Selection of an enantioselective process for the preparation of a CGRP receptor inhibitor

Article

pubs.acs.org/OPRD

Selection of an Enantioselective Process for the Preparation of a

CGRP Receptor Inhibitor

Reginald O. Cann,* Chung-Pin H. Chen,^†Qi Gao, Ronald L. Hanson, Daniel Hsieh, Jun Li, Dong Lin,

Rodney L. Parsons, Yadagiri Pendri,^§R. Brent Nielsen, William A. Nugent,^‡William L. Parker,

Sandra Quinlan, Nathan P. Reising,^⊥Brenda Remy, Justin Sausker, and Xuebao Wang^∥

Chemical Development, Bristol-Myers Squibb Company, One Squibb Drive, New Brunswick, New Jersey 08903, United States

ABSTRACT: (R)-N-(3-(7-Methyl-1H-indazol-5-yl)-1-(4-(1-methylpiperidin-4-yl)piperazine-1-yl)-1-oxopropan-2-yl)-4-(2-oxo-

1,2-dihydroquinolin-3-yl)piperidine-1-carboxamide (1) is a potent calcitonin gene-related peptide (CGRP) receptor antagonist.

We have developed a convergent, stereoselective, and economical synthesis of the hydrochloride salt of 1 and demonstrated the

synthesis on a multikilogram scale. Two diﬀerent routes to the chiral indazolyl amino ester subunit were developed utilizing

either a Rh-catalyzed asymmetric hydrogenation or a biocatalytic process to install the single chiral center. The advantages and

disadvantages of each of these process routes are discussed, as are challenges addressed in the assembly of the ﬁnal drug

substance.

available. An eﬃcient synthesis of the known³1-(1-methyl-

1. INTRODUCTION

piperidin-4-yl)piperazine 4 was developed but will not be

Long-term treatment of migraine headaches is a signiﬁcant

discussed here. Instead our focus will be on the development of

unmet medical need. The discovery that the neuropeptide

an enantioselective route to the α-aminoester 3 and its

calcitonin gene-related peptide (CGRP) is involved in the

elaboration to 1.

pathophysiology of migraine¹prompted intense research eﬀorts

Many synthetic approaches to chiral α-amino acids and esters

have been reported.⁴Among these, catalytic protocols using

enantioselective transition metal catalysis or enzymatic syn-

thesis are especially process friendly. During our development

eﬀorts we explored both of these types of catalytic routes for

the preparation of 3 and were then faced with the need to

choose between the two approaches. Since α-amino esters are

frequently encountered as pharmaceutical building blocks, we

anticipate that our experience may be useful to others.

to identify selective CGRP receptor antagonists. The medicinal

chemistry eﬀorts at Bristol-Myers Squibb²resulted in the

identiﬁcation of 1 as a potent and eﬀective competitive CGRP

antagonist for clinical evaluation. To support these clinical

trials, an eﬃcient large-scale synthesis of 1 was developed as

described in this paper.

As shown in the retrosynthetic analysis in Scheme 1, the

structure of 1 lends itself to straightforward assembly from

three fragments. Of these, piperazine 2 is commercially

2. RESULTS AND DISCUSSION

Scheme 1. Retrosynthetic analysis for 1

2.1. Synthesis of 3 Using Asymmetric Hydrogenation.

For the synthesis of amino ester 3, our Discovery Chemistry

colleagues employed the reaction sequence shown in Scheme

2.⁵Initially the indolazine ring was introduced earlier in the

sequence, but its presence was found to interfere with the

asymmetric hydrogenation. Consequently, installation of the

indolazine was deferred until after the hydrogenation step. The

general strategy of Scheme 2 was adopted for our initial

approach to 3. However, several signiﬁcant issues in the

Discovery Chemistry route needed to be addressed prior to

scale-up.

Preparation of Enamide 8. It was ﬁrst necessary to establish

a reliable synthesis of amidoacrylate 7. We noted that unless 7

was puriﬁed chromatographically, impurities in the crude

material interfered with its crystallization. Moreover, it became

evident that isolated 7 was prone to polymerization. Therefore,

we developed a through-process which provides a solution of 7

Received: October 30, 2012

Published: November 14, 2012

1953

dx.doi.org/10.1021/op3003097 | Org. Process Res. Dev. 2012, 16, 1953−1966

Organic Process Research & Development

Article

Scheme 2. The Discovery Chemistry synthesis of 3

Scheme 4. Process for the preparation of puriﬁed 8

was then free-based prior to use by treatment with aqueous

potassium carbonate. At an 18 kg scale, enamide 8 was obtained

in 52% overall yield and 95.2 wt %/wt purity.

Asymmetric Hydrogenation of Enamide 8. As noted

previously (Scheme 2) an Et-DuPHOS-Rh catalyzed⁸hydro-

genation had been used by the Discovery Chemistry group to

produce amino ester 9 in excellent yield and selectivity.

However, several issues weighed against further scale-up using

this catalyst. The hydrogenation was slow and limited catalyst

life necessitated a high catalyst loading (molar substrate-to-

catalyst ratio) of 200. In addition, the reaction rate varied

signiﬁcantly for diﬀerent batches of enamide 8.

(Scheme 3) that is taken into the subsequent Heck reaction

without isolation.

Scheme 3. Process for the preparation of amidoacrylate 7

One contributor to the slow reaction rate when using the

DuPHOS catalyst may be competitive binding of the aniline

nitrogen atom. In support of this hypothesis, addition of a

strong acid such as methanesulfonic acid (MSA) to protonate

the amino group led to a faster reaction. This is illustrated using

chromatographically puriﬁed enamide 8 in Figure 1 (compare

reaction proﬁles 2 and 3).

Commercially available serine methyl ester hydrochloride 12

was ﬁrst Cbz protected under Schotten Baumann conditions⁶

to produce 13. After aqueous washes, the MTBE solvent was

exchanged via vacuum distillation to THF, and the alcohol was

then activated with methanesulfonyl chloride (MsCl) and

underwent elimination in the presence of triethylamine at 0 °C.

Concentration of the organic phase after aqueous workup

provided a THF solution of 7.

We next focused on the preparation of iodoaniline 6

(Scheme 4). The iodination reagent used by the Discovery

Chemistry group, iodine monochloride, generated a highly

colored HCl salt of iodoaniline 6, and the color persisted into

the product of the Heck reaction. This issue was circumvented

by replacing ICl with commercially available benzyltrimethyl-

ammonium dichloroiodate⁷(BTMACl₂I) which aﬀorded 6 as

an oﬀ-white hydrochloride salt. For the kilogram-scale

preparation, a solution of 5 in methanol was added to a slurry

of BTMACl₂I and potassium carbonate in dichloromethane to

provide addition control of the reaction exotherm. After

ﬁltration, an aqueous wash, and azeotropic drying, addition of a

solution of HCl in 2-propanol allowed isolation of 6 as its HCl

salt.

Figure 1. Reactivity of Et-FerroTANE-Rh catalyst vs Et-DuPHOS-Rh

catalyst.

Also shown in Figure 1 (reaction proﬁle 1), an even greater

improvement in reaction rate was achieved by replacing the Et-

DuPHOS ligand in the catalyst with the Et-FerroTANE ligand.⁹

The structures of the two ligands are shown in Scheme 5. Using

the Et-FerroTANE catalyst, the reaction time for hydro-

genation of chromatographically puriﬁed 8 decreased to 1 h.

The hydrochloride salt of 6 smoothly underwent palladium-

(II) acetate-catalyzed Heck coupling with the aminoacrylate 7

in aqueous THF to provide crude enamide 8 (Scheme 4). The

product could be puriﬁed by crystallization as a MSA salt and

1954

dx.doi.org/10.1021/op3003097 | Org. Process Res. Dev. 2012, 16, 1953−1966

Organic Process Research & Development

Article

Scheme 5. Comparison of ligand structures

Scheme 7. Preparation of the aminoester 3 as its bis-HCl salt

This higher activity was obtained despite a lower catalyst

loading of the FerroTANE vs the DuPHOS catalyst.

The ﬁnal step of the sequence is deprotection of the Cbz

group by palladium on carbon hydrogenolysis. This appears

conceptually simple but was complicated by the tendency of the

ester to undergo hydrolysis under these conditions, presumably

as a result of water in the carbon or methanol. The extent of the

hydrolysis was greater when ﬁltration of the mixture through a

Celite bed for removal of the palladium catalyst was prolonged.

The problem was alleviated by diluting the mixture so that the

hydrolysis product, amino acid 14, precipitated and was

simultaneously removed during ﬁltration. In this way the

amount of 14 in the isolated hydrogenation product was

diminished to <0.2%.¹¹

Even with this alternative catalyst, variability in reaction rate

and catalyst life between diﬀerent batches of enamide 8

remained an issue. We suspected that this was due to impurities

deactivating the catalyst. In particular, it is known that residual

chloride ion can have this eﬀect.¹⁰Therefore, an additional

puriﬁcation step was introduced: Hot ﬁltration of a solution of

8 in ethyl acetate followed by crystallization aﬀorded puriﬁed

enamide 8 in 86% yield.

Using this puriﬁed material the asymmetric hydrogenation of

8 using Et-FerrroTANE-Rh at a 0.25% catalyst loading

performed consistently (Scheme 6). At the conclusion of the

Another signiﬁcant issue with the scale-up of the hydro-

genolysis step was inconsistency of the yield which varied from

53 to 89%. The problem was traced to carbon dioxide, a

byproduct generated during the reaction. The carbon dioxide

formed an amine carbonate salt precipitate with the amino ester

3 which was inadvertently removed during the ﬁltration of the

heterogeneous catalyst. (The precipitate was not fully

characterized, and alternative formulation as a carbamic acid

is possible.) Purging the reaction mixture at the end of the

hydrogenation with nitrogen removed the carbon dioxide, and

the carbonate salt disproportionated to regenerate amino ester

3. This nitrogen purging solved the inconsistent yield issue and

was incorporated into the process.

The freebase amino ester 3 was found to be an unstable, low-

melting solid; we therefore chose to isolate it as the bis-

hydrochloride salt, which had superior solid handling proper-

ties. To accomplish this, the ﬁltered solution of 3 in methanol

and ethyl acetate from the hydrogenolysis step was treated with

a solution of HCl in 2-propanol to aﬀord 3 bis-HCl, which was

collected by ﬁltration and vacuum dried. Two 1.8 kg batches of

3 bis-HCl were prepared following this protocol (83−86%

yield, 98 wt % purity, >99.9% ee.

2.2. Biocatalytic Route to Amino Ester 3. Our enzymatic

approach to 3 is summarized in Scheme 8. Since the indazole

functionality is expected to be compatible with the biocatalytic

transformation, the heterocycle can be introduced at the

beginning of the synthesis. The stereogenic center is established

by direct conversion of the indazolyl keto acid 15 to the

indazolyl amino acid 14 promoted by a ^D-transaminase.

Preparation of Keto Acid Substrate 15. As shown in

Scheme 9, commercially available aniline 16 was treated with

tert-butyl nitrite to generate the indazole ring.

Scheme 6. Process for the preparation of the MSA salt of 9

reaction the solution was subjected to polish ﬁltration and

solvent exchanged into ethyl acetate prior to precipitation of 9

as its MSA salt. A total of 14 kg of the MSA salt of 9 was

produced in 96% overall yield and ∼99% ee.

Preparation of Amino Ester 3. The indazole ring closure

was based on the nitrosation route previously shown in Scheme

2. However, we noticed that initial batches of indazole 10

prepared in the laboratory had a dark brown coloration that

would carry into the subsequent intermediate, amino ester 3.

The extent of this color formation correlated with the presence

of trace levels of rhodium and palladium in the starting material

9 indicating the need to remove these contaminants prior to

the nitrosation step. This was accomplished by ﬁltering the

ethyl acetate solution obtained upon free-basing through a

Cuno carbon zeta pad. After azeotropic drying, this solution

was used directly in the nitrosation.

Addition of isoamyl nitrite to the resulting solution after

acidiﬁcation with acetic acid smoothly transformed the aniline 9

to indazole 10 (Scheme 7). After neutralization of the acetic

acid with aqueous potassium carbonate solution, 10 was

isolated by solvent exchange into toluene and addition of

heptane.

For the subsequent halogen−metal exchange, in order to

avoid formation of the desbromo compound 21, it was

1955

dx.doi.org/10.1021/op3003097 | Org. Process Res. Dev. 2012, 16, 1953−1966

Organic Process Research & Development

Article

prior to addition of water to crystallize N-benzoyl enamide 20.

The ﬁnal hydrolysis to aﬀord the enzyme substrate 15

proceeded readily upon treatment with 6 M hydrochloric acid

at 95 °C. However, we initially found that removal of the

benzoic acid side product was challenging.

Scheme 8. Chemoenzymatic synthesis of aminoester 3

To purge benzoic acid from keto acid 15 we took advantage

of the relative solubility diﬀerence between the acids. Keto acid

15 has limited solubility in hot water or hot dilute acid

compared with benzoic acid; thus, ﬁltering the reaction mixture

at >95 °C allowed the selective isolation of 15. Under these

conditions, 15 crystallized as its enol tautomer and contained

<3% of residual benzoic acid. At this level residual benzoic acid

did not interfere with the subsequent biocatalytic step.

Enzymatic Synthesis of Amino Acid 14. Initial eﬀorts to

convert keto acid 15 to amino acid 14 were carried out using a

commercial ^D-transaminase from Biocatalytics, Inc. (now

Codexis) and ^DL-alanine as the nitrogen source. Using this

enzyme, the reaction tended to stall at ∼90% conversion

apparently due to inhibition by the side product pyruvic acid.

We initiated a research eﬀort to identify an improved enzyme.

While these eﬀorts were underway we elected to scale up the

process using the commercial enzyme.

Scheme 9. Preparation of enzyme substrate 15

Using this enzyme, the indazolyl acid 15 underwent

transamination with ^DL-alanine serving as an amino donor at

pH 7.5. At the end of the conversion, the reaction mixture was

acidiﬁed with aqueous sulfuric acid to dissolve the reaction

products while simultaneously precipitating the denatured

enzyme. Following ﬁltration of the acidic reaction mixture

through Amberlite XAD16 resin, the ﬁltrate was heated to

∼100 °C and was neutralized with sodium hydroxide to eﬀect

the crystallization of indazolyl amino acid 14 as a monohydrate.

The enzymatic process was demonstrated on a 600-g scale in

two batches to produce indazolyl amino acid 14 in 79% yield

and >99% ee. Process details for this step were published

previously.¹³

We subsequently discovered a D-transaminase from a soil

organism identiﬁed as Bacillus thuringiensis which exhibited

signiﬁcantly diminished product inhibition. The puriﬁed

enzyme was cloned and expressed in E. coli as described

elsewhere.¹³In laboratory scale experiments, this enzyme was

shown to oﬀer additional advantages. In particular, the reaction

time of 72 h for the commercial enzyme was decreased to 21 h

using the BMS cloned enzyme. Moreover, the volume eﬃciency

was improved from 33 L/kg to 17 L/kg.

advantageous to ﬁrst deprotonate the acidic NH proton of

indazole 17 with 1 equiv of n-butyllithium prior to treatment

with sec-butyllithium. After reaction of the resulting aryllithium

with DMF, aldehyde 18 was readily separated from any

unreacted bromoindazole 17 or indazole 21 by formation of the

bisulﬁte adduct 22 and was extracted into the aqueous phase.

Aldehyde 18 was then regenerated by treatment with aqueous

sulfuric acid and was crystallized from a mixture of ethyl acetate

and heptane.

Preparation of 3 via Fischer Esteriﬁcation. Since we again

wished to isolate the bis-hydrochloride of ester 3, hydrogen

chloride was employed as the esteriﬁcation catalyst. Anhydrous

HCl was conveniently generated by addition of acetyl chloride

to methanol. Two issues needed to be addressed in this

otherwise straightforward transformation.

First, attempts to directly esterify amino acid 14 as its

monohydrate stalled at ∼95% conversion. Apparently the

system reaches an equilibrium conversion of 14 and 3 as a

result of the extra equivalent of water present in the

monohydrate. Consistent with this explanation, when the

monohydrate was replaced with the anhydrous bis-hydro-

chloride of amino acid 14, the esteriﬁcation proceeded to >99%

completion.

The aldehyde 18 was next subjected to the Erlenmeyer

synthesis¹²with hippuric acid in the presence of acetic

anhydride and potassium acetate to aﬀord 2-phenylazlactone

19. When the reaction was complete, excess acetic anhydride

was decomposed by charging water to the mixture. The 2-

phenylazlactone 19, which crystallized out of the reaction

mixture, was found to be contaminated with the N-acetylated

azlactone. To address this impurity we simply aged the

crystallization slurry for at least 2 h to hydrolyze any N-

acetylated azlactone prior to product ﬁltration.

The remaining issue was formation of an impurity derived

from N-methylation of the indazole ring. Impurity 23 was not

purged when the 3 bis-HCl salt was isolated from solvent

methanol. When present, 23 gave rise to additional down-

Ring-opening of 2-phenylazlactone 19 was eﬀected in

methanol with a catalytic amount of sodium methoxide.

Unreacted sodium methoxide was neutralized with acetic acid

1956

dx.doi.org/10.1021/op3003097 | Org. Process Res. Dev. 2012, 16, 1953−1966

Organic Process Research & Development

Article

stream impurities which were diﬃcult to remove and could

potentially contaminate the API.

laboratory results are highly predictive for subsequent scale-

up. To be conservative, we have retained the yield data for the

large-scale run but expect that this would also be improved

using our cloned enzyme. It can be seen from Table 1 that each

technology exhibited particular advantages in this application.

The rhodium-catalyzed asymmetric hydrogenation provided

a fast, clean reaction with excellent volumetric eﬃciency. The

overall synthetic sequence required fewer process steps than for

the enzymatic process. However, these advantages came at a

cost. The Et-FerroTANE-Rh catalyst is proprietary and thus of

limited availability.¹⁴In addition, substrate purity was critical to

achieve a high turnover number (TON), necessitating a labor-

intensive multistage puriﬁcation procedure. In addition to these

considerations, there was concern regarding the removal of Pd

and Rh impurities from the intermediates and API. As a result,

the enzymatic route was selected for future scale-up of 3.

We also made an eﬀort to compare the “greenness” of these

two approaches but this was not as straightforward as might be

imagined. The hydrogenation of 8 is an extremely clean

reaction which produces no side-products whatsoever.

However, the asymmetric hydrogenation here employs a

soluble catalyst containing the scarce and precious metal

rhodium. In contrast, the enzymatic reaction generates alanine-

derived side-products which require disposal. The asymmetric

hydrogenation involves the use of organic solvents (methanol

and ethyl acetate) while the enzymatic process is carried out in

water. However, this apparent advantage is counterbalanced by

solvent disposal for the extra steps¹⁵required for substrate

preparation in the enzymatic route.

To address this issue, screening studies were undertaken to

identify reaction conditions (equivalents of acetyl chloride,

concentration, and temperature) where formation of 23 was

minimized. The optimized conditions resulted in a reaction

proﬁle with <0.5% of 23. It was further discovered that 1:1

MeOH/MTBE is an eﬀective crystallization solvent which

provides the bis-HCl salt of 3 with <0.2% of 23. These

elements were combined as shown in Scheme 10. This process

was used to produce 2.2 kg of 3 (bis-HCl salt) in 98.6% purity

with only 0.11% of 23 in the product.

Scheme 10. Conversion of 14 to the bis-HCl salt of 3 by

Fischer esteriﬁcation

2.4. The Synthetic Endgame. Assembly of API 1 from

amino ester 3 involved two coupling steps as shown in Scheme

11. In particular, the initial CDI coupling proved sensitive to

2.3. Comparing the Two Routes. By this point it was

evident that either of the two routes could provide kilogram

quantities of amino ester 3. To choose one route for further

scale-up, we compared the two approaches across a number of

criteria and process parameters.

Scheme 11. Endgame strategy for manufacture of API 1

Regarding Table 1, we should note that the reaction time and

volume eﬃciency data for the BMS cloned enzyme have been

used. These results were obtained in laboratory scale (6 g)

conditions. However, our experience has been that such

Table 1. Comparison of asymmetric hydrogenation and

biocatalytic routes

factor/catalyst

Et-FerroTANE-Rh

96

D-transaminase

yield of enantioselective step

(%)

79

a

b

linear steps to 3 (bis-HCl salt)

6

8

a

b

overall yield to 3 (bis-HCl salt) 25

34

(%)

enantiomeric excess (%)

product purity (%)

>99.8

99.6

1

>99.7

the reaction conditions. This reﬂects the fact that 3 is an

ambident nucleophile: both the NH₂group of the amino ester

moiety and the NH of the indazole ring are reactive so that

chemoselectivity is an important consideration.

Preparation of Penultimate 11 by CDI Coupling. As noted

in Scheme 11, the CDI coupling and subsequent ester

hydrolysis were telescoped for process eﬃciency. An inves-

tigation into this synthetic sequence showed that a variety of

impurities were present in the isolated 11 depending on the

reaction conditions and isolation protocol.

97

reaction time (h)

21

volumetric eﬃciency (L/kg)

catalyst availability

8

17

limited

high

readily available

less critical

required substrate purity

potential inhibitors

chloride, carbonate, pyruvate (side

etc.

product)

oxygen sensitivity

highly oxygen

sensitive

not oxygen sensitive

a

b

When the coupling was run at ambient temperature, up to

2% of 26 was formed (Scheme 12) and did not purge during

Based on aniline 5 as starting material. Based on bromide 16 as

starting material.

1957

dx.doi.org/10.1021/op3003097 | Org. Process Res. Dev. 2012, 16, 1953−1966

Organic Process Research & Development

Article

followed by coupling with two molecules of piperidinyl

Scheme 12. Formation of process impurity 26

quinolinone 4 (Scheme 14).

Scheme 14. Process impurity 30 formed by overcharging

CDI

Process impurities 32 and 33 were formed when amino acid

14 was present in 3 (bis-HCl salt). These presumably arise by

activation of the carboxylic acid moiety by CDI followed by

undesired peptide coupling with either 4 or an additional

molecule of 3 (Scheme 15). These impurities were controlled

to <0.2% by setting the amino acid 14 speciﬁcation for the

starting material 13 to <0.5%. At this level any resulting

downstream impurities were purged during isolation of the API.

Given the large impact of small deviations from precise

stoichiometry, a tactical decision was made to use a slight excess

(1.2 equiv) of CDI. This approach left 30 as the sole process

impurity, which we were able to control by taking advantage of

the selectivity observed when the various components of the

reaction mixture react with water. After generation of the

activated intermediate 24 water selectively reacts with the

excess CDI as well as the doubly activated intermediate 29

which leads to 30. In contrast, the activated intermediate 24

was stable for up to an hour in the presence of water (5 equiv)

at −10 °C.¹⁶The modiﬁed process incorporating these

conditions consistently aﬀorded penultimate 11 containing

<0.2% of impurity 30.

isolation of 11. Laboratory studies revealed that the indazolyl

urea linkage is unstable at higher temperatures. Consequently,

impurity 26 was controlled by simply running the coupling at

40 °C.

Stoichiometry was also critical. Undercharging of CDI led to

some unreacted amino ester 3, which then coupled with

activated amino ester 24 to generate the undesired symmetric

urea impurity 27 and the hydantoin 28 (Scheme 13).

Conversely, overcharging CDI aﬀorded process impurity 30,

which resulted from activation of the indazolyl nitrogen atom

Scheme 13. Impurities generated from undercharging CDI

The amorphous nature of compound 11 made it susceptible

to entrapment of salts during isolation, thus forming gummy

solids. This could be avoided by slow addition of the reaction

mixture to an aqueous hydrochloric acid quench with eﬃcient

agitation. This reverse addition protocol generated well

dispersed solids with acceptable ﬁltration properties. After

ﬁltration, the cake was washed with copious amounts of water

to remove salts and dimethylamine (formed from the cleavage

of DMF by the lithium hydroxide).

A ﬁnal issue was the decomposition of 11 during drying. The

penultimate 11 degraded by urea bond cleavage to form 2−3%

of amino acid 14 and piperidinyl quinolinone 4 during drying at

50 °C. Laboratory experimentation revealed that any potential

1958

dx.doi.org/10.1021/op3003097 | Org. Process Res. Dev. 2012, 16, 1953−1966

Organic Process Research & Development

Article

Scheme 15. Impurities generated from amino acid 14 in 3

(bis-HCl salt)

Scheme 16. Zwitterionic form of 11

Fortunately, automated high-throughput screening studies

identiﬁed the hydrochloride salt as an acceptable form.

Preparation of 1 requires amidation of the carboxylic acid 11

with piperazine 2 as previously shown in Scheme 11. Screening

several commercial coupling reagents identiﬁed EDAC/HOBT

as a promising system.

The principal challenge for this step was achieving a high

volume eﬃciency. Here we beneﬁtted from solvent screening

studies, which revealed that 1 (HCl salt) has a very low

solubility in a mixture of dimethyl acetamide (DMAc) and

acetone. We envisioned the use of DMAc as reaction solvent so

that subsequent addition of acetone and hydrochloric acid

would allow the crystallization of the product when the

amidation was complete. We were pleasantly surprised to

observe that adding acetone to the reaction mixture and

seeding with 1 (HCl salt) crystals caused the desired API

crystals to form without the need for any extra hydrochloric

acid. Evidently the free base of 1 generated from the reaction

equilibrates with DIPEA-HCl to form the HCl salt of 1 and

DIPEA free base (Scheme 17).¹⁷This direct precipitation of

the HCl salt of 1 resulted in a satisfactory volume eﬃciency of

∼25 L/kg.

residual levels of HCl, LiOH, dimethylamine or LiCl were not

the root cause of this degradation. This led us to speculate that

the degradation may be facilitated by the carboxylic acid moiety

1

of 11. This possibility is supported by H NMR studies, which

showed that the indazole N−H resonance at 13.04 ppm in 11

was a rather broad peak of two protons between 12.4 and 13.4

ppm. This implies the potential existence of 11 as a zwitterion

(Scheme 16), and the carboxylate anion may serve as the

nucleophile to attack the urea carbonyl group to break the urea

bond.

Scheme 17. In situ formation of 1 (HCl salt)

Decreasing the oven temperature to 40 °C limited

decomposition to <0.1% over 63 h of drying while Dynochem

modeling revealed that the rate of drying was controlled by

moisture transport from inside the wet cake. To minimize

drying time we utilized an agitated ﬁlter-dryer in a two stage

protocol. The wet cake was ﬁrst deliquored for at least 12 h at

ambient temperature with minimum agitation, thus avoiding

the formation of gummy balls. This was followed by warming

to 40 °C with intermittent agitation. The telescoped process

was then utilized to produce 1.7 kg of penultimate 11 in 80%

yield and a purity of 95.6% and 99.3% ee.

Final Coupling to Produce 1. Initial polymorph screening

revealed that 1 can crystallize in several neat and solvated

forms, none of which were appropriate for development.

1959

dx.doi.org/10.1021/op3003097 | Org. Process Res. Dev. 2012, 16, 1953−1966

Organic Process Research & Development

Article

Solvent screening also revealed that, in contrast to

enantiomerically pure 1 (HCl salt), the racemic salt is virtually

insoluble in ethanol. Thus ﬁltration of an ethanol solution of

the product reduced the levels of the undesired enantiomer of 1

below the limits of detection. Simple addition of heptane to

recrystallize the product routinely provided 1 (HCl salt) of

>99.9% ee.

This amidation/recrystallization sequence was scaled up to

produce 1.8 kg of 1 (HCl salt) in 89% yield with a purity of

99.9% and >99.9% ee. The overall route was deemed suitable

for future pilot plant campaigns to produce the HCl salt of 1.

reactor, and the resulting slurry was cooled to 5 °C. A solution

of 2,6-dimethylaniline 5 (23.5 kg, 194.2 mol) in methanol (177

L) was charged over 1 h at <15 °C. The charging line was

rinsed with methanol (59 L) into the reactor. The reaction

mixture was warmed to 20 °C over 30 min and was stirred for

an additional 30 min to complete consumption of 5. The

reaction mixture was ﬁltered through a 20 μm polypropylene

cloth to remove the inorganic solids. The reactor and ﬁlter cake

were rinsed with dichloromethane (71 L), and the combined

ﬁltrate was distilled under reduced pressure at 20 °C to a

volume of ∼235 L. Ethyl acetate (235 L) was charged, and the

distillation was continued under reduced pressure to an end

point of <5% v/v of methanol. Ethyl acetate (235 L) was

charged to adjust the volume to ∼470 L. A solution of sodium

chloride (12 kg) in water (202 L) was charged to the solution,

and the mixture was stirred for 20 min. After phase separation,

the aqueous layer (bottom) was discarded. The organic

solution was washed with a solution of sodium chloride (7

kg) in water (134 L), the phases were separated, and the

solution was distilled under reduced pressure at 20 °C to a

volume of ∼235 L. Ethyl acetate (235 L) was charged, and the

distillation was continued under reduced pressure at 20 °C to

an end point of <2% v/v of water. Ethyl acetate (282 L) was

charged to the solution to adjust the volume to ∼518 L, and the

solution was polish ﬁltered through a 10 μm in-line ﬁlter. The

ﬁltrate was adjusted with ethyl acetate to a ﬁnal volume of

∼588 L. The solution was cooled to 15 °C, and a solution of

HCl in IPA (5 M, 46.0 kg, 252 mol) was charged over 30 min,

while maintaining the temperature at <25 °C. The resulting

slurry was stirred at 20 °C for 2 h and ﬁltered. The cake was

washed with ethyl acetate (2 × 117 L) and dried by vacuum

suction on the ﬁlter for 1 h. The product cake was further dried

under vacuum (35 °C, 25 mmHg) to aﬀord 4-iodo-2,6-

dimethylaniline hydrochloride (6) (47.8 kg, 81% yield, 96.0

area %, 93.3 wt.%) as an oﬀ-white crystalline solid; mp 168.4

°C; ¹H NMR (500 MHz DMSO-d₆) δ 9.67 (br, 3 H), 7.74 (s, 2

H), 2.33 (s, 6 H); ¹³C NMR (100 MHz DMSO-d₆) δ 136.7,

133.2, 130.4, 91.9, 17.3; IR (KBr pellet) ν_max3435, 2924, 2704,

2650, 2509, 2472, 2218, 1981, 1732, 1615, 1586, 1567, 1518,

1474, 1463, 1454, 1402, 1386, 1290, 1264, 1176, 1105, 1093,

1037, 1027, 915, 877, 858, 838, 704; Anal. Calcd for

C₈H₁₁ClIN: C, 33.89; H, 3.91; Cl, 12.50; I, 44.76; N, 4.94.

Found: C, 34.19; H, 3.84; Cl, 12.93; I, 43.21; N, 4.84.

Methyl 2-(Benzyloxycarbonylamino)acrylate (7). So-

dium bicarbonate (44.4 kg, 527.2 mol) and water (410 L) were

charged to a reactor, and the mixture was agitated for 15 min.

The resulting solution was treated with a solution of serine

methyl ester hydrochloride, 12 (82.0 kg, 159.8 mol), in water

(328 L) over 1 h at such a rate as to control oﬀ-gassing. The

solution was stirred at 20 °C for 15 min, and benzyl

chloroformate (27.3 kg, 159.8 mol) was charged over 30 min.

The charging line was rinsed with MTBE (214 L) into the

reactor. The biphasic solution was stirred at 20 °C over 5 h to

complete consumption of benzyl chloroformate. MTBE (528

L) was charged to the reactor, and the mixture was stirred for

an additional 15 min. After the phases were separated, the

aqueous layer (bottom) was discarded. A solution of sodium

chloride (16 kg) in water (164 L) was added, and the mixture

was stirred for 15 min. After the phases were separated, the

aqueous layer (bottom) was discarded, and the solution was

concentrated under reduced pressure at <25 °C to a volume of

250 L. THF (643 L) was charged, and the distillation was

continued under reduced pressure at 20 °C to an end point of

3. CONCLUSION

We have described development of a robust and practical route

to the CGRP antagonist 1 (HCl salt) based on a convergent

synthesis, which utilizes readily available starting materials,

makes use of diﬀering physical properties to facilitate isolations,

and is applicable to further scale up. As a part of this eﬀort, two

processes were developed to prepare multikilogram quantities

of amino ester 3 using either asymmetric transition metal

catalysis or biocatalysis, each route oﬀering diﬀerent advantages

from a process chemistry perspective. The biocatalytic route

was selected for further development based on three

considerations: (1) Exhaustive puriﬁcation of enamide substrate

8 was required to achieve a high turnover number (TON) in

the asymmetric hydrogenation. In contrast, the biocatalytic

route is more tolerant of impurities down to 95% starting

material purity. (2) The transition metal-catalyzed route

required use of a proprietary catalyst available from a single

vendor. In contrast, ^D-transaminase enzymes are widely

available for the biocatalytic route. Furthermore a highly active

D-transaminase enzyme was developed and expressed in

recombinant E. coli and was produced eﬃciently by

fermentation at BMS. (3) The biocatalyic route eliminated

concerns about the removal of Pd and Rh residues from the

API.

Three aspects of the endgame are noteworthy. The ﬁrst is the

use of water to quench excess CDI while not decomposing the

activated intermediate 24. Our approach should have general

utility for reactions involving CDI and ambident nucleophiles.

The other two points underscore the importance of

determining the physical properties of reaction intermediates

when designing synthetic protocols. In order to maximize

process throughput, the low solubility of the ﬁnal product 1

(HCl salt) in a mixture of DMAc and acetone was exploited by

adding acetone to the reaction mixture to crystallize the

product directly. In addition, the low solubility of the HCl salt

of rac-1 in ethanol was utilized to further upgrade the

enantiomeric purity. The optimized process operations were

shown to be robust and allowed production of the active

pharmaceutical ingredient 1 (HCl salt) in multikilogram

quantities.

4. EXPERIMENTAL PROCEDURES

The experiments outlined below provide representative

procedures for what was run on scale in the kilo/pilot-plant

facilities. NMR chemical shifts (δ) are given in ppm and

infrared data (ν_max) are given in cm⁻¹.

4-Iodo-2,6-dimethylaniline Hydrochloride (6). Benzyl-

trimethylammonium dicholoroiodate (BTMACl₂I, 68.3 kg,

196.1 mol), potassium carbonate (30.1 kg, 217.5 mol), and

dichloromethane (604 L) were sequentially charged to a

1960

dx.doi.org/10.1021/op3003097 | Org. Process Res. Dev. 2012, 16, 1953−1966

Organic Process Research & Development

Article

<750 ppm of water. THF (370 L) was charged to adjust the

volume to ∼620 L. The solution of the alcohol 13 was cooled

to 0 °C, and methanesulfonyl chloride (22.0 kg, 191.7 mol) was

charged over 30 min. The mixture was stirred for 30 min, and

the triethylamine (50.1 kg, 495 mol) was charged while

maintaining the internal temperature <25 °C. The reaction

mixture was warmed to 25 °C and was stirred for 2 h to

complete consumption of the intermediate mesylate. A solution

of ammonium chloride (41.0 kg) in water (300 L) was charged,

and the mixture was stirred vigorously for 15 min. After the

phases were separated, the aqueous layer (bottom) was

discarded, and the solution was concentrated under reduced

pressure at <25 °C to a volume of ∼90 L. The concentration of

the solution was adjusted with THF (26 L) to produce a THF

solution of methyl 2-(benzyloxycarbonylamino)acrylate, 7 (137

kg of solution, 24.8 wt % of 7, 90.4% solution yield).

ﬁltered. The cake was washed with dichloromethane (57 L) and

dried by vacuum suction on the ﬁlter for ∼3 h. The partially

dried MSA salt was charged back into the reactor followed by

methanol (201 L) and water (23 L). The mixture was stirred at

20 °C for 15 min, and a solution of potassium carbonate (11.2

kg, 81.1 mol) in water (135 L) was charged while maintaining

the temperature at ∼20 °C. The slurry was stirred at 20 °C for

1 h and ﬁltered. The cake was washed with a solution of

methanol (36 L) in water (33 L) and was dried by vacuum

suction of the ﬁlter for 1 h. The product cake was further dried

under vacuum (50 °C, 25 mmHg) to aﬀord methyl 3-(4-amino-

3,5-dimethylphenyl)-2-(benzyloxycarbonylamino)acrylate, 8,

(18.3 kg, 79.6% yield, 97.6 area %, 95.2 wt %) as a yellow,

crystalline solid..

Puriﬁcation by Hot Filtration. Methyl 3-(4-amino-3,5-

dimethylphenyl)-2-(benzyloxycarbonylamino)acrylate, 8 (17.1

kg, 48.3 mol), and ethyl acetate (282 L) were charged to a

reactor. The mixture was heated to ∼70 °C and stirred for 15

min to generate a solution. The hot solution was ﬁltered

through a preheated (65 °C) plate ﬁlter containing a 2 in. bed

of Celite (2 kg). The reactor and plate ﬁlter were washed with

preheated ethyl acetate (34 L), and the combined ﬁltrate was

concentrated under reduced pressure at 55 °C to a ﬁnal volume

of ∼68 L. The slurry was warmed to 60 °C and stirred for 20

min. The slurry was cooled to 20 °C over 3 h and ﬁltered. The

product cake was washed with ethyl acetate (34 L) and dried by

vacuum suction of the ﬁlter for 1 h. The product cake was

further dried under vacuum (50 °C, 25 mmHg) to aﬀord

puriﬁed methyl 3-(4-amino-3,5-dimethylphenyl)-2-

(benzyloxycarbonylamino)acrylate, 8 (14.7 kg, 86% yield, 98

area %, 97 wt %), as a bright-yellow crystalline solid; mp 143.6

An aliquot was concentrated in vacuo for characterization. ¹H

NMR (400 MHz CDCl₃) δ 7.41−7.35 (m, 5 H), 7.30 (bs, 1

H), 5.82 (d, J = 1.52 Hz, 1 H), 5.19 (s, 2 H), 3.85 (s, 3 H); ¹³

C

NMR (400 MHz CDCl₃) δ 164.1, 153.1, 135.8, 130.9, 128.5,

128.2, 106.0, 67.0, 52.9.

Methyl 3-(4-Amino-3,5-dimethylphenyl)-2-

(benzyloxycarbonylamino)acrylate (8). Heck Reaction to

Produce Crude 8. 4-Iodo-2,6-dimethylaniline (6) (32.9 kg,

108.3 mol), tetra-n-butyl ammonium chloride (TBAC, 15.0 kg,

54.1 mol), potassium carbonate (K₂CO₃, 52.4 kg, 378.9 mol),

and water (100 L) were sequentially charged to the reactor.

The solution of 7 in THF (128.5 kg, 31.83 kg of 7, 135.3 mol)

was charged, and the mixture was stirred for 20 min. The

palladium(II) acetate (1.21 g, 5.4 mmol) was charged, and the

reaction mixture was heated to 70 °C. The reaction mixture was

stirred at 65−70 °C over 5 h until complete consumption of 6.

The reaction mixture was cooled to <40 °C and was mixed with

water (33 L) and THF (165 L). The mixture was stirred for 15

min and ﬁltered through a 2 in. Celite bed (2 kg) followed by

polish ﬁltration though a 10 μm Cuno cartridge in-line ﬁlter to

remove the insoluble palladium residue. The reactor, Celite

cake, and cartridge were rinsed with THF (37 L), and the

combined ﬁltrates were allowed to settle. After the phases were

separated, the aqueous layer (bottom) was discarded, and the

solution was washed with a solution of sodium chloride (3 kg)

in water (33 L). The solution was distilled under reduced

pressure at 35 °C to a volume of ∼165 L. Methanol (334 L)

was charged to the solution, and the distillation continued

under reduced pressure at 35 °C to an end point of <2% v/v of

THF. Methanol (215 L) was charged to adjust the volume to

∼380 L. The resulting slurry was cooled to 25 °C, and water

(165 L) was charged while maintaining the internal temper-

ature between 20 and 25 °C. The slurry was stirred at 20 °C for

1 h and was ﬁltered. The cake was washed with a solution of

methanol (66 L) in water (66 L) and was dried by vacuum

suction of the ﬁlter for 1 h. The product cake was further dried

under vacuum (50 °C, 25 mmHg) to aﬀord the crude 8 (25.2

kg, 65.7% yield, 94.1 area %) as a dark-yellow, crystalline solid.

Procedure for Formation of MSA Salt of 8 and Free

Basing. Crude 8 (23 kg, 64.9 mol), dichloromethane (206 L)

and isopropyl alcohol (19 L) were sequentially charged to a

reactor, and the resulting solution was cooled to 5 °C. The

solution was stirred for 15 min, and methanesulfonic acid (6.9

kg, 71 mol) was charged while maintaining the internal

temperature <20 °C. The mixture was warmed to 20 °C, and

the salt began to precipitate after approximately 30 min of

stirring. The slurry was stirred for additional 3 h and was

1

°C; H NMR (400 MHz CDCl₃) δ 8.80 (s, 1 H), 7.55−7.30

(m, 8 H), 5.14 (m, 2 H), 3.69 (s 3 H), 2.07 (s 6 H); ¹³C NMR

(400 MHz CDCl₃) δ 165.8, 154.5, 146.6, 137.0, 135.5, 130.7,

128.0, 127.4, 127.1, 120.1, 119.78, 119.3, 65.1, 51.5, 17.5; IR

(KBr pellet) ν_max3456, 3376, 3313, 3062, 3002, 2949, 2903,

2853, 1960, 1822, 1723, 1686, 1646, 1622, 1587, 1506, 1494,

1464, 1454, 1440, 1423, 1381, 1369, 1341, 1326, 1279, 1228,

1166, 1126, 1079, 1053, 1026, 999, 980, 964, 944, 906, 886,

876, 801, 775, 739, 697, 649; Anal. Calcd for C₂₀H₂₂N₂O₄: C,

67.78; H, 6.26; N, 7.90. Found: C, 67.51; H, 6.31; N, 7.97.

Methyl (R)-3-(4-Amino-3,5-dimethylphenyl)-2-(benzy-

loxycarbonylamino)-propanoate Methanesulfonate Salt

[9 (MSA)]. The enamide 8 (1.7 kg, 4.7 mol), methanol (7 L)

and dichloromethane (7 L) were sequentially charged to 20-L

glass-lined reactor, and the mixture was stirred for 15 min. The

mixture was purged three times at 30 psi with nitrogen, and the

catalyst, (+)-1,1′-bis-(2R,4R)-2,4-diethylphosphotano)-

ferrocene-(1,5-cyclooctadiene)rhodium(I)tetraﬂuoroborate

(Et-ferrotane-Rh, 8.6 g, 11.7 mmol, 0.25 mol %), was charged.

The charging ﬂask was rinsed with dichloromethane (0.5 L)

into the reactor. The slurry was again purged three times at 30

psi with nitrogen followed by hydrogen at 30 psi. The hydrogen

pressure was set at 45 psi, and the reaction mixture was stirred

at 25 °C for 1 h to complete consumption of 8. The reaction

mixture was ﬁltered through a 10 μm polypropylene polish

ﬁlter cloth. The reactor and polish ﬁlter cloth were rinsed with

ethyl acetate (16 L), and this rinse was combined with the

ﬁltrate.

The hydrogenation reactor was cleaned with dilute aqueous

nitric acid followed by a methanol rinse in between batches.

Three additional batches of 1.65 kg input per batch were

performed, and the ﬁltrates of the four batches were combined.

1961

dx.doi.org/10.1021/op3003097 | Org. Process Res. Dev. 2012, 16, 1953−1966

Organic Process Research & Development

Article

The solution was distilled under reduced pressure at 35 °C to a

volume of ∼20 L. Ethyl acetate (67 L) was charged, and the

distillation was continued under reduced pressure at 35 °C to

an end point of <0.5% v/v of methanol. Ethyl acetate (42 L)

was charged to adjust the volume to ∼59 L. A solution of

methanesulfonic acid (0.49 kg, 5.1 mol) in ethyl acetate (29 L)

was charged at ∼25 °C over 1 h. The resulting slurry was stirred

at 20 °C for 2 h and was ﬁltered. The cake was washed with

ethyl acetate (9 L) and was dried by vacuum suction on the

ﬁlter for 1 h. The product cake was further dried under vacuum

(40 °C, 25 mmHg) to aﬀord methyl (R)-3-(4-amino-3,5-

dimethylphenyl)-2-(benzyloxycarbonylamino)propanoate

methanesulfonate salt [9 (MSA)] (8.1 kg, 96.1% yield, 99.6

area %, 99.3 wt %, 99.0% ee) as a white crystalline solid.

Another set of three batches of 1.65 kg input per batch were

performed to produce an additional 6.1 kg of the MSA salt of 9

(96.5% yield, 99.5 area %, 98.9 wt %, 98.9% ee); mp 180.3 °C;

[α]²_D⁵+17.03 (c 1.00, MeOH). ¹H NMR (400 MHz DMSO-d₆)

δ 7.84 (d, J = 8.1 Hz, 1 H), 7.42−7.24 (m, 5 H), 7.02 (s, 2 H),

4.99 (dd, J = 5.0, 8.9 Hz, 2 H), 4.23 (ddd, J = 5.1, 8.2, 10.0 Hz,

1 H), 2.98−2.79 (m, 2 H), 2.34 (s, 3 H), 2.29 (3, 6 H); ¹³C

NMR (125 MHz DMSO-d₆) δ 172.2, 155.9, 136.8, 136.1,

130.4, 129.6, 128.3, 127.8, 127.6, 79.3, 78.9, 78.6, 65.4, 55.4,

51.9, 35.6, 17.6; IR (KBr pellet) ν_max3355 (br), 2956, 2607,

1740, 1694, 1659, 1631, 1610, 1531, 1491, 1446, 1383, 1321,

1294, 1242, 1162, 1068, 1040, 1024, 868, 777, 699, 560; Anal.

Calcd for C₂₁H₂₈N₂O₇S: C, 55.73; H, 6.23; N, 6.19; S, 7.08.

Found: C, 55.91; H, 6.30; N, 6.08; S, 6.85.

Methyl (R)-2-(Benzyloxycarbonylamino)-3-(7-methyl-

1H-indazol-5-yl)-propanoate (10). Methyl (R)-3-(4-amino-

3,5-dimethylphenyl)-2-(benzyloxycarbonylamino)-propanoate

methanesulfonate salt, 9 (MSA salt) (6.07 kg, 13.4 mol) and

ethyl acetate (74 L) was charged to a reactor, and the mixture

was stirred at 20 °C for 10 min. A solution of potassium

carbonate (2.0 kg, 14.1 mol) in water (30 L) was charged while

maintaining internal temperature at 20 °C. The charging line

was rinsed with water (5 L) to the reactor, and the mixture was

stirred for 30 min. The mixture was ﬁltered through a 2 in. bed

of Celite (1.2 kg) followed by polish ﬁltration though a 10 μm

Cuno cartridge in-line ﬁlter to remove any insoluble material.

The reactor, Celite cake and cartridge were rinsed with ethyl

acetate (13 L) to the reactor. After the phases were separated,

the aqueous (bottom) was discarded. The solution was washed

with a solution of sodium chloride (1.5 kg) in water (30 L).

The ethyl acetate solution was ﬁltered through a Cuno Zeta

pad (R55SP) at a rate of 1.5 L/min to remove residual metals.

The reactor and the Zeta pad were rinsed with ethyl acetate (13

L), and the rinse was combined with the ﬁltrate. An aliquot of

the combined ﬁltrates was analyzed to ensure the residual

palladium and rhodium levels were <10 ppm. The solution was

distilled under reduced pressure at 35 °C to a volume of ∼55 L

and a water content of <2.3%. Potassium acetate (1.64 kg,

16.76 mols) was added followed by acetic acid (1.21 kg, 20.12

mols, 1.5 equiv), and the mixture was stirred for 10 min.

Isoamyl nitrite (96%, 1.8 kg, 14.8 mol) was charged over 30

min, and the charging lines were rinsed with ethyl acetate (26

L) to the reactor. The reaction was heated to 65 °C and was

stirred for ∼5 h to ensure complete consumption of 9. The

mixture was cooled to 25 °C and a solution of potassium

carbonate (4.0 kg, 29.1 mol) in water (30 L) was added. The

mixture was stirred for 15 min, the phases were separated, and

the aqueous layer was discarded. A solution of sodium chloride

(1.5 kg) in water (30.4 L) was added, and the mixture was

stirred for 20 min. After the phases were separated, the aqueous

layer (bottom) was discarded, and the solution was distilled

under reduced pressure at ∼35 °C to volume of ∼22 L.

Toluene (61 L) was added, and the distillation was continued

under reduced pressure at 35 °C to an end point of <0.5% v/v

of ethyl acetate. Toluene (20 L) was charged to adjust the

volume to ∼42 L, and the slurry was heated to 70 °C. The light

slurry was stirred at ∼70 °C for 30 min and was cooled to 25

°C over 2 h. n-Heptane (30 L) was added over 30 min, and the

slurry was stirred at 25 °C for 4 h. The slurry was ﬁltered, and

the cake was washed with a mixture of toluene (9 L) and n-

heptane (21 L). The cake was dried by vacuum suction on the

ﬁlter for ∼1 h and was further dried under vacuum (50 °C, 25

mmHg) to aﬀord methyl (R)-2-(benzyloxycarbonylamino)-3-

(7-methyl-1H-indazol-5-yl)-propanoate (10) (3.6 kg, 73%

yield, 98.3 area %, 98.0 wt %, >99.9% ee) as an oﬀ-white

crystalline solid.

A second batch of 6.07 kg input was performed to produce

3.5 kg of 10 in 71% isolated yield and chemical purity (98.1

area %, 97.4 wt %) and chiral purity of 99.9% ee; mp 150.9 °C;

[α]²_D⁵+7.89 (c 1.03, MeOH). ¹H NMR (500 MHz DMSO-d₆) δ

13.08 (s, 1 H), 7.98 (s, 1 H), 7.83 (d, J = 8.2 Hz, 1 H), 7.41 (s,

1 H), 7.30−7.22 (m, 5 H), 7.02 (s, 1 H), 4.97 (s, 2 H), 4.31−

4.28 (m, 1 H), 3.62 (s, 3 H), 3.10 (dd, J = 5.0, 8.9 Hz, 1 H),

2.92 (dd, J = 3.5, 10.0 Hz, 1 H), 2.47 (s, 3 H); ¹³C NMR (125

MHz DMSO-d₆) δ 172.5, 156.0, 139.2, 136.9, 133.4, 129.4,

128.3, 127.7, 127.5, 127.3, 122.7, 119.6, 117.7, 65.3, 56.0, 51.9,

36.5, 16.8; IR (KBr pellet) ν_max3329, 3207, 3154, 3067, 2979,

2947, 2929, 2862, 2837, 2765, 1950, 1746, 1696, 1624, 1604,

1589, 1528, 1453, 1431, 1380, 1362, 1333, 1284, 1260, 1251,

1203, 1179, 1151, 1072, 1037, 1030, 1013, 991, 952, 909, 877,

847, 818, 779, 752, 738, 697, 672, 642; Anal. Calcd for

C₂₀H₂₁N₃O₄: C, 65.38; H, 5.76; N, 11.44. Found: C, 65.66; H,

5.48; N, 11.59.

Methyl (R)-2-Amino-3-(7-methyl-1H-indazol-5-yl)-

propanoate Dihydrochloride (3 bis-HCl). Method A.

Palladium on carbon (10%, 0.2 kg, 0.1 mol), 10 (2.6 kg, 7.1

mol), methanol (13 L) and ethyl acetate (7 L) were

sequentially charged to a hydrogenation reactor. The reactor

was purged repeatedly with nitrogen and then pressurized with

hydrogen to 45 psig. The reaction mixture was heated to 25 °C

and was stirred for 2 h to complete the consumption of 10.

Ethyl acetate (13 L) was charged, and the mixture was purged

with nitrogen for ∼30 min. The mixture was ﬁltered

sequentially through a 2 in. bed of Celite 545 (2 kg) and a

10 μm polypropylene ﬁlter, rinsing with a solution of ethyl

acetate (10 L) and methanol (11 L), and the rinse was

combined with the ﬁltrate. The combined ﬁltrate was ﬁltered

through a Cuno Zeta pad (R53SP) at a rate of 1.5 L/min to

remove the residual metals. The reactor and the Zeta pad were

rinsed with ethyl acetate (9 L), and the rinse was combined

with the ﬁltrate. An aliquot of the ﬁltrate was taken and

analyzed to ensure the residual palladium and rhodium levels

were <10 ppm. The solution was polish ﬁltered through a 0.5

μm in-line polish ﬁlter and ethyl acetate (27 L) was added to

adjust the ethyl acetate/methanol ratio to 3:1. A solution of

HCl in IPA (4.5 M, 3.3 kg, 15.6 mol) was added over 1 h, and

the resulting slurry was stirred at 20 °C for 3 h. The slurry was

ﬁltered, and the cake was washed with ethyl acetate (4 L). The

cake was dried by cacuum suction on the ﬁlter for 1 h and was

further dried under vacuum (35 °C, 25 mmHg) to aﬀord (R)-

methyl 2-amino-3-(7-methyl-1H-indazol-5-yl)propanoate dihy-

drochloride (3 bis-HCl) (1.85 kg, 85.6% yield, 99.5 area %, 98

1962

dx.doi.org/10.1021/op3003097 | Org. Process Res. Dev. 2012, 16, 1953−1966

Organic Process Research & Development

Article

wt %, >99.9% ee) as a white crystalline solid with less than 5

ppm of both palladium and rhodium.

A second batch with a 2.6 kg input was performed to

produce 1.80 kg of 3 bis-HCl in 83% isolated yield, chemical

purity 98 wt %, and chiral purity >99.9% ee.

over 30 min while maintaining the internal temperature at <50

°C. The reaction mixture was stirred at 45 °C for 6 h to ensure

the complete consumption of 16. The reaction mixture was

cooled to 25 °C and was charged with ethyl acetate (60 L)

followed by water (54 L). The mixture was stirred at 25 °C for

20 min, and aqueous sodium hydroxide solution (4 M, 20 L)

was charged to adjust the pH to ∼9.0. The mixture was stirred

for an additional 1 h,, and the phases were separated. The

aqueous layer was extracted with ethyl acetate (32 L), and the

ethyl acetate extract was combined with the organic phase. The

combined organic phase was washed with water (54 L), and the

solution was distilled under reduced pressure at 55 °C to a

volume of ∼43 L. Toluene (63 L) was charged, and the

distillation was continued under reduced pressure at 35 °C to

an end point of <5% v/v of ethyl acetate. The resulting slurry

was heated to 85 °C and n-heptane (65 L) was charged over ∼1

h. The slurry was cooled to 25 °C over 2 h and stirred for an

additional 6 h. The slurry was ﬁltered, and the cake was washed

with n-heptane (2 × 32 L). The cake was dried by vacuum

suction on the ﬁlter for 1 h and was further dried under vacuum

(35 °C, 25 mmHg) to aﬀord 5-bromo-7-methyl-1H-indazole,

17, (9.43 kg, 81.7% yield, 99.2 wt %) as a light-brown

crystalline solid; mp 185.7 °C. ¹H NMR (500 MHz DMSO-d₆)

δ 13.52 (br, 1 H), 8.14 (s, 1 H), 7.89 (s, 1 H), 7.36 (s, 1 H),

2.63 (s, 3 H); ¹³C NMR (125 MHz DMSO-d₆) δ 138.9, 133.3,

128.1, 124.0, 122.6, 120.1, 112.6, 16.6; IR (KBr pellet) ν_max

3449, 3151, 3068, 2936, 2875, 2839, 2778, 1712, 1695, 1618,

1592, 1572, 1494, 1449, 1432, 1412, 1374, 1341, 1300, 1277,

1241, 1205, 1165, 1104, 1087, 1070, 1039, 1014, 985, 946, 908,

870, 856, 851, 822, 779, 739, 612; HRMS Calcd for C₈H₇BrN₂:

210.9865 [(M + H)⁺], found 210.9862.

7-Methyl-1H-indazole-5-carbaldehyde (18). 5-Bromo-

7-methyl-1H-indazole, 17, (8.6 kg, 40.5 mol) and THF (128 L)

were charged to a reactor. The mixture was stirred for 20 min,

and the resulting solution was cooled to −70 °C. A solution of

n-butyllithium in hexanes (2.5 M, 16.9 kg, 60.8 mol) was

charged while maintaining the internal temperature at <−60

°C. The charging lines were rinsed with THF (3 L) followed by

cyclohexane (2 L) into the reactor. The mixture was stirred at

−70 °C for ∼1 h, and a solution of sec-butyllithium in

cyclohexane (2.0 M, 54.1 kg, 101.3 mol) was charged while

maintaining the internal the temperature <−60 °C. The

charging lines were rinsed with cyclohexane (3 L) followed

by THF (2 L) into the reactor. The mixture was stirred at −70

°C for ∼1 h to ensure the complete consumption of 17.

Dimethylformamide (11.8 kg, 162.0 mol,) was charged while

maintaining the internal temperature <−55 °C. The reaction

mixture was warmed to 5 °C over 2 h and was charged with

water (86 L) while maintaining the internal temperature <20

°C. The mixture was stirred for 15 min, and the phases were

separated. The organic layer was concentrated by distillation at

reduced pressure and 40 °C to a ﬁnal volume of ∼55 L. The

concentrated solution was extracted with aqueous sodium

bisulﬁte solution (10%, 5 × 26 L). The product rich aqueous

extracts were combined. Ethyl acetate (128 L) was charged, and

the pH of the mixture was adjusted to between pH 8 to 9 with

aqueous sulfuric acid (2 M, 17 L). Sodium chloride (43 kg) was

charged and mixture was stirred for 30 min. After the phases

were separated, the aqueous layer (bottom) was discarded. The

solution was concentrated by distillation at reduced pressure at

45 °C to a volume of ∼26 L. n-Heptane (60 L) was charged,

and the distillation was continued under reduced pressure at

∼45 °C to an end point of <5% v/v of ethyl acetate. n-Heptane

Method B. Step 1: Formation of the bis-HCl Salt of 14.

Indazolyl amino acid 14 (2.0 kg, 8.8 mol) and methanol (16 L)

were charged to a reactor, and the mixture was stirred for 10

min. The mixture was cooled to <10 °C, and acetyl chloride

(1.45 kg, 18.5 mol) was charged while maintaining the internal

temperature at <20 °C. The charging lines were rinsed with

methanol (1 L) into the reactor. The mixture was stirred at 20

°C for 30 min, and MTBE (34 L) was charged over 1 h. The

slurry was stirred at 20 °C for 2 h and was ﬁltered. The cake

was washed with MTBE (11 L) and was dried by vacuum

suction on the ﬁlter for 12 h to aﬀord (R)-2-amino-3-(7-

methyl-1H-indazol-5-yl)propanoic acid dihydrochloride (14

1

bis-HCl) (2.38 kg, 94.5% yield, 97.5 area %). H NMR (400

MHz CDCl₃) δ 8.73 (s, 1 H), 7.78 (s, 1 H), 7.48 (s, 1 H), 4.38

(dd, J = 1.9, 5.7 Hz, 1 H), 4.38 (dd, J = 5.6, 8.9 Hz, 1 H), 2.65

(s, 3 H); ¹³C NMR (400 MHz CDCl₃) δ 171.4, 141.4, 133.9,

132.9, 132.2, 124.3, 123.0, 121.7, 55.3, 37.4, 17.4.

Step 2: Fischer Esteriﬁcation. (R)-2-Amino-3-(7-methyl-1H-

indazol-5-yl)propanoic acid dihydrochloride (14 bis-HCl)

(2.34 kg, 8.0 mol) and methanol (55 L) were charged to a

reactor, and the mixture was stirred at 20 °C for 30 min. The

resulting solution was recirculated through a 10 μm Cuno

polish ﬁlter for 1 h. The polish ﬁlter and the circulating lines

were rinsed with methanol (2 L) into the reactor. The

combined solution was cooled to <10 °C, and acetyl chloride

(3.1 kg, 40.0 mol) was charged while maintaining the internal

temperature at <20 °C. The charging lines were rinsed with

methanol (1 L) into the reactor. The reaction mixture was

heated to 40 °C and was stirred for 16 h to ensure complete

consumption of 14. The reaction mixture was cooled to <25 °C

and was concentrated by distillation under reduced pressure at

<30 °C to a volume of ∼21 L. The resulting slurry was warmed

to 35 °C and was charged with MTBE (19 L) over 1 h. The

slurry was stirred at 20 °C for 2 h and was ﬁltered. The cake

was washed with MTBE (5.8 L) and dried by vacuum suction

on the ﬁlter for ∼1 h. The cake was further dried under vacuum

(25 °C, 25 mmHg) to aﬀord methyl (R)-2-amino-3-(7-methyl-

1H-indazol-5-yl)propanoate dihydrochloride (3 bis-HCl) (2.2

kg, 90.7% yield, 98.6 area %, 74.7 wt %, 99.0% ee) as a white

crystalline solid; mp 162.9 °C; [α]²_D⁵−7.06 (c 1.03, MeOH). ¹H

NMR (500 MHz DMSO-d₆) δ 8.66 (s, 1 H), 7.73 (s, 1 H), 7.43

(s, 1 H), 4.34 (dd, J = 1.9, 5.6 Hz, 1 H), 3.44 (dd, J = 5.7, 9.1

Hz, 1 H), 3.29 (dd, J = 1.9, 6.7 Hz, 1 H), 2.61 (s, 3 H); ¹³C

NMR (125 MHz DMSO-d₆) δ 169.3, 139.5, 133.2, 127.4,

126.7, 122.6, 120.2, 118.4, 53.5, 52.4, 35.7, 16.8; IR (KBr

pellet) ν_max3426, 2948, 2841, 2702, 2614, 2535, 2437, 1742,

1637, 1588, 1529, 1448, 1433, 1391, 1360, 1302, 1269, 1238,

1217, 1139, 1113, 1065, 1007, 983, 951, 923, 897, 879, 843,

752, 713, 622; Anal. Calcd for C₁₂H₁₇Cl₂N₃O₂: C 47.07, H

5.59, Cl 23.15, N 13.72. Found: C 47.07, H 5.59, Cl 22.94, N

13.70.

5-Bromo-7-methyl-1H-indazole (17). 4-Bromo-2,6-dime-

thylaniline, 16 (10.8 kg, 53.7 mol), potassium acetate (8.0 kg,

81.6 mol,) and toluene (162 L) were sequentially charged to a

reactor. The mixture was stirred for 30 min and was charged

with acetic acid (4.6 kg, 76.8 mol) while maintaining internal

temperature at <40 °C. The mixture was stirred at 40 °C and

tert-butyl nitrite (90% purity, 7.1 kg, 61.8 mol) was charged

1963

dx.doi.org/10.1021/op3003097 | Org. Process Res. Dev. 2012, 16, 1953−1966

Organic Process Research & Development

Article

(34 L) was charged to adjust the volume to ∼60 L. The slurry

was cooled to 5 °C over 2 h and stirred for an additional 2 h.

The slurry was ﬁltered, and the cake was washed with n-heptane

(9 L). The cake was dried by vacuum suction on the ﬁlter for

∼1 h and was further dried under vacuum (40 °C, 25 mmHg)

to aﬀord 7-methyl-1H-indazole-5-carbaldehyde, 18, (5.84 kg,

90.3% yield, 98.5 wt %) as a white crystalline solid; mp 183.5

°C; ¹H NMR (500 MHz DMSO-d₆) δ 13.61 (s, 1 H), 9.97 (s, 1

H), 8.32 (s, 1 H), 8.24 (s, 1 H), 7.59 (s, 1 H), 2.56 (s, 3 H);

¹³C NMR (125 MHz DMSO-d₆) δ 192.5, 142.8, 136.5, 130.6,

125.7, 123.6, 122.5, 121.4, 16.9; IR (KBr pellet) ν_max3516,

3354, 3250, 3075, 2937, 2827, 2797, 2717, 1816, 1798, 1688,

1669, 1620, 1602, 1503, 1489, 1462, 1440, 1383, 1376, 1351,

1333, 1324, 1283, 1243, 1213, 1125, 1083, 1060, 998, 941, 889,

860, 834, 802, 760, 705, 613; Anal. Calcd for C₉H₈N₂O: C,

67.49; H, 5.03; N, 17.49. Found: C, 67.17; H, 4.85; N, 17.53.

4-((7-Methyl-1H-indazol-5-yl)methylene)-2-phenylox-

azol-5(4H)-one (19). 7-Methyl-1H-indazole-5-carbaldehyde,

18, (5.4 kg, 33.9 mol) and potassium acetate (4.6 kg, 47.1 mol)

were charged to a reactor. Acetic anhydride (86.5 kg, 847 mol)

was charged to the mixture while maintaining the internal

temperature at <40 °C. The mixture was stirred at 25 °C for 1

h, and hippuric acid (12.2 kg, 67.8 mol) was charged. The

reaction mixture was heated to 60 °C over 30 min and was

stirred at 60 °C for 3 h to ensure the complete consumption of

18. The reaction mixture was cooled to 25 °C and water (38 L)

was charged while maintaining the internal temperature at <40

°C. The slurry was heated to 60 °C and was stirred for at 2 h.

The slurry was cooled to ∼20 °C over 1 h and ﬁltered. The

cake was washed with a (1:1) methanol/water mixture (109 L)

and dried by vacuum suction on the ﬁlter for 1 h. The cake was

further dried under vacuum (40 °C, 25 mmHg) to aﬀord 4-((7-

methyl-1H-indazol-5-yl)methylene)-2-phenyloxazol-5(4H)-

one, 19, (10.95 kg, 93.9% yield, 99.3 wt %) as a yellow

crystalline solid; mp 225.2 °C. ¹H NMR (500 MHz DMSO-d₆)

δ 13.32 (s, 1 H), 10.06 (s, 1 H), 8.11 (s, 1 H), 8.03 (s, 1 H),

8.02 (s, 1 H), 7.97 (s, 1 H), 7.64−7.61 (m, 1 H), 7.58−7.55 (m,

3 H), 7.53 (s, 1 H), 3.75 (s, 3 H), 2.46 (s, 3 H); ¹³C NMR (125

MHz DMSO-d₆) δ 166.3, 165.8, 140.2, 134.9, 134.8, 133.6,

131.9, 128.6, 127.7, 127.2, 126.2, 124.4, 122.8, 121.2, 120.0,

52.2, 17.0; IR (KBr pellet) ν_max3426, 3231, 3061, 2945, 2850,

2784, 1731, 1718, 1646, 1615, 1580, 1509, 1483, 1437, 1373,

1361, 1318, 1248, 1191, 1160, 1134, 1104, 1084, 1027, 979,

946, 915, 902, 871, 835, 797, 766, 750, 714, 692, 633, 616;

Anal. Calcd for C₁₉H₁₇N₃O₃: C, 68.05; H, 5.11; N, 12.53.

Found: C, 68.03; H, 5.12; N, 12.59.

2-Hydroxy-3-(7-methyl-1H-indazol-5-yl)acrylic Acid

(15). Methyl 2-benzamido-3-(7-methyl-1H-indazol-5-yl)-

acrylate, 20, (1.8 kg, 5.1 mol), water (2 L), and 30% aqueous

hydrochloric acid (49 kg) were sequentially charged to a

reactor. The reaction mixture was heated to ∼100 °C and was

stirred for 6 h to ensure a complete consumption of 20. The

reaction mixture was cooled to 10 °C over 2 h, and the slurry

was ﬁltered. The cake was washed with water (17 L) and was

charged back to the reactor. Water (20 L) was charged, and the

mixture was warmed to 90 °C. The slurry was stirred at 90 °C

for 2 h and was ﬁltered through a preheated ﬁlter (90 °C). The

cake was washed with hot water (90 °C, 10 L) and was dried by

vacuum suction on the ﬁlter for 1 h. The cake was dried under

vacuum (50 °C, 25 mmHg) to aﬀord an E/Z mixture of 2-

hydroxy-3-(7-methyl-1H-indazol-5-yl)acrylic acid (15) (0.89

1

kg, 74.6% yield, 97.2 area %, 96.10 wt %) as a white solid. H

NMR (500 MHz CDCl₃) E-isomer: 8.11 (s, 1 H), 8.0 (d, J =

1.3 Hz, 1 H), 7.68 (d, J = 1.9 Hz, 1 H), 7.54 (t, J = 1.9, 5.0 Hz,

1 H), 6.57 (s, 1 H), 2.56 (s, 3 H); Z-isomer: 8.13 (s, 1 H), 8.02

(d, J = 1.0 Hz, 1 H), 7.69 (s, 1 H), 7.58 (t, J = 1.6, 8.8 Hz, 1 H),

2.56 (s, 3 H); ¹³C NMR (125 MHz CDCl₃) 167.4, 166.6,

140.2, 139.4, 134.3, 132.9, 130.8, 129.3, 128.6, 127.8, 127.7,

123.0, 119.6, 119.0, 110.9, 16.9; IR (KBr pellet) ν_max3473,

3445, 3071, 2972, 2838, 2670, 2603, 2557, 1788, 1687, 1617,

1602, 1583, 1518, 1496, 1453, 1424, 1369, 1327, 1291, 1258,

1240, 1186, 1152, 1128, 1095, 1073, 1026, 993, 963, 945, 890,

876, 845, 833, 810, 768, 751, 707; HRMS Calcd for

C₁₁H₁₁N₂O₃: 219.0764 [(M + H)⁺], found 219.0764.

Preparation of (R)-Amino Acid 14 from Keto Acid 15.

[Note: this procedure was previously reported in ref 13 but is

reproduced here for the convenience of the reader.] Water

(445 mL), keto acid 15 (30 g, 0.14 mol), ^DL-alanine (120 g,

1.35 mol), K₂HPO₄(6.97 g, 0.04 mol), and KH₂PO₄(1.36 g,

0.01 mol) were charged to a 2-L reactor followed by 30%

sodium hydroxide (8 M, 17.3 mL), and the suspension was

stirred at 200 rpm for 10 min to dissolve most of the solids.

Additional water (445 mL) was added to bring the total volume

to 1 L, and stirring was continued for 20 min to completely

dissolve the solids. The pH was adjusted to 7.5 with a few drops

of 25% NaOH. Dithiothreitol (154 mg, 1 mmol) in water (2

mL) and pyridoxal phosphate monohydrate (26.5 mg, 0.1

mmol) in pH 7.5 buﬀer (2 mL) were then added to the

solution which was stirred at 50 rpm and 30 °C. ^D-

Transaminase (300 mg, 4.4 U/mg, Biocatalytics) dissolved in

25 mL of 0.1 M potassium phosphate buﬀer pH 7.5 was then

added to the stirred solution. After precipitation of 14

commenced (∼3 h), stirring was periodically increased to 150

rpm so that aliquots for HPLC analysis could be obtained.

Reaction was continued until the area % of 15 was <10 area %

relative to that for 14, which took about 72 h. The pH was

adjusted to 1 with sulfuric acid at which point product 14

1

crystalline solid.; mp 224.8 °C; H NMR (500 MHz DMSO-

d₆) δ 13.49 (s, 1 H), 8.52 (s, 1 H), 8.25 (s, 1 H), 8.19 (s, 1 H),

8.13 (s, 1 H), 8.12 (d, J = 1.3 Hz, 1 H), 7.72−7.69 (m, 1 H),

7.65−7.62 (m, 2 H), 7.41 (s, 1 H), 2.58 (s, 3 H); ¹³C NMR

(125 MHz DMSO-d₆) δ 167.1, 161.8, 141.0, 135.8, 133.3,

132.7, 130.7, 129.3, 128.8, 127.8, 126.5, 125.3, 125.2, 123.2,

120.7, 17.0; IR (KBr pellet) ν_max3434, 3232, 3146, 3057, 2939,

2851, 1789, 1770, 1755, 1650, 1609, 1597, 1583, 1557, 1501,

1491, 1450, 1369, 1325, 1298, 1258, 1209, 1173, 1144, 1106,

1085, 1062, 1009, 979, 943, 913, 888, 877, 858, 776, 757, 719,

699, 686, 673, 655, 632, 619; Anal. Calcd for C₁₈H₁₃N₃O₂: C,

71.28; H, 4.32; N, 13.85. Found: C, 70.80; H, 3.85; N, 13.77.

Methyl 2-Benzamido-3-(7-methyl-1H-indazol-5-yl)-

acrylate (20). 4-((7-Methyl-1H-indazol-5-yl)methylene)-2-

phenyloxazol-5(4H)-one, 19, (10.9 kg, 31.4 mol) and methanol

(163 L) were charged to a reactor. The mixture was stirred for

10 min, and a 30% solution of sodium methoxide in methanol

(0.6 kg, 3.1 mol) was charged while maintaining the internal

temperature at <30 °C. The reaction mixture was stirred at 25

°C for 3 h to ensure complete consumption of 19. Acetic acid

(0.4 kg, 6.3 mol) was charged to the reaction mixture, and the

mixture was stirred for 30 min. The mixture was charged with

water (163 L) while maintaining the internal temperature at

<30 °C. The resulting slurry was cooled to 5 °C and was stirred

for 1 h. The slurry was ﬁltered, and the cake was washed with

water (43 L). The cake was dried by vacuum suction on the

ﬁlter for 1 h and was further dried under vacuum (50 °C, 25

mmHg) to aﬀord methyl 2-benzamido-3-(7-methyl-1H-inda-

zol-5-yl)acrylate, 20, (10.3 kg, 94.8% yield, 96.6 wt %) as white

1964

dx.doi.org/10.1021/op3003097 | Org. Process Res. Dev. 2012, 16, 1953−1966

Organic Process Research & Development

Article

redissolved. The solution was ﬁltered, and the ﬁltrate was

passed through a 2.5 cm i.d. × 9.1 cm column of Amberlite F

XAD-16 resin, washing with 0.1 M H₂SO₄to remove nonpolar

impurities. The column eﬄuent was heated to 100 °C, and the

pH was slowly adjusted to 7 with NaOH. The mixture was

cooled to room temperature, stirred for 15 h, and ﬁltered,

washing the cake with deionized water. The cake was dried in

vacuum at room temperature, giving 24.6 g of 14 as the

monohydrate. The yield from keto acid 15 was 77%, correcting

for the purity of the keto acid (94.2 wt %) and isolated product

(96.5 wt %). [α]²_D⁹+17.3 (c 0.08, HOAc); ¹H NMR (400 MHz,

acetic acid-d₄) 2.54 (s, 3H), 3.27 (dd, J = 14.5, 8.2 Hz, 1H),

3.45 (dd, J = 14.7, 4.6 Hz, 1H), 4.39 (dd, J = 7.9, 4.8 Hz, 1H),

7.16 (s, 1H), 7.54 (s, 1H), 8.11 (s, 1H). Anal. Calcd for

C₁₁H₁₃N₃O₂.H₂O: C, 55.68, H, 6.37, N, 17.71; found: C, 55.62;

H, 6.45, N, 17.63. Water, calculated for monohydrate 7.59;

found: 7.50 (Karl Fischer).

Telescoped Preparation of (R)-3-(7-Methyl-1H-inda-

zol-5-yl)-2-(4-(2-oxo-1,2-dihydroquinolin-3-yl)-

piperidine-1-carboxamido)propanoic Acid (11). Step 1:

CDI Activated Urea Coupling. Methyl (R)-2-amino-3-(7-

methyl-1H-indazol-5-yl)propanoate hydrochloride, 3 bis-HCl,

(1.4 kg, 4.6 mol), N,N-dimethylformamide (DMF, 6 L), and

N,N-diisopropylethylamine (DIPEA, 1.2 kg, 9.6 mol) were

sequentially charged to a reactor. The mixture was stirred for 20

min, and the resulting solution was cooled to −20 °C. A slurry

of 1,1′-carbonyldiimidazole (CDI, 0.9 kg, 5.5 mol) in DMF (4

L) was charged while maintaining the temperature at <−10 °C.

The mixture was stirred at −10 °C for 15 min to ensure the

complete consumption of 3. A solution of water (0.4 kg, 22.9

mol) in DMF (1 L) was charged while maintaining the

temperature at <−10 °C. The resulting mixture was stirred at

−10 °C for 15 min to ensure the complete quench of CDI. 3-

(Piperidin-4-yl)quinolin-2(1H)-one hydrochloride (4 HCl salt)

(1.3 kg, 4.7 mol), DIPEA (0.7 kg, 5.0 mol), and methylene

chloride (3 L) were sequentially charged to the reactor. The

reaction mixture was heated to 40 °C and was stirred for 12 h

to ensure the complete consumption of 24. The reaction

mixture was cooled to 20 °C and was charged with aqueous

hydrochloric acid (2 M, 6.5 kg, 13.1 mol) to adjust the pH of

the mixture to 2. The resulting mixture was extracted with

methylene chloride (8 L), and the product-rich methylene

chloride phase was washed with water (14 L).

(s, 1H), 7.37−7.47 (m, 2H), 7.26 (d, J = 8.3 Hz, 1H), 7.11−7.2

(m, 1H), 7.04 (s, 1H), 6.67 (d, J = 7.8 Hz, 1H), 4.03−4.14 (m,

2H), 3.04−3.14 (m, 1H), 2.94−3.03 (m, 1H), 2.83−2.94 (m,

1H), 2.61−2.79 (m, 2H), 2.46 (s, 3H), 1.67−1.81 (m, 2H),

1.18−1.38 (m, 2H); ¹³C NMR (100 MHz, DMSO-d₆) δ 173.4,

160.6, 156.0, 138.4, 136.6, 136.2, 133, 132.6, 129.8, 128.6,

126.9, 126.8, 122.0, 121, 118.7, 117, 114, 55.8, 44.2, 44, 36.6,

35.5, 30.8, 17.1; IR (KBr pellet) ν_max3215 (br), 2930, 2856,

1716, 1646, 1615, 1571, 1517, 1426, 1314, 1267, 1222, 1084,

983, 944, 870, 755, 714, 590, 467; HRMS Calcd for

C₂₆H₂₈N₅O₄: 474.2136 [(M + H)⁺], found 474.2063.

(R)-N-(3-(7-Methyl-1H-indazol-5-yl)-1-(4-(1-methylpi-

peridin-4-yl)piperazin-1-yl)-1-oxopropan-2-yl)-4-(2-oxo-

1,2-dihydroquinolin-3-yl)piperidine-1-carboxamide (1

HCl salt). (R)-3-(7-methyl-1H-indazol-5-yl)-2-(4-(2-oxo-1,2-

dihydroquinolin-3-yl)-piperidine-1-carboxamido)propanoic

acid, 11, (1.81 kg, 3.8 mol), 1-hydroxybenzotriazole mono-

hydrate (HOBT, 0.6 kg, 3.8 mol), dimethylacetamide (DMAc,

3 L), diisopropylethylamine (DIPEA, 1.0 kg, 7.6 mol) and

piperazine 2 (0.84 kg, 4.6 mol) were sequentially charged to a

reactor while maintaining the temperature <45 °C. The mixture

was stirred at 45 °C for 15 min, and 1-[3-(dimethylamine)-

propyl]-3-ethylcarbodiimide (EDAC-HCl, 1.0 kg, 4.6 mol) was

added portionwise to ensure a good dispersion, rinsing with

DMAc (1 L). The reaction mixture was cooled to 25 °C and

was stirred for ∼24 h to consume 11. The reaction mixture was

polish ﬁltered through a 10 μm in-line polypropylene ﬁlter,

rinsing the reactor and the ﬁlter lines with ethanol (200 proof, 2

L). The combined ﬁltrate was heated to 45 °C over 30 min.

Acetone (11 L) was charged while maintaining the temperature

>41 °C. The solution was seeded with a slurry of 1 (HCl salt)

(4.5 g, 0.3 wt %) in acetone (0.2 L), and the mixture was stirred

at 45 °C for ∼1 h. Acetone (43 L) was charged over 30 min,

and the resulting slurry was stirred at 45 °C for 1 h. The slurry

was cooled to 20 °C over 30 min and was stirred for 3 h. The

slurry was ﬁltered, and the cake was washed with acetone (7 L).

The cake was dried by vacuum suction on the ﬁlter for 1 h and

was further dried under vacuum (45 °C, 25 mmHg) to aﬀord

crude (R)-N-(3-(7-methyl-1H-indazol-5-yl)-1-(4-(1-methylpi-

peridin-4-yl)piperazine-1-yl)-1-oxopropan-2-yl)-4-(2-oxo-1,2-

dihydroquinolin-3-yl)piperidine-1-carboxamide (1 HCl salt)

(2.32 kg, 90% yield, 95.5 wt %, 99.9% ee) as a white crystalline

solid.

Step 2: Saponiﬁcation. A solution of lithium hydroxide

(0.16 kg, 7.1 mol) in water (4 L) was charged to the product-

rich methylene chloride solution. The mixture was stirred at 20

°C for 40 min to ensure complete hydrolysis. After the phases

were separated, the methylene chloride phase was discarded.

Water (8 L) was charged, and the diluted aqueous phase was

transferred into aqueous hydrochloric acid (1 M, 10.2 kg, 10.1

mol) while maintaining the temperature below 15 °C. The

resulting slurry with pH between 1 and 2 was stirred at <20 °C

for 2 h and was ﬁltered. The cake was washed with water (4 ×

14 L) to reduce the conductivity of the ﬁltrate to below 2.0 mS.

The cake was further washed with MTBE (14 L) and was

deliquored (20 °C, 25 mmHg) in an agitation drier with

minimum stirring for 16 h. The drying was continued under

vacuum (35 °C, 25 mmHg) to aﬀord (R)-3-(7-methyl-1H-

indazol-5-yl)-2-(4-(2-oxo-1,2-dihydroquinolin-3-yl)-piperidine-

1-carboxamido)propanoic acid, 11, (1.90 kg, 84% yield, 95.6 wt

%, 99.3% ee) as a white amorphous solid; [α]²_D⁵+28.7 (c 1.00,

A second batch with a 1.81 kg input of 11 was performed to

produce 2.24 kg of 1 (HCl salt) in 86% isolated yield and

chemical purity (98.8 area %, 95.5 wt %) and chiral purity of

99.9% ee.

Recrystallization of 1 (HCl salt). Crude (R)-N-(3-(7-methyl-

1H-indazol-5-yl)-1-(4-(1-methylpiperidin-4-yl)piperazin-1-yl)-

1-oxopropan-2-yl)-4-(2-oxo-1,2-dihydroquinolin-3-yl)-

piperidine-1-carboxamide (2.0 kg, 3.0 mol) and ethanol (40 L)

were charged to a reactor, and the mixture was heated to 73 °C

over 30 min. The resulting solution was cooled to 45 °C and

was polish ﬁltered through a 10 μm Cuno in-line ﬁlter, rinsing

with EtOH (2 L), and the combined ﬁltrate was heated to 45

°C over 30 min and was charged with acetone (3 L). The

solution was stirred for 15 min and was seeded with a slurry of

1 HCl salt (15 g, 0.8 wt %) and acetone (367 mL). The mixture

was stirred at 45 °C for 1 h, and acetone (39 L) was charged

over 30 min. The slurry was stirred at 45 °C for 1 h, cooled to

20 °C over 30 min, and stirred for 12 h. The slurry was ﬁltered,

and the cake was washed with acetone (3 × 10 L). The cake

was dried by vacuum suction on the ﬁlter for 1 h and was

1

DMSO); H NMR (400 MHz, DMSO-d₆) δ 12.4−13.4 (br,

2H), 11.76 (s, 1H), 7.97 (s, 1H), 7.61 (d, J = 6.9 Hz, 1H), 7.54

1965

dx.doi.org/10.1021/op3003097 | Org. Process Res. Dev. 2012, 16, 1953−1966

Organic Process Research & Development

Article

(6) Baer, E.; Maurukus, J.; Jonas, D. D. Can. J. Chem. 1956, 34,

1182−1188.

further dried under vacuum (65 °C, 25 mmHg) to aﬀord (R)-

N-(3-(7-methyl-1H-indazol-5-yl)-1-(4-(1-methylpiperidin-4-

yl)piperazine-1-yl)-1-oxopropan-2-yl)-4-(2-oxo-1,2-dihydroqui-

nolin-3-yl)piperidine-1-carboxamide (1, HCl salt) (1.81 kg,

89% yield, 99.8 wt %, >99.9% ee) as a white crystalline solid;

(7) Kajigaeshi, S.; Kakinami, T.; Yamasaki, T.; Fujisaki, S.; Okamoto,

T. Bull. Chem. Soc. Jpn. 1988, 61, 600−602.

(8) Burk, M. J.; Feaster, J. E.; Nugent, W. A.; Harlow, R. L. J. Am.

Chem. Soc. 1993, 115, 10125−10138.

(9) Berens, U.; Burk, M. J.; Gerlach, A.; Hems, W. Angew. Chem., Int.

Ed. 2000, 29, 1981−1984.

(10) Cobley, C. J.; Lennon, I. C.; Praquin, C.; Zenotti-Gerosa, A.;

Appell, R. B.; Goralski, C. T.; Sutterer, A. C. Org. Process Res. Dev.

2003, 7, 407−411.

(11) Product purities and relative percent impurities are HPLC area

% unless otherwise indicated.

1

mp 276.6 °C; [α]²_D⁵−27.3 (c 1.04, MeOH); H NMR (400

MHz DMSO-d₆) δ 13.07 (s, 1 H), 11.78 (s, 1 H), 7.97 (s, 1 H),

7.64 (d, J = 7.8 Hz, 1H), 7.60 (s, 1 H), 7.43 (t, J = 7.4 Hz, 1 H),

7.38 (s, 1 H), 7.38 (s, 1 H), 7.29 (d, J = 8.3 Hz, 1 H), 7.15 (t, J

= 7.8 Hz, 1 H), 7.03 (s, 1 H), 6.71 (d, J = 7.8 Hz, 1 H), 4.79

(dd, J = 7.3, 7.8 Hz, 1 H), 4.15 (d, J = 12.2 Hz, 1 H), 3.55−2.67

(m, 15 H), 2.62 (s, 3 H), 2.49 (s, 3 H), 2.35−2.00 (m, 4 H),

1.79−1.75 (m, 2 H), 1.71−1.45 (m, 5 H), 1.38−1.29 (m, 2 H);

¹³C NMR (125 MHz, CD₃OD) δ 173.8, 164.7, 159.6, 141.5,

139.0, 138.1, 137.0, 135.7, 131.6, 131.5, 130.4, 129.3, 125.0,

124.2, 122.4, 122.1, 120.2, 116.6, 59.7(br), 55.2(br), 53.6, 47.3,

46.2, 44.0(br), 43.6, 40.4, 37.6, 32.6, 26.8(br), 26.7(br), 17.7;

IR (KBr pellet) ν_max3600−3200 (br), 1658, 1637, 1627, 1611,

1533, 1469, 1227 Anal. Calcd for C₃₆H₄₇ClN₈O₃: C, 64.03; H,

7.02; Cl, 5.25; N, 16.59. Found: C, 64.07; H, 7.76; Cl, 5.30; N,

16.50.

(12) Boekelheide, V.; Schamm, L. M. J. Org. Chem. 1949, 14, 298−

311.

(13) Hanson, R. L.; Davis, B. L.; Goldberg, S. L.; Johnston, R. M.;

Parker, W. L.; Tully, T. P.; Montana, M. A.; Patel, R. N. Org. Process

Res. Dev. 2008, 12, 1119−1129.

(14) At the time this work was completed, Dow Chemical Company

was the only bulk supplier.

(15) Although the step count in the enzymatic route begins with

commercially available bromide 16, the bromination of 5 to aﬀord 16

must be considered when assessing the global environmental impact.

(16) Recently, water has been proposed as an advantageous solvent

for some CDI reactions: Padiya, K. J.; Gavade, S.; Kardile, B.; Tiwari,

M.; Bajare, S.; Mane, M.; Gaware, V.; Harel, D.; Kurhade, S. Org. Lett.

2012, 14, 2814−2817.

(17) For a somewhat related example where an insoluble product

overcomes the expected equilibrium between two amine hydro-

chlorides, see: Anderson, N. G.; Ary, T. D.; Berg, J. L.; Bernot, P. J.;

Chan, Y. Y. Org. Process Res. Dev. 1997, 1, 300−310.

AUTHOR INFORMATION

■

Corresponding Author

*reginald.cann@bms.com

Present Addresses

^†Hovione LLC, East Windsor, New Jersey.

^‡Vertex Pharmaceuticals, Inc., Cambridge, Massachusetts.

^§Escientia Life Sciences, South Glastonbury, Connecticut.

^⊥Red Spot Paint and Varnish Co., Evansville, Indiana.

^∥NetChem, Inc., New Brunswick, New Jersey.

NOTE ADDED AFTER ASAP PUBLICATION

■

This paper was published on the Web on November 26, 2012.

Scheme 16 has been updated and the corrected version was

reposted on December 13, 2012.

Notes

The authors declare no competing ﬁnancial interest.

ACKNOWLEDGMENTS

■

We thank Drs. Jaan Pesti, Robert Waltermire, Animesh

Goswami, David Conlon, and Prashant Deshpande for useful

suggestions during the preparation of this manuscript. We also

acknowledge the important contributions of Mr.Victor Rosso

and the BMS Laboratory Automation group as well as analytical

support provided by Mr. Michael Peddicord for assistance with

mass spectrometry.

REFERENCES

■

(1) (a) Edvinsson, L. CNS Drugs 2001, 15, 745−753. (b) Grant, A.

D. Br. J. Pharmacol. 2002, 135, 356−362. (c) Durham, P. L.; Russo, A.

F. Pharmacol. Ther. 2002, 94, 77.

(2) (a) Chaturvedula, P. V.; Chen, L.; Civiello, R.; Conwy, C. M.;

Degnan, A. P.; Dubowchik, G. M.; Han, X.; Karageorg, G. N.; Luo, G.;

Macor, J. E.; Poindexter, G.; Vig, S. Calcitonin gene related peptide

receptor agonists. WO 2003104236 (A1) 2003, CAN 140:42208.

(b) Macor, J. E. 2010 Spring Pain Research Conference, Grand

Cayman, April, 2010.

(3) Burgey, C. S.; Stump, C. A.; Williams, T. A. Preparation of

benzodiazepine CGRP receptor antagonists. WO 20052000807 (A2),

2005, CAN 142:114110.

(4) For a recent review, see: Asymmetric Synthesis and Application

of α-Amino Acids. , Soloshonok, V. A., Izawa, K., Eds.; ACS

Symposium Series 1009; American Chemical Society: Washington,

DC, 2009.

(5) Chaturvedula, P. V.; Han, X.; Jiang, X. Novel process for

preparation of CGRP receptor antagonists and intermediate thereof.

WO 2006060678 (A2, A3) 2006, CAN 145:46084.

1966

dx.doi.org/10.1021/op3003097 | Org. Process Res. Dev. 2012, 16, 1953−1966

Article Doi

Selection of an enantioselective process for the preparation of a CGRP receptor inhibitor

DOI: 10.1021/op3003097

Source and publish data:

Authors:

Article abstract of DOI:10.1021/op3003097

Full text of DOI:10.1021/op3003097

Products guided by the article

R&D Labs maybe for 1414976-16-1

Relevant to this article

Hot Product