyl-pyridin-3-yloxy)-5-((S)-2-hydroxy-1-
methyl-ethoxy)-N-(1-methyl-1H-pyrazol-3-yl)-benzamide
Methanesulfonic Acid Salt (MK-0941, 1). To a solution of 7 (20
kg, 45.7 mol) in MeCN (120 L) and H2O (80 L) at 0ꢀ5 °C, were
added pyridine (1.1 kg, 13.7 mol) and 3-amino-1-methylpyrazole
(8, 5.3 kg, 54.86 mol). EDC-HCl (10.5 kg, 54.86 mol) was charged,
and the resulting solution was aged at 0ꢀ5°C for 30 min. HCl (1 M,
45.7 L, 45.72 mol) was charged at 0ꢀ5 °C over 3 h. The resulting
biphasic solution was stirred at 0ꢀ5 °C for 2ꢀ4 h. The reaction was
allowed to warm to 22 °C and diluted with i-PrOAc (160 L), H2O
(140 L) and 1 M HCl (13.7 L). The aqueous layer was separated
and extracted with i-PrOAc (160 L). The combined organic layer
was washed with 2% citric acid/20% NaCl (prepared from 2.4 kg of
citric acid, 24 kg of NaCl, and 93.6 kg of H2O) followed by 25%
NaCl (100 L). The solution was concentrated and azeotropically
dried with MeCN (batch temperature <25 °C). The resulting
mixture was filtered to remove inorganic salts and diluted with
MeCN (∼24 wt % of 1). Toluene (80 L) was added, and the batch
was heated to 30 °C, followed by charging methanesulfonic acid
(1.1 kg, 11.43 mol). The resulting solution was seeded (1.3 kg of
MK-0941 MsOH salt), and the resulting mixture was stirred at
25ꢀ35 °C for 2 h to form a seed bed. A solution of methanesulfonic
acid (3.7 kg, 38.86 mol) in MeCN (40 L) and toluene (40 L) was
charged at 25ꢀ35 °C over 12 h. The resulting slurry was stirred at
25ꢀ35 °C for 1 h and allowed to cool to 5ꢀ10 °C over 2 h, followed
by stirring at 5ꢀ10 °C for 6 h before filtration. The wet cake was
displacement washed with cold 1:1 MeCN/toluene (80 L,
5ꢀ10 °C), 1:9 MeCN/toluene (80 L) and MTBE (160 L) and
dried in vacuum oven at 45 °C to afford 1 (22.9 kg) as an off-white
solid with >99% purity. 90% yield. 1H NMR (400 MHz, DMSO-d6)
δ 8.63 (d, J = 2.8 Hz, 1 H), 8.15 (brs, 2 H), 8.06 (d, J = 8.7 Hz, 1 H),
7.66 (dd, J = 8.7, 2.8 Hz, 1 H), 7.61 (d, J = 2.3 Hz, 1 H), 7.57ꢀ7.50
(R)-1-(Triisopropylsilyloxy)propan-2-ol (22). To a solution of
(R)-1,2-propanediol (21, 10 kg, 131.4 mol) and imidazole
(11.6 kg, 170.8 mol) in acetonitrile (60 L) at 0 °C was added
triisopropylchlorosilane (26.6 kg, 138 mol) over 3 h at 0ꢀ5 °C. The
resulting slurry was stirred at 0ꢀ5 °C for additional 1 h followed by
at 20ꢀ25 °C for 1ꢀ3 h until the reaction was deemed complete
(1,2-propanediol <1.0% by GC). The reaction was quenched by
addition of 15% NaCl (100 L) and toluene (80 L). The organic
layer was separated and washed with 15% NaCl (50 L). GC assay:
28.28 kg of 22. 93% yield. Analytically pure sample could be
1
prepared by distillation under reduced pressure. H NMR (400
MHz, CDCl3) δ 3.98ꢀ3.80 (m, 1 H), 3.68 (dd, J = 9.7, 3.6 Hz,
1 H), 3.44 (dd, J = 9.7, 7.7 Hz, 1 H), 2.50 (br, 1 H), 1.15ꢀ1.04 (m,
21 H), 1.13 (d, J = 6.4 Hz, 3 H). 13C NMR (100 MHz, CDCl3)
δ 68.81, 68.05, 18.14, 17.91, 11.86. Anal. Calcd for C12H28O2Si: C,
62.01; H, 12.14. Found: C, 61.78; H, 12.26.
(R)-1-(Triisopropylsilyloxy)propan-2-yl Methanesulfonate
(19). The crude solution of 22 (28.28 kg assay, 121.7 mol) was
azeotropically concentrated (batch temperature <60 °C) to ∼50 L
and diluted with toluene to 310 L. The resulting solution was cooled
to 0 °C. Triethylamine (17.2 kg, 170.3 mol) was charged followed
by addition of methanesulfonyl chloride (16.7 kg, 146 mol) over 2 h
at 0ꢀ5 °C. The resulting slurry was stirred at 0ꢀ5 °C until the
reaction was deemed complete by GC assay. The reaction was
quenched by addition of water (170 L), and the resulting mixture
was allowed to warm to 20ꢀ25 °C. The organic layer was washed
with H2O (85 L). The resulting solution was azeotropically
concentrated under vacuum (batch temperature <60 °C) to give
19 as a clear solution (GC assay: 60.6 wt %, 35.5 kg of 19, 94%
yield). Analytically pure sample could be prepared by silica gel
chromatography. 1H NMR (400 MHz, CDCl3) δ 4.82ꢀ4.74 (m, 1
H), 3.81 (dd, J = 11.1, 6.9 Hz, 1 H), 3.74 (dd, J = 11.1, 3.8 Hz, 1 H),
3.04 (s, 3 H), 1.41 (d, J = 6.5 Hz, 3 H), 1.10ꢀ1.05 (m, 21 H). 13C
NMR (100 MHz, CDCl3) δ80.63, 66.35, 38.37, 17.92, 17.82, 11.88.
HRMS: m/z [M þ H]þ calcd for C13H30O4SSi: 311.1712. Found:
311.1713.
3-(6-Ethanesulfonyl-pyridin-3-yloxy)-5-((S)-2-hydroxy-1-
methyl-ethoxy)-benzoic Acid DABCO salt (2:1) (7). A crude
solution of 19 in toluene (19.57 kg, 60.6 wt %, 38.2 mol) was
diluted with dry AcNMe2 (59.5 L). Cesium carbonate
(powdered, 14.36 kg, 44.1 mol) was charged followed by 4
(9.91 kg, 29.4 mol) while maintaining a vigorous agitation. The
reaction mixture was stirred vigorously at 80 °C for 8ꢀ12 h until
>99% conversion was achieved. The reaction mixture was then
cooled to 0 °C and diluted with MTBE (79 L). H2O (40 L) was
charged slowly at <10 °C. After a phase cut at ambient
temperature, the organic phase (HPLC assay: 15.15 kg of 20,
93% yield) was solvent switched to THF at a final volume of 89
L. MeOH (30 L) was added and the batch was cooled to 0 °C.
NaOH (5 N, 8.8 L, 43.9 mol) was charged at <5 °C. The
resulting solution was stirred at 0ꢀ5 °C for 1 h followed by at
20ꢀ25 °C for 2ꢀ6 h until >99.5% conversion was achieved.
HCl (4 M, 27.5 L, 109.8 mol) was added. The resulting hazy
829
dx.doi.org/10.1021/op200068c |Org. Process Res. Dev. 2011, 15, 824–830

Organic Process Research & Development
ARTICLE
(m, 1 H), 7.37ꢀ7.35 (m, 1 H), 7.02 (t, J = 2.0 Hz, 1 H), 6.57 (d, J =
2.3 Hz, 1 H), 4.64ꢀ4.56 (m, 1 H), 3.77 (s, 3 H), 3.55 (dd, J = 11.3,
5.8 Hz, 1 H), 3.51 (dd, J = 11.3, 4.6 Hz, 1 H), 3.41 (q, J = 7.3 Hz, 2
H), 2.48 (s, 3 H), 1.24 (d, J = 6.2 Hz, 3 H), 1.14 (t, J = 7.3 Hz, 3 H).
13C NMR (100 MHz, DMSO-d6) δ 163.13, 159.77, 156.46, 155.52,
150.10, 146.81, 141.27, 137.05, 131.31, 125.95, 124.38, 111.63,
111.06, 97.56, 75.34, 64.28, 46.36, 39.76, 38.48, 16.54, 6.95. Anal.
Calcd for C22H28N4O9S2: C, 47.47; H, 5.07; N, 10.07. Found: C,
47.35; H, 5.18; N, 10.08.
(8) Johnstone, C.; Mckerrecher, D.; Pike, K. G.; Waring, M. J. PCT
Int. Appl. WO/2005/121110, 2005, CAN 144:69853.
(9) This is consistent with the observed trend of the ratio of 4:9.
Further discussion about the reaction kinetics is beyond the scope of this
manuscript, because the kinetic profile is affected by multiple compo-
nents including the reactivity of the mono- and di- anion of 3 and related
multiple acidꢀbase equilibriums.
(10) The use of a larger excess of 3 resulted in increased levels of 10
and could also cause difficulties in the removal of unconsumed 3.
(11) The monoanion of 3 seemed to be more reactive or at least as
effective as the corresponding dianion for SNAr displacement, which
could be attributed to its solubility or actual concentration of the species
in the reaction solution, as the O-arylation still gave >98% conversion
and ∼75% yield even with a mole ratio for t-BuOK/3 as low as 1.4:1. As
evidence, the use of a large excess of monoanion of 3 gave an excellent
conversion and selectivity, although it was not practical.
’ ASSOCIATED CONTENT
1
Supporting Information. H and 13C NMR spectra. This
S
b
material is available free of charge via the Internet at http://pubs.
acs.org.
(12) In line with this conclusion, the formation of 9 also depended
on the water content in the reaction mixture, which partially quenched t-
BuOK. However, the reaction could still tolerate a moisture level (KF) of
up to ∼1700 ppm under optimal conditions, while the formation of 9
was suppressed to <0.5%.
(13) Interestingly, higher levels of impurity 9 were also observed
when the reaction was conducted in the presence of air, implying that the
formation of 9 could occur via radical-type mechanism. This would
rationalize the fact that reducing the amount of base to less than 2 mol
equiv made the radical pathway more competitive and hence generated
substantially higher levels of 9.
(14) The use of 2-propanol (∼10%) made the crystallization of 4
reproducible, presumably by keeping the unconsumed 3 from
precipitating out.
(15) Upon heating, CsHCO3 could turn into Cs2CO3 and water with
evolution of CO2. See: Pushnyakova, V. A.; Berger, A. S.; Kirgintsev, A. N.
Zh. Neorg. Khim. 1973, 18, 311.
(16) 1,2-Propandiol could be converted to the corresponding bis-
mesylate in the subsequent step, which could then react with 4 to
generate (R)-OH-isomer 15.
(17) The use of other bases such as Na2CO3 or DBU gave lower
conversions. The use of 2 equiv of powered K2CO3 and 1.3 equiv of 19
in AcNMe2 at 100 °C afforded 20 in 83% assay yield (10% lower than the
use of Cs2CO3) after 20 h. However, the formation of 11 was increased
to 3% (vs <1% with Cs2CO3).
’ AUTHOR INFORMATION
Corresponding Author
*naoki_yoshikawa@merck.com; feng_xu@merck.com
’ ACKNOWLEDGMENT
We thank Dr. P. Dormer for assistance with NMR studies and
Dr. T. Chasse and Ms. E. Kolodziej for analytical assistance.
’ REFERENCES
(1) For recent reviews, see: (a) Iynedjian, P. B. Cell. Mol. Life Sci.
2009, 66, 27. (b) Kawai, S.; Mukai, T.; Mori, S.; Mikami, B.; Murata, K.
J. BioSci. Bioeng. 2005, 99, 320 and references cited therein.
(2) For recent reviews, see: (a) Matschinsky, F. N. Nat. Rev. Drug
Discovery 2009, 8, 399. (b) Pal, M. Curr. Med. Chem. 2009, 16, 3858.
(3) Iino, T.; Hashimoto, N.; Nakajima, H.; Takahashi, K.; Nishimura,
T.; Eiki, J. PCT Int. Appl. WO/2004/076420, 2004, CAN 141:260754.
(4) Asakawa, K.; Sawada, N.; Tsuritani, T.; Itoh, T.; Mase, T.;
Takahashi, K.; Xu, F.; Yoshikawa, N. PCT Int. Appl. WO/2009/
041475, 2009, CAN 150:398533.
(5) For recent examples of direct mono-O-arylation of resorcinol
derivatives via a SNAr reaction, see: (a) Yates, M. H.; Koenig, T. M.;
Kallman, N. J.; Ley, C. P.; Mitchell, D. Org. Process Res. Dev. 2009,
13, 268.(b) Nilsson, P.; Katkevics, M.; Pelcman, B. PCT Int. Appl. WO/
2009/127822, 2009, CAN 151:490887. (c) Rossom, W. V.; Ovaere, M.;
Meervelt, L. V.; Dehaen, W.; Maes, W. Org. Lett. 2009, 11, 1681. (d) Xie,
C.; Sullivan, K. A.; Laurila, M. E.; Mitchell, D. N.; Pu, Y. J. Synth.
Commun. 2008, 38, 21.(e) Martin, N. G.; McKerrecher, D.; Pike, K. G.;
Waring, M. J. PCT Int. Appl. WO/2008/050117, 2008, CAN
148:472080. (f) Ryono, D. E.; Cheng, P. T. W.; Bolton, S. A.; Chen,
S. S.; Shi, Y.; Meng, W.; Tino, J. A.; Zhang, H.; Sulsky, R. B. PCT Int.
Appl. WO/2008/005964, 2008, CAN 148:144884. (g) Pearson, A. J.;
Ciurea, D. V.; Velankar, A. Tetrahedron Lett. 2008, 49, 1922. (h)
Lyubimtsev, A.; Vagin, S.; Syrbu, S.; Hanack, M. Eur. J. Org. Chem.
2007, 2000. (i) Yu, C. R.; Xu, L. H.; Tu, S.; Li, Z. N.; Li, B. J. Fluor. Chem.
2006, 127, 1540.(j) Caulkett, P. W. R.; McKerrecher, D.; Newcombe,
N. J.; Pike, K. G.; Waring, M. J. PCT Int. Appl. WO/2006/125958, 2006,
CAN 146:27848. (k) Wu, K.; Wang, Y.; Zheng, H. CN1597657, 2005,
CAN 144:170781. (l) Wen, Y.; Fang, Z.; Liu, W. J. Zhejian Univ., Sci.
2004, 5, 956. (m) Storm, J. P.; Andersson, C. M. J. Org. Chem. 2000,
65, 5264.
(18) The unconsumed 4 could be transformed to 26 upon hydro-
lysis, which could further react with 7 and 8 to form 27, subsequently.
(19) For example, byproduct 28 could be formed.
(6) For a synthesis of differentially substituted resorcinol derivatives
using 1,3-difluorobenzene as the starting material, see: Kim, A.; Powers,
J. D.; Toczko, J. F. J. Org. Chem. 2006, 71, 2170.
(7) For a synthesis of m-phenoxyphenol from resorcinol and
bromobenzene, see: Mil’to, V. I.; Orlov, V. Y. RU2287516, 2006,
CAN 145:488993.
830
dx.doi.org/10.1021/op200068c |Org. Process Res. Dev. 2011, 15, 824–830