Route development and multikilogram GMP delivery of a somatostatin receptor antagonist

Article abstract of DOI:10.1021/op300128c

Route development and demonstration on multikilogram scale for the first GMP delivery of MK-4256 are described. Key aspects of the convergent route include a regioselective green iodination, one-pot oxadiazole synthesis, and an efficient ketone Pictet-Spengler reaction with diastereomeric upgrade via crystallization to afford 6 kg of API. A recycle procedure augmented the yield of desired diastereomer in the Pictet-Spengler reaction from a mixture of diastereomers heavily enriched in the undesired diastereomer.

Full text of DOI:10.1021/op300128c

Concept Article
pubs.acs.org/OPRD
Route Development and Multikilogram GMP Delivery of a
Somatostatin Receptor Antagonist
Rebecca T. Ruck,*,† Mark A. Huffman,† Gavin W. Stewart,‡ Ed Cleator,‡ Wynne A. Kandur,†
Mary M. Kim,† and Dalian Zhao†
†Department of Process Chemistry, Merck Research Laboratories, Merck & Co., Inc., Rahway, New Jersey 07065, United States
‡Department of Process Chemistry, Merck Sharp and Dohme Research Laboratories, Hertford Road, Hoddesdon, Hertfordshire
EN11 9BU, United Kingdom
S
* Supporting Information
ABSTRACT: Route development and demonstration on multikilogram scale for the first GMP delivery of MK-4256 are
described. Key aspects of the convergent route include a regioselective green iodination, one-pot oxadiazole synthesis, and an
efficient ketone Pictet−Spengler reaction with diastereomeric upgrade via crystallization to afford 6 kg of API. A recycle
procedure augmented the yield of desired diastereomer in the Pictet−Spengler reaction from a mixture of diastereomers heavily
enriched in the undesired diastereomer.
oximinoacetate (16), is commercially available on small scale.
The anhydride/pyridine conditions used for most oxadiazole
formations are known to be less successful with this substrate;4
in our hands, acetylation proceeded cleanly to generate
intermediate 17, but cyclization to furnish oxadiazole ethyl
ester 18 performed poorly under the pyridine conditions. As an
alternative approach, after isolation of 17, acetic acid was
identified from a screen of solvents for the desired cyclization
reaction. Acetic acid was then examined for the acetylation
reaction and found to be suitable for that step as well, allowing
for a one-pot process. The cost and availability of ethyl α-
aminooximinoacetate on-scale proved prohibitive, necessitating
preparation of amidoxime 16. A literature report described the
treatment of ethyl cyanoformate (15) with hydroxylamine free
base in dichloromethane to provide this intermediate in modest
yield.5 We found acetic acid to be a superior solvent for this
reaction as well; hydroxylamine hydrochloride salt could be
used in the presence of sodium acetate as base. This result
allowed all three reactions to be combined into a one-pot
process to convert ethyl cyanoformate to oxadiazole ethyl ester
18 (Scheme 3). Headspace monitoring showed negligible HCN
formation during this chemistry, although a cyanide monitor
was placed in close proximity to the vessel, and airline hood
repirators were employed to avoid potential exposure. In
collaboration with our EPSE group, we identified specific safety
considerations for this process that are detailed in the
Experimental Section. This reaction was run successfully on-
scale in a single flask (7.24 kg, 95% assay yield), and an ethyl
acetate solution of compound 18 was carried forward for
conversion to oxadiazole K-salt 4.
INTRODUCTION
■
Substituted β-carboline derivatives, including MK-4256 (1)
have been shown to be selective antagonists of the somatostatin
subtype receptor (SSTR3) and useful for treatment of type 2
diabetes and its associated conditions. The compounds are also
useful for the treatment of depression and anxiety.1 The
discovery synthesis of 12 provided a convergent route to API
(Scheme 1) via two key intermediates: tryptamine 2 and ketone
3. Ketone 3 was, in turn, derived from oxadiazole K-salt 4 and
iodopyrazole 5.
Four key areas were identified to improve the viability of this
process on-scale (Scheme 2):
(1) Oxadiazole methyl ester 9 was prepared in only 19%
yield over three steps employing POCl3 as solvent for
one step, necessitating a new synthesis of this
intermediate.3
(2) 4-Iodo-N-methylpyrazole (5) was purchased for earlier
efforts; the cost and lead-time on-scale required
identification of a cost-effective preparation.
(3) Tryptamine derivative 2 was previously synthesized as
the HCl salt with no formal isolation; since this material
was to be prepared in bulk in England and used for
processing in the United States, identification of an
isolable, crystalline salt was critical for shipping.
(4) The final Pictet−Spengler reaction and subsequent
chromatography were inefficient, providing 1 in <10%
yield; success toward improving the chemical productiv-
ity of this reaction and the efficient separation of
diastereomers was paramount for the completion of this
multikilogram delivery.
Replacement of the methyl ester in the medicinal chemistry
route with the ethyl ester in the process chemistry route
simplified isolation of oxadiazole K-salt 4. The ethyl acetate
solution of oxadiazole ethyl ester 18 was solvent switched to
RESULTS AND DISCUSSION
■
Oxadiazoles are known products of the reactions of amidoximes
with carboxylic anhydrides, generally in the presence of
pyridine. To generate the necessary oxadiazole ethyl ester
derivative here, the requisite amidoxime, ethyl α-amino-
Received: May 17, 2012
© XXXX American Chemical Society
A
dx.doi.org/10.1021/op300128c | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Concept Article
Scheme 1. Convergent synthesis of MK-4256 (1)
Scheme 2. Discovery Chemistry route to 1
Scheme 3. Oxadiazole ethyl ester 18 synthesis
absence of a hydrogen bond-donating solvent. Maintaining a
satisfactory concentration of ethanol served to prevent
evaporation of the product. Saponification in ethanol as solvent
was efficient, leading to direct crystallization of oxadiazole K-
salt 4. Use of 10 N KOH limited the water concentration,
leading to a high recovery (Scheme 4). Oxadiazole K-salt was
isolated in 77% yield over two steps (6.33 kg).
ethanol; great care was taken during this solvent switch as
compound 18 was volatile under these conditions in the
The medicinal chemistry route had employed a two-step,
one-pot sequence for the preparation of oxadiazole Weinreb
B
dx.doi.org/10.1021/op300128c | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Concept Article
Scheme 4. Preparation of oxadiazole Weinreb amide 10
amide 10: the K-salt 4 was converted to the corresponding acid
chloride, using oxalyl chloride in dichloromethane, and
Weinreb amide was formed under Schotten−Bauman con-
ditions. The presence of excess oxalyl chloride led to formation
of the corresponding bis-Weinreb amide 19, which complicated
the subsequent ketone formation step; a flushing procedure
with toluene was employed to remove excess reagent. Efforts to
reproduce this chemistry using one equivalent of oxalyl chloride
led to decreased conversion to the acid chloride and low yield
of the desired Weinreb amide. It was found that reacting
oxadiazole K-salt and Weinreb amine·HCl with EDC·HCl in
dichloromethane led to rapid formation of oxadiazole Weinreb
amide 10 (Scheme 4). The effectiveness of this reaction using
the three salts in DCM was advantageous because of the facile
solvent switch to THF for the next step. In one batch, 6.31 kg
of oxadiazole Weinreb amide was isolated as a 42 wt % solution
in THF after solvent switch (98% assay yield).
iodination of N-methylpyrazole (20). (1) N-Iodosuccinimide
and catalytic trifluoroacetic acid in acetonitrile were successful
reaction conditions, though the cost of NIS was deemed a
liability (Scheme 5a). (2) Use of 0.5 equiv each of iodine and
ceric ammonium nitrate in acetonitrile also worked, although
the considerable quantity of cerium byproduct rendered the
workup extremely challenging (Scheme 5b). (3) In the best
procedure, the second option was modified to employ 0.6 equiv
of hydrogen peroxide as oxidant and water as the only solvent.8
This result provided a completely green system for the
iodination of N-methylpyrazole (Scheme 5c); the product is
directly filtered from the reaction mixture, and water is the only
byproduct. EPSE evaluation of the chemistry identified that the
reaction could be conducted safely at a temperature below 70
°C. This procedure was used to prepare 16.2 kg of 4-iodo-N-
methylpyrazole in a single batch (91% yield). These conditions
were subsequently reported for the successful 4-selective
iodination of a variety of pyrazole compounds.9
Preparation of ketone 3 followed the medicinal chemistry
procedure, with only a minor change to the workup and
development of a crystallization protocol that led to a dramatic
improvement in yield. 4-Iodo-N-methylpyrazole (5) underwent
magnesium−halogen exchange with isopropylmagnesium
chloride. Addition of a THF solution of oxadiazole Weinreb
amide 10 led to clean conversion to form pyrazole−oxadiazole
ketone 3. A small amount of direct isopropyl addition was also
identified in the reaction mixture (Scheme 6). Dichloro-
4-Iodo-N-methylpyrazole (5) is commercially available and
had been purchased for medicinal chemistry’s efforts. The price
was high and the lead time prohibitively long for the quantity
required for the first GMP delivery (Table 1). Two approaches
Table 1. Cost and lead-time comparison of pyrazole starting
materials
cmpd
lead time (wks)
cost/kg
cost/mol
4-iodo-N-methylpyrazole (5)
4-iodopyrazole
6
4
2
$3360
$830
$956
$699
$161
N-methylpyrazole (20)
$78.50
Scheme 6. Preparation of ketone 3
were considered to rapidly access compound 5: (1)
methylation of 4-iodopyrazole and (2) iodination of 1-methyl
pyrazole.6 The starting material for the former approach also
proved expensive, especially since addition of the methyl group
does not have much impact on the molecular weight of this
intermediate. On the basis of this cost analysis and in-stock
availability, N-methylpyrazole emerged as the preferred starting
material.
methane replaced ethyl acetate as workup solvent, due to low
product solubility in ethyl acetate (as well as all other organic
solvents considered). Crystallization of the ketone occurred
easily from a variety of solvent combinations. Given the
There are many known conditions for iodination of
pyrazoles, though each one suffers from its own drawbacks.7
Several sets of conditions were shown to be effective for the
Scheme 5. Iodination of N-methylpyrazole 5
C
dx.doi.org/10.1021/op300128c | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Concept Article
Scheme 7. Tryptamine 2 synthesis
Scheme 8. Pictet−Spengler reaction products formed
presence of THF in the organic extracts from the reaction, a
THF/heptane solvent system was employed for crystallization.
Two batches of ketone solution were combined for solvent
switch to THF and isolation. Pyrazole−oxadiazole ketone 3 was
isolated in 87% yield over two steps (6.26 kg).
The medicinal chemistry procedure for Boc-deprotection of
compound 14 employed HCl in dioxane to afford the
corresponding HCl salt. However, isolation was difficult, and
it was necessary to dry the product by vacuum over several days
to isolate a foam. We decided to investigate whether Boc-
deprotection mediated by a sulfonic acid and concomitant salt
formation would represent a viable alternative on scale. Initially,
a THF solution of 14 was heated to 60 °C in the presence of
MSA. These conditions furnished the desired amine after one
hour, but the product failed to crystallize from the reaction
mixture. When the same protocol was applied using
TsOH·H2O, the desired product crystallized to afford a 48%
yield of the bis-TsOH salt. The isolated yield of the bis-tosylate
salt of 2 was improved to 94% by using toluene as antisolvent;
however, the small particle size of the resultant crystals led to a
slow filtration that was impractical for large-scale implementa-
tion. A solvent screen of this deprotection/salt formation
sequence identified acetonitrile as the optimal solvent. Thus, a
solution of Boc-amine 14 in acetonitrile was heated to 60 °C in
the presence of 3 equiv of TsOH·H2O until complete
deprotection was observed. The reaction mixture was then
cooled to 20 °C, leading to crystallization of salt 2. This solid
was composed of much larger particles and was filtered with
ease. An 85% isolated yield (29.4 kg) of this bis-tosylate salt of
2 was achieved on-scale without the use of antisolvent.
The Pictet−Spengler reaction represented a convergent
method for the construction of 1. This final step also served
to identify the GMP range for this synthesis, with ketone 3 and
tryptamine 2 serving as GMP starting materials. There are
limited examples of Pictet−Spengler reactions using ketones as
substrates in the literature; moreover, these reactions generally
require strong acid catalysis and furnish products in poor
diastereoselectivities. The original synthesis of 1 employed
harsh conditions to effect this transformation: the two reaction
components were heated in pyridine at 70−85 °C for 3 d
(Scheme 8). An excess of ketone 3 was used because it was
removed more easily than the tryptamine 2 during chromatog-
raphy. The diastereomers 1 and 22 were formed in variable
The synthesis of tryptamine 2 supplied by medicinal
chemistry (identified as the HCl salt of 2) was deemed fit for
purpose, albeit with a few requisite process-related improve-
ments. 2-Bromo-4′-fluoroacetophenone (12) had been used as
the medicinal chemistry starting material; however, bulk
supplies of this reagent were not available in an adequate
timeframe for this delivery. Hence, for this campaign, the
reaction with the corresponding chloride 21 was investigated.
The coupling of Boc-D-Trp-OH (11) with chloride 21
proceeded in quantitative yield in less than 2 h at 20 °C in
DMF. Compound 13 was isolated on small scale by
concentrating the reaction mixture to dryness after aqueous
workup. Since this was not an option on scale, we decided to
telescope this reaction with the subsequent imidazole
formation. This through process was achieved by partitioning
the reaction mixture between water and ethyl acetate then
switching the organic solvent to toluene: the resulting stream of
compound 13 was used directly in the next step. This protocol
had two additional benefits: the imidazole formation performed
very well in toluene (a more desirable solvent than p-xylene
previously used), and the equivalents of ammonium acetate
required to effect this transformation could be reduced (five vs.
ten in the prior procedure). The medicinal chemists partitioned
imidazole 14 between water and ethyl acetate then partially
concentrated the organics for crystallization. This crystallization
was suboptimal: it was necessary to chromatograph the mother
liquors to recover product in adequate yield. An alternative
process was developed that involved partitioning the reaction
mixture between water and ethyl acetate then solvent switch to
toluene for crystallization. These changes were implemented
on-scale (Scheme 7): Boc-D-Trp-OH (11) (19.0 kg) was
processed to afford imidazole 14 in 83% yield over two steps
with excellent retention of enantiopurity (99.6% ee).
D
dx.doi.org/10.1021/op300128c | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Concept Article
ratios up to a maximum of 1.5:1. At this point, we identified a
major impurity (23) arising from elimination of the oxadiazole
and oxidation to the corresponding pyridine, in addition to
many small impurities likely resulting from decomposition of
the tryptamine fragment. Assay yield for the Pictet−Spengler
reaction was variable from 15 to 30%, a fact potentially
attributable to excess HCl present in the unpurified
tryptamine·HCl salt. The isolated yield was further compro-
mised by inefficient chromatography to separate the two
diastereomers; isolated yields ranged from 5 to 10% of
amorphous free base after chromatography and lyophilization
from acetonitrile/water.
salt (25) under the strongly basic workup conditions and could
be washed into the aqueous layer along with water-soluble
oxadiazole byproduct (Scheme 9).
Scheme 9. Ketone workup hydrolysis
Extensive screening efforts to identify improved conditions
for the Pictet−Spengler reaction identified DMSO and ethanol
as viable alternatives to pyridine. A new impurity, correspond-
ing to elimination of the oxadiazole without oxidation (24),
also formed in some solvents and formed almost exclusively in
acetic acid. Ultimately, DMSO was chosen as solvent because
the reaction rate was faster relative to that in ethanol, and the
reaction itself could be carried out under more concentrated
conditions.
The final Pictet−Spengler reaction conditions used for the
first GMP delivery are illustrated in Scheme 10. Two batches
were completed, each providing 1 and its diastereomer 22 in a
better than 58:42 ratio and a combined >88% assay yield.
Identification of an effective crystallization protocol obviated
the need for chromatographic separation of diastereomers. The
combination of toluene and different cosolvents (methanol,
isopropanol, dichloromethane, acetonitrile) provided the first
crystalline forms of 1. On crystallization from 94/6 v/v
toluene/acetonitrile, the supernatant contained >5:1 22/1,11
efficiently removing 22 to produce 1, following filtration, in
high diastereomeric purity (dr ≈ 99.5:0.5) and good recovery
(41−42% isolated yield, 6.01 kg over two batches).
The crystalline solid obtained from the diastereomeric
upgrade above was identified as the toluene solvate Form I.
On drying in the vacuum oven, Form I lost toluene and
converted to a new form characterized as Form III. In our prep
lab ovens, this drying procedure was inefficient and required ∼3
w at 69 °C at 120 Torr with a nitrogen sweep to achieve a
specification of <0.149 wt % toluene. However, drying at 90 °C
led to formation of the oxadiazole elimination byproduct 24;
thus, we decided to retain the lower temperature point. The
final process is illustrated in Scheme 11.
The diastereomers 22 and 1 were found to interconvert
under acidic conditions without enantiomeric erosion. Acetic
acid and acetonitrile were suitable solvents for the epimeriza-
tion using sulfonic acids or trifluoroacetic acid. Greater than
one equivalent of acid was necessary as the first equivalent was
neutralized by the basic imidazole group. A procedure was
developed in which the filtrate from the initial 1 crystallization,
containing ∼5.5:1 ratio of the undesired diastereomer 22 to 1,
was equilibrated to a 1:1 ratio, and a second crop of API was
collected by repeating the diastereoselective crystallization. This
allowed for recovery of more than 2 kg of additional 1 from the
campaign in a single recycle.
The particular acid used to catalyze the Pictet−Spengler
reaction played a pivotal role in preventing byproduct
formation. Lower pKa’s led to faster reaction rates, but also
faster decomposition rates and lower diastereoselectivities.
Trifluoromethanesulfonic acid with tryptamine free base and
tryptamine·2TsOH emerged as the two best acid candidates.
Since tryptamine 2 was isolated as the bis-TsOH salt, there was
a driving force to employ this material as-is without the
requirement to break the salt. Additionally, this option offered
greater ease of handling, making a compelling case for entering
into the reaction with tryptamine·2TsOH (2) directly. The
stoichiometry of acid used was found to have a similar effect on
the Pictet−Spengler reaction as the pKa of the acid used: the
presence of more acid led to a faster reaction rate, but also to
faster elimination to byproduct 24. One equivalent of strong
acid provided the optimal balance between reaction rate and
purity. With two equivalents of p-TsOH entering into the
system, one equivalent of base was used to attenuate the acidity.
Alkali acetates and hydroxides (sodium, potassium, cesium) and
tert-butoxides (sodium, potassium) were screened, and sodium
acetate as base provided the best reaction purity and greatest
ease of handling (lowest hygroscopicity) while simultaneously
leading to the beneficial formation of one equivalent of the
conjugate weak acid.
With base and solvent now identified, final details for the
optimization could be set. One equivalent of tetraethyl
orthosilicate [Si(OEt)4] served to increase the reaction rate,
likely by favoring imine formation and sequestering water.10
The presence of a silicon byproduct complicated the workup of
this reaction: this byproduct exhibited different physical
properties at varying pH values. Ultimately, the strongly basic
conditions employed for workup led to formation of a dense
silicate oil that formed a third layer in the extraction vessel and
could be removed from the reaction mixture. The reaction
temperature was decreased from 100 to 75 °C, minimizing
impurity formation, while still allowing an improved reaction
time of 40 h. The final stoichiometry used was one equivalent
of tryptamine·2TsOH (2) and 1.1 equivalents of pyrazole−
oxadiazole ketone 3. The goal was to maximize consumption of
the tryptamine fragment, and 96−97% conversion was achieved
under these conditions. The ketone was shown to hydrolyze to
the corresponding N-methylpyrazole-4-carboxylic acid sodium
CONCLUSIONS
■
We have presented the chemistry toward the first GMP delivery
of 1. Improved chemistry was employed to isolate the
tryptamine fragment as a bis-TsOH salt, enabling its direct
use in the final Pictet−Spengler reaction. New chemistry was
developed for synthesis of the requisite oxadiazole fragment
that employed a safe, multicomponent sequence conducted
fully in acetic acid. A novel green 4-iodination of N-methyl
pyrazole was identified and applied to a series of pyrazole
compounds. Optimized conditions for the ketone Pictet−
Spengler reaction led to a high assay yield, and a
diastereoselective crystallization of 1 afforded the desired API
in high purity. A recycle protocol was developed to isolate
additional 1 from a mixture enriched in the undesired
E
dx.doi.org/10.1021/op300128c | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Concept Article
Scheme 10. Final Pictet−Spengler reaction conditions
Scheme 11. Final GMP delivery route
diastereomer. A total of 8 kg of API was obtained using these
EXPERIMENTAL SECTION
■
General. HPLC analysis was performed on an Agilent 1100
instrument with Zorbax Eclipse Plus C18 4.6 mm × 50 mm;
MeCN/H2O (0.1% H3PO4) 10/90 to 90/10 0−5 min: hold at
90/10 5−6 min; 1.5 mL/min; 210 nm; 25.0 °C, 8 min run
two methods.
F
dx.doi.org/10.1021/op300128c | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Concept Article
time. All compounds described are fully characterized in the
literature, and data obtained here matched literature
information; relevant references are provided.
1,2,4-Oxadiazole-3-carboxylic acid, 5-methyl-ethyl ester
(18).12 A 100-L RBF in a water bath was charged with
hydroxylamine·HCl (4.23 kg, 60.8 mol), followed by acetic acid
(24.86 L). Ethyl cyanoformate (4.95 kg, 49.4 mol) was added,
then sodium acetate (4.99 kg, 60.8 mol) was added over 75
min.
salt as a white crystalline solid matching spectroscopic data
from the literature14 and that of material purchased
commercially. The mother liquor contained 185 g (2.2%)
product and 626 g (8%) unreacted ethyl ester by HPLC assay.
1,2,4-Oxadiazole-3-carboxamide,N-methoxy-N,5-dimeth-
yl- (10).1,2 A 100-L glass extraction vessel was charged with
dichloromethane (20 L), followed by oxadiazole K-salt 4 (6.25
kg, 37.6 mol). N,O-Dimethylhydroxylamine·HCl (4.68 kg, 47.0
mol) was added to the slurry. Additional dichloromethane
(11.25 L) was added to rinse the vessel, and the batch
temperature was lowered to 8 °C. EDC·HCl (8.29 kg, 43.3
mol) was added portionwise over 1.5 h, with the temperature
increasing to 22 °C. After an additional 10 min, Weinreb amide
formation was complete by HPLC (>97% conversion, Weinreb
amide/carboxylic acid); DCM (11.6 L) and H2O (18.6 L) were
added to the reaction mixture which was stirred vigorously. The
layers were separated. The DCM layer was returned to the glass
extraction vessel and washed with 4.2% NaHCO3(aq) (14.6 L).
The DCM solution was then transferred, via an inline filter, to a
100-L RBF and concentrated under vacuum. THF was added
during the distillation to complete the solvent switch. The
Weinreb amide was isolated as a solution in THF (total mass =
14.9 kg, 42.3 wt %). Spectroscopic data were consistent with
that reported in the literature.1,2
CAUTION: Ethyl cyanoformate is typically used as an
acylating agent with cyanide as byproduct. Even though
headspace monitoring suggested the absence of this competing
pathway, this reaction sequence was conducted in the presence
of a cyanide monitor to ensure avoidance of exposure to HCN.
On the basis of the decomposition temperature of α-
hydroxylamine·HCl in the presence of sodium acetate, our
EPSE group identified that this step was safe to run at T < 35
°C. The reaction was complete after 2 h at 18−28 °C. The
reaction mixture was cooled in an ice bath to 15 °C and then
exhibited a slow exotherm during acetylation; our EPSE group
recommended that the addition be conducted at T < 30 °C,
and that heatup proceed slowly and under well-vented
conditions to avoid pressure buildup. Acetic anhydride (8.76
kg, 86 mol) was added via addition funnel over 90 min. The
temperature increased to 26 °C during the addition. The
reaction mixture was stirred for an additional 15 min.
Acetylation to form 17 was complete, with no detectable
oxime 16. The reaction mixture was heated at 99 °C for 12 h.
HPLC analysis of the reaction mixture showed 99.66%
conversion to oxadiazole 18 (call point: >99% conversion).
The slurry was cooled to room temperature, and acetic acid was
removed under vacuum. Evaporation was discontinued when a
ratio of 1.6:1 of acetic acid/18 was observed in the reaction
4-Iodo-1-methylpyrazole (5).10 A 100-L RBF with a water
bath was charged with N-methylpyrazole (7.0 kg, 85.3 mol) and
water (30 L), followed by iodine (11.0 kg, 43.5 mol). After a 10
min age, 4-iodo-N-methylpyrazole (15 g) was charged as seed.
30% H2O2 (5.2 L, 51.2 mol) was charged over 30 min, the
temperature increasing from 18 to 32 °C. After 42 h of stirring,
the slurry was cooled to 10 °C. Ten percent NaHSO3(aq) (20 L)
was slowly charged as quench, keeping the internal temperature
below 20 °C. The resulting light-tan slurry was stirred at room
temperature for 3 h and isolated by filtration. The wet cake was
washed with water (5 × 20 L). The wet cake was dried in the
vacuum oven (30−35 °C, ∼200 mm vac, N2 sweep) for 3 d.15
The isolated yield was 91% (16.2 kg, 98.1 wt % by HPLC).
This compound is commercially available; spectroscopic data
were consistent with that of authentic material.
Methanone, (5-methyl-1,2,4-oxadiazol-3-yl)(1-methyl-1H-
pyrazol-4-yl)- (3).1,2 A 100-L glass extraction vessel was
charged with 1-methyl-4-iodo-pyrazole (4.49 kg, 21.2 mol)
and THF (9.0 L) under N2. The reaction mixture was cooled to
−19 °C. Isopropylmagnesium chloride (2.0 M in THF, 11.1 L,
22.1 mol) was added over 1.75 h; the temperature was
maintained below 5 °C during the addition. The solution was
assayed by quench of a sample into methanol for 100%
conversion of the iodo compound to the protio compound by
HPLC. Oxadiazole Weinreb amide 10 (3.15 kg, 18.4 mol) was
then added as a solution in THF (5.33 L) over 25 min, the
temperature increasing from −9 to 21 °C. After the addition,
the reaction mixture was stirred at room temperature. Ketone
formation was complete by HPLC after 10 min; the reaction
mixture was cooled to 8 °C. HCl (2 N, ∼3 L) was added, the
temperature increasing to 21 °C. DCM (4.0 L) was added to
aid stirring, and the slurry was recooled to 14 °C. Additional
HCl (2 N, 9.1 L) was added over 5 min. DCM (20.0 L) and
H2O (11.0 L) were added to the reaction mixture, and the
layers were separated. The aqueous layer was extracted with
additional DCM (8.0 L). The combined organic layers were
returned to the glass extraction vessel and washed with H2O
(4.0 L). This procedure was performed in two batches; the
organic extracts from both batches were combined for solvent
1
mixture by H NMR spectroscopy. Ethyl acetate (25 L) and
water (5 L) were added to the reaction mixture. The solution
was neutralized with 30% K2CO3 (17.5 L) to pH 7. The batch
was transferred to a 100-L glass extraction vessel via an inline
filter. The flask was rinsed twice with ethyl acetate (7.5 L, 5 L),
and the rinses were also transferred to the glass extraction
vessel. The aqueous layer was removed. The organic layer was
washed with water (5 L). The organic layer was determined to
contain 7.24 kg (95%) product by HPLC. Compounds 16, 17,
and 18 are all commerically available, with spectra matching
those provided in the literature11 and of known authentic
material.
1,2,4-Oxadiazole-3-carboxylic acid, 5-methyl-, potassium
salt (4).13 The isolated organic layer containing oxadiazole ester
18 (7.24 kg, 46.4 mol) was transferred to a 100-L RBF with
ethanol (20 L), and the solution was solvent switched to
ethanol. When the ratio of product: ethyl acetate was 8.6:1 by
1H NMR spectroscopy, distillation was stopped. The total
ethanol volume was adjusted to 46.8 L. The batch was heated
to 56 °C, and 10 N KOH (430 mL) was added via addition
funnel over 5 min. White crystals immediately began to form in
the reaction mixture. After the first addition was complete, 10
N KOH (3.87 L) was added over 10 min; the temperature
increased to 76 °C. Additional 10 N KOH (350 mL) was
added, and the slurry was left to age and cool to room
temperature overnight. Supernatant HPLC assay showed
approximately 206 g (2.5%) product and 823 g (10.7%) ethyl
ester. The batch was filtered and rinsed with ethanol, followed
by heptane. The product was dried under vacuum and nitrogen
for 2 d, affording 6.33 kg (77% over two steps) of oxadiazole K-
G
dx.doi.org/10.1021/op300128c | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Concept Article
switch to afford 6.78 kg of ketone (95.8% yield).16 The ketone
solution was transferred via an inline filter to a 100-L RBF. The
solution was concentrated under vacuum and flushed with
additional THF (∼40 L) to a low volume (16.75 L). The RBF
was placed under N2 and warmed to 44 °C. Heptane (36 L)
was added via addition funnel in 4-L portions, allowing the
slurry to cool during addition. Crystallization of off-white
needles occurred immediately. The slurry was stirred at room
temperature overnight, then filtered, washing with 5:4 heptane/
THF (18.0 L) followed by heptane. The solid was dried under
nitrogen and vacuum overnight. Assay of the dried, off-white
crystals showed 6.26 kg of product (88.4% yield, 100 LCAP).
The yield over two steps, from oxadiazole K-salt 4 to ketone 3,
was 86.6%. Spectroscopic data were consistent with that
previously reported in the literature.1,2
The enantiopurity was >99%. Spectroscopic data were
consistent with that previously reported in the literature.1,2
1H-Pyrido[3,4-b]indole, 3-[5-(4-fluorophenyl)-1H-imida-
zol-2-yl]-2,3,4,9-tetrahydro-1-(5-methyl-1,2,4-oxadiazol-3-
yl)-1-(1-methyl-1H-pyrazol-4-yl)-, (1R,3R)- (1).1,2 A 100-L
glass extraction vessel was charged with pyrazole−oxadiazole
ketone 3 (2.96 kg, 15.4 mol), tryptamine·2TsOH 2 (9.42 kg,
14.2 mol), and sodium acetate (1.16 kg, 14.2 mol). DMSO
(14.1 L) and tetraethyl orthosilicate (2.95 kg, 14.2 mol) were
added, and heating was applied to 73.5−75.5 °C. The reaction
mixture was homogeneous above 58 °C. The reaction
proceeded to 96% conversion after 40 h (dr = 58.7:41.3).
The reaction mixture was cooled to room temperature. 2-
Methyltetrahydrofuran (38 L) was added, and the mixture was
cooled to 18.5 °C. NaOH(aq) (5 N, 30 L) was added in 4 L
portions, the temperature rising to 32 °C. The mixture was
stirred for 3 h, until no white solid remained. On settling, three
layers formed: the bottom silicon layer was removed, and
NaOH(aq) (1N, 16 L) was added to the biphasic mixture with
additional stirring. The aqueous layer was separated. The
organic layer was washed with 2.5% brine (16 L) The organic
layer contained 3.63 kg 1 (51.8% assay yield by HPLC); it also
contained 2.56 kg of L-274 (36.5%) for a dr of 58.7:41.3 and a
combined yield of 6.19 kg (88.3%). The solution was recharged
into a 100-L RBF via inline filter. The solution was solvent-
switched into toluene. Aceteonitrile (3.55 L) was charged; the
amount equal to ∼6% of toluene volume. KF was 50 ppm. The
batch was heated to 60 °C, crystallization occurred, and the
batch was allowed to cool to room temperature overnight.
Supernatant assay showed a 5.5:1 ratio of 22/1. The batch was
filtered, washing with 3% acetonitrile/toluene (60 L), to afford
1 as a white crystalline solid. The wet cake was 99.6:0.4 1/22 by
LC. The solid was transferred to trays and placed in a vacuum
oven at 69 °C and ∼120 Torr for 3 w to remove toluene to 0.11
wt %. 1 was isolated as a white crystalline solid (2.97 kg, 6.01
mol, 42.4% yield). Spectroscopic data were consistent with that
previously reported in the literature.1,2 Residual metals were
<10 ppm. Chiral method: Chiralcel OD-H, 250 mm × 4.6 mm,
40 °C, 1 mL/min, 260 nm, 30 min run time, 20% (1:1 IPA/
MeOH) in heptane +0.1% TEA isocratic: rt (1): 7.61 min, rt
(enantiomer-1): 14.45 min. By HPLC assay, final product was
99.60 LCAP 1, 0.17 LCAP 22, 0.24 LCAP enantiomer-22,
enantiomer-1 was undetectable.
Epimerization of Filtrate and Crystallization of Second
Crop of 1. The filtrates and initial washes from 1 crystallization
batches 1 and 2 contained about 1 kg of 1 and about 5.4 kg of
the diastereomer for a total of about 6.4 kg (12.9 mol) in about
160 L of toluene/acetonitrile. This solution was concentrated
to a volume of 35−40 L. Acetonitrile (32 L) was added, and the
solution was transferred to a 100-L glass extraction vessel, and
trifluoroacetic acid (3.68 kg, 32.3 mol) was added, causing the
temperature to rise from 19 to 28 °C. The solution was heated
to 55 °C for 18 h, giving an approximate 48:52 ratio of isomers.
The solution was cooled to 17 °C and quenched with aqueous
2.5 N NaOH (26 L, mild exotherm caused a 5 °C temperature
rise). The aqueous layer was removed. The organic layer was
washed with water (26 L). The final organic layer was
drummed off and assayed to contain 6.05 kg of the combined
diastereomers (51.3 kg at 11.8 wt % by HPLC). The solution
was transferred through an inline filter into a 100-L round-
bottom flask and solvent switched into toluene. Acetonitrile
was used to rinse the drums forward and keep the product in
solution until water was azeotropically removed. Distillation
Carbamic acid, N-[(1R)-1-[5-(4-fluorophenyl)-1H-imida-
zol-2-yl]-2-(1H-indol-3-yl)ethyl]-, 1,1-dimethylethyl ester
(14).1,2 Boc-D-Trp-OH 11 (19.0 kg, 62.4 mol) was dissolved
in DMF (89.7 kg). Cesium carbonate (10.2 kg, 31.2 mol) was
added, followed by 2-chloro-4′-fluoroacetophenone (10.8 kg,
62.4 mol). The mixture was aged at 20 °C for 16 h. Water (95.0
kg) was charged such that the temperature was less than 25 °C.
Ethyl acetate (84.9 kg) was added, and the resulting phases
were separated. The aqueous phase was extracted with ethyl
acetate (84.9 kg). The combined organic layers were then
washed with a 5% aqueous LiCl solution (4.8 kg LiCl in 95.0 kg
of water) to aid in removal of DMF. Toluene (164.7 kg) was
added and the solution distilled under vacuum until
approximately 100 L remained in the vessel. Ammonium
acetate (24.1 kg, 312.2 mol) was added and the reaction heated
at 110 °C under Dean−Stark conditions for 2.5 h (>99%
conversion by HPLC). The reaction was cooled to 20 °C and
diluted with ethyl acetate (84.9 kg) and water (95.0 kg).
Product crystallised in the upper organic layer at this point. The
lower aqueous phase was cut away. NaOH (1 N aq.) solution
was added to the batch such that the temperature was less than
35 °C. The organics were washed with half-saturated brine
(95.0 kg) followed by water (10 kg). The reaction mixture was
distilled under vacuum until approximately 95 L remained in
the vessel. The mixture was cooled to 20 °C and aged for 30
min. The batch was sampled, and the mother liquors were
found to contain approximately 3% of the theoretical product
yield by HPLC assay. The slurry was filtered and the cake
washed with toluene (20 kg). The solid was pumped dry on the
filter for 12 h. A total of 23.9 kg of a white solid was isolated
(91.2 LCWP due to residual toluene); 21.8 kg of product were
isolated in 83% corrected yield over two steps. The
enantiopurity of the solid was 99.6%. Spectroscopic data were
consistent with that previously reported in the literature.1,2
1H-Indole-3-ethanamine, α-[5-(4-fluorophenyl)-1H-imida-
zol-2-yl]-, (α-R)- (2).1,2 The Boc-imidazole compound 14 (21.8
kg, 51.9 mol) was slurried in acetonitrile (85.7 kg). p-
Toluenesulfonic acid monohydrate (29.6 kg, 155.5 mol) was
added and the mixture heated to 60 °C. Bubbling, but no
foaming, was observed. Greater than 99% conversion to
product was achieved within 45 min by HPLC; the batch was
cooled to 20 °C. The slurry was aged for a further 30 min at
this temperature. The slurry was filtered and the cake washed
with acetonitrile (17.1 kg). The solid was dried under vacuum
at 45 °C under a sweep of nitrogen to constant mass. A total of
29.6 kg of a white solid was isolated in high purity (99.3
LCWP). This corresponds to 29.4 kg of product in 85% yield.
H
dx.doi.org/10.1021/op300128c | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Concept Article
(6) Methylation of 4-iodopyrazole using methyl iodide and several
carbonate bases afforded the desired product in quantitative yield.
(7) (a) Elguero, J. In Comprehensive Heterocyclic Chemistry; Potts, K.
T., Ed.; Pergamon Press: Oxford, 1984; Vol. 5, pp 167−303.
(b) Cheng, D.-P.; Chen, Z.-C.; Zheng, Q.-G. Synth. Commun. 2003,
33, 2671. (c) Felding, J.; Kristensen, J.; Bjerregaard, T.; Sander, L.;
Vedsø, P.; Bergtrup, M. J. Org. Chem. 1999, 64, 4196. (d) Tretyakov,
E. V.; Vasilevsky, S. F. Mendeleev Commun. 1995, 233. (e) Ohsawa, A.;
Kaihoh, T.; Itoh, T.; Okada, M.; Kawabata, C.; Yamaguchi, K.; Igeta,
H. Chem. Pharm. Bull. 1988, 36, 3838. (f) Giles, D.; Parnell, E. W.;
was completed at a volume of 60 L (KF = 19 ppm), and the
acetonitrile content was adjusted to 5.5 vol %. The mixture was
heated to 65 °C, with crystallization initiating during heatup at
around 50 °C. The suspension was allowed to cool to room
temperature. HPLC assay of the supernatant showed a 1:5 ratio
of 1 to its diastereomer in solution. The crystalline 1 was
collected by filtration, rinsing with 3% acetonitrile in toluene
(56 L). Solid was dried in a vacuum oven at 70 °C, giving 2.2 kg
(34%).
Renwick, J. D. J. Chem Soc. (C) 1966, 1179. (g) Huttel, R.; Schafer, O.;
Jochum, P. Liebigs. Ann. Chem. 1955, 593, 200.
XRPD Data of 1. Slurry analysis (green, Form I) and
evaluation of Form III (red) and mixture (blue) )Figure 1.
̈
̈
(8) These conditions have been shown useful in electrophilic
aromatic substitution reactions of electron-rich benzene derivatives:
Pavlinac, J.; Zupan, M.; Stavber, S. Synthesis 2006, 2603.
(9) Kim, M. M.; Ruck, R. T.; Zhao, D.; Huffman, M. A. Tetrahedron
Lett. 2008, 49, 4026.
(10) Tetraethyl orthosilicate is a known dehydrating agent: Litvic,
M.; Cepanec, I.; Vinkovic, V. Heterocycl. Commun. 2003, 9, 385.
(11) Crystallization from DCM/toluene afforded crystals of a 1:1
mixture of diastereomers.
(12) Hellmann, H.; Piechota, H.; Scwiersch, W. Chem. Ber. 1961, 94,
757.
(13) Moussebois, C.; Eloy, F. Helv. Chim. Acta 1964, 47, 838.
(14) Ruccia, M.; Vivona, N. Ann. Chim. 1968, 58, 484.
(15) The product was shown to sublime when left to dry on a funnel
under vacuum and nitrogen.
(16) Assay yields are based on 6.31 kg oxadiazole Weinreb amide
(98% yield) from the previous oxadiazole Weinreb amide step.
Figure 1.
ASSOCIATED CONTENT
* Supporting Information
Additional HPLC parameters. This material is available free of
charge via Internet at http://pubs.acs.org.
■
S
AUTHOR INFORMATION
Corresponding Author
*rebecca_ruck@merck.com
■
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
We thank Mohammad Al-Sayah for analytical support, Jenna
Terebitski and Sachin Lohani for crystal analysis, Richard Ball
for single-crystal x-ray structure analysis, and Khateeta Emerson
for safety evaluation.
REFERENCES
■
(1) Dobbelaar, P. H.; Du, W.; Guo, L.; Hagmann, W. K.; He, S.; Jian,
T.; Liu, J.; Nargund, R. P.; Pasternak, A.; Shah, S. K.; Truong, Q. T.;
Ye, Z.; Dellureficio, J.; Bakshi, R. WO/2009/011836 A1, 2009.
(2) He, S.; Ye, Z.; Truong, Q.; Shah, S.; Du, W.; Guo, L.; Dobbelaar,
P. H.; Lai, Z.; Liu, J.; Jian, T.; Qi, H.; Bakshi, R. K.; Hong, Q.;
Dellureficio, J.; Pasternak, A.; Feng, Z.; deJesus, R.; Yang, L.; Reibarkh,
M.; Bradley, S. A.; Holmes, M. A.; Ball, R. G.; Ruck, R. T.; Huffman,
M. A.; Wong, F.; Samuel, K.; Reddy, V. B.; Mitelman, S.; Tong, S. X.;
Chicchi, G. G.; Tsao, K.-L.; Trusca, D.; Wu, M.; Shao, Q.; Trujillo, M.
E.; Eiermann, G. J.; Li, C.; Zhang, B. B.; Howard, A. D.; Zhou, Y.-P.;
Nargund, R. P.; Hagmann, W. K. ACS Med. Chem. Lett. 2012, 3, 484.
(3) Kmetic,
̌
M.; Stanovnik, B. J. Heterocycl. Chem. 1995, 32, 1563.
(4) Borg, S.; Estenne-Bouhtou, G.; Luthnan, K.; Csoregh, I.;
̈
Hesselink, W.; Hacksell, U. J. Org. Chem. 1995, 60, 3112.
(5) Branco, P. S.; Prabhakar, S.; Lobo, A. M.; Williams, D. J.
Tetrahedron 1992, 48, 6335.
I
dx.doi.org/10.1021/op300128c | Org. Process Res. Dev. XXXX, XXX, XXX−XXX