Development of a large scale asymmetric synthesis of the glucocorticoid agonist BI 653048 BS H3PO4

Article

pubs.acs.org/joc

Development of a Large Scale Asymmetric Synthesis of the

Glucocorticoid Agonist BI 653048 BS H₃PO₄

Jonathan T. Reeves,*^,†Daniel R. Fandrick,^†Zhulin Tan,^†Jinhua J. Song,^†Sonia Rodriguez,^†Bo Qu,^†

Soojin Kim,^†Oliver Niemeier,^§Zhibin Li,^†Denis Byrne,^†Scot Campbell,^‡Ashish Chitroda,^†

Phil DeCroos,^†Thomas Fachinger,^§Victor Fuchs,^†Nina C. Gonnella,^‡Nelu Grinberg,^†Nizar Haddad,^†

Burkhard Jager,^§Heewon Lee,^†Jon C. Lorenz,^†Shengli Ma,^†Bikshandarkoil A. Narayanan,^†

̈

Larry J. Nummy,^†Ajith Premasiri,^†Frank Roschangar,^†Max Sarvestani,^†Sherry Shen,^†Earl Spinelli,^†

Xiufeng Sun,^†Richard J. Varsolona,^†Nathan Yee,^†Michael Brenner,^§and Chris H. Senanayake^†

^†Chemical Development, Boehringer Ingelheim Pharmaceuticals, Inc., 900 Old Ridgebury Road, P.O. Box 368, Ridgeﬁeld,

Connecticut, 06877-0368, United States

^‡Analytical Development, Boehringer Ingelheim Pharmaceuticals, Inc., 900 Old Ridgebury Road, P.O. Box 368, Ridgeﬁeld,

Connecticut, 06877-0368, United States

^§Boehringer Ingelheim GmbH & Co. KG, Binger Strasse 173, 55216 Ingelheim am Rhein, Germany

S

* Supporting Information

ABSTRACT: The development of a large scale synthesis of the glucocorticoid agonist BI 653048 BS H₃PO₄(1·H₃PO₄) is

presented. A key triﬂuoromethyl ketone intermediate 22 containing an N-(4-methoxyphenyl)ethyl amide was prepared by an

enolization/bromine−magnesium exchange/electrophile trapping reaction. A nonselective propargylation of triﬂuoromethyl

ketone 22 gave the desired diastereomer in 32% yield and with dr = 98:2 from a 1:1 diastereomeric mixture after crystallization.

Subsequently, an asymmetric propargylation was developed which provided the desired diastereomer in 4:1 diastereoselectivity

and 75% yield with dr = 99:1 after crystallization. The azaindole moiety was eﬃciently installed by a one-pot cross coupling/

indolization reaction. An eﬃcient deprotection of the 4-methoxyphenethyl group was developed using H₃PO₄/anisole to produce

the anisole solvate of the API in high yield and purity. The ﬁnal form, a phosphoric acid cocrystal, was produced in high yield and

purity and with consistent control of particle size.

present our results on the development of a large scale,

INTRODUCTION

■

asymmetric propargylation based synthesis of 1·H₃PO₄.³

Traditional anti-inﬂammatory agents used for the treatment of

rheumatoid arthritis are steroids such as prednisolone and

dexamethasone.¹While eﬀective, these compounds unfortu-

nately can cause undesirable side eﬀects due to activation of the

glucocorticoid receptor. It is therefore desirable to identify

nonsteroidal glucocorticoids with increased selectivity that

avoid the side eﬀects of traditional steroidal agents. Compound

1·H₃PO₄(BI 653048 BS H₃PO₄) was identiﬁed by our

Medicinal Chemistry Department as a candidate for develop-

ment for the treatment of rheumatoid arthritis (Figure 1).²To

support development work and clinical studies, a safe, scalable,

and eﬃcient synthesis of 1·H₃PO₄was required. Herein we

RESULTS AND DISCUSSION

■

The Medicinal Chemistry synthetic route to 1 is shown in

Scheme 1.²The route relies on a diastereoselective addition of

lithiated (S)-p-tolyl methyl sulfoxide 7 to triﬂuoromethyl

ketone 6 to set the chiral center with 2:1 diastereoselectivity.⁴

Other notable features of the synthesis are the elaboration of an

Received: January 21, 2013

A

dx.doi.org/10.1021/jo400079z | J. Org. Chem. XXXX, XXX, XXX−XXX

The Journal of Organic Chemistry

Article

Scheme 2. Key Challenges for Scale-up of the Medicinal

Chemistry Synthesis

Figure 1. Structure of BI 653048 BS H₃PO₄.

aryl methyl group to the primary carboxamide, the conversion

of the β-hydroxy sulfoxide into a homopropargylic alcohol, and

the synthesis of the azaindole via Sonogashira cross coupling of

alkyne 19 with iodide 20 followed by a base-mediated

indolization of 21.⁵The synthesis proceeds in 17 linear steps

with an overall yield of 0.8%.

After extensive route-scouting, we arrived at the “chiral

amide” route as the most promising and direct synthesis of 1.

This route addressed the two key issues identiﬁed in the

Medicinal Chemistry route, while maintaining the cross-

coupling/cyclization strategy for late-stage heterocycle intro-

duction. The overall concept for this route is shown in Scheme

3. A propargylation of triﬂuoromethyl ketone 22 would directly

install the requisite homopropargyl alcohol in 23 and would

avoid the need for any functional group manipulations as with

the chiral sulfoxide addition. A chiral N-phenethyl amide in the

triﬂuoromethyl ketone substrate 22 would serve multiple

purposes. First, it would potentially render the propargylation

diastereoselective. Second, the diastereomeric hydroxy amides

produced in the propargylation reaction might be separable by

crystallization. Lastly, the phenethyl group could be removed

under acidic conditions to give the required primary

carboxamide directly, thereby avoiding the extensive functional

group manipulations and oxidation state adjustments of the

Medicinal Chemistry route.⁹Employing the same cross

coupling/cyclization sequence for late-stage installation of the

On evaluation of the discovery route for potential scale-up,

several key concerns were identiﬁed (Scheme 2).⁶First, the

introduction of the chiral center proceeded with moderate

selectivity, employing the expensive chiral sulfoxide 7,⁷and

several steps were required to elaborate the β-hydroxy sulfoxide

8 into the homopropargylic alcohol moiety necessary for the

Sonogashira cross coupling. Second, the conversion of the aryl

methyl group into the primary carboxamide required several

steps, including a radical bromination. The radical bromination

posed a signiﬁcant safety concern as a potential runaway

reaction.⁸Finally, the length and low overall yield of the

synthesis meant a large investment in raw materials and

manpower would be required for direct scale-up of the current

route. On the other hand, the late stage installation of the

azaindole unit by a cross coupling/cyclization sequence was

attractive, as was the general strategy for installing the chiral

center by addition of a nucleophile to a triﬂuoromethyl ketone.

With these points in mind, we set out to design a more eﬃcient

and scalable route to 1.

Scheme 1. Medicinal Chemistry Synthesis of 1

B

dx.doi.org/10.1021/jo400079z | J. Org. Chem. XXXX, XXX, XXX−XXX

The Journal of Organic Chemistry

Article

Scheme 3. Concept for Chiral Amide Route

Scheme 4. Synthesis of Triﬂuoromethyl Enone 28

Scheme 5. Unsuccessful Conjugate Additions to Enone 28

Scheme 6. Conjugate Addition to Enone 28

azaindole would provide 24, from which the phenethyl group

could be removed with acid to give 1.

prepared by adding the morpholine amide 26 to a 5 °C solution

of Grignard reagent 27. The isolation of this enone was

challenging due to its low boiling point and solubility in water.

A special workup/isolation protocol had to be devised. After

aging for 0.5 h, the reaction was quenched with aqueous HCl.

Dodecane was added, and the aqueous layer removed.

Additional washes with water eﬀected the removal of most of

the THF, while the product enone remained in the dodecane

layer. The enone was then distilled (bp 108 °C at 1 atm) from

the higher boiling dodecane (bp 217 °C at 1 atm) either by

simple vacuum distillation on lab scale or by thin ﬁlm

distillation on kilogram scale. This process provided enone

28 as a 45−75 wt% solution in THF with small amounts of

dodecane. This process was used to prepare 10 kg of 28.

The ﬁrst challenge in the demonstration of the chiral amide

route was the development of a synthesis of the amide

triﬂuoromethyl ketone 22. The conjugate addition of aryl

Grignard reagents to enone 28 was developed in our Medicinal

Chemistry Department and appeared to be a potential route to

ketone 22 (Scheme 4).^1,2,10The enone 28 was prepared by

addition of commercially available 2-methyl-1-propenylmagne-

sium bromide 27 to the N-triﬂuoroacetyl Weinreb amide 25.

After routine reaction safety analysis showed the reagent 25 to

be a high energy compound, we switched to the analogous

morpholine amide 26, which eliminated any safety concerns

and gave the same yield and quality as 25.¹¹Enone 28 was

C

dx.doi.org/10.1021/jo400079z | J. Org. Chem. XXXX, XXX, XXX−XXX

The Journal of Organic Chemistry

Article

The conjugate addition of arylmetal reagents already

functionalized with the amide or an equivalent oxidation state

functional group to enone 28 was initially explored (Scheme 5).

Unfortunately, the reagents derived from halogen-metal

exchange of halides 29−32 in the presence of various copper

salts led to either very low yields, exclusive 1,2-addition, or no

reaction at all under a variety of reaction conditions.

We next explored the conjugate addition of the aryl Grignard

derived from commercially available 2-bromo-4-ﬂuoro-1-

iodobenzene 33 (Scheme 6). If successful, the aryl bromide

could potentially be elaborated into an amide by a subsequent

aminocarbonylation reaction. The Knochel procedure for

generation of the requisite aryl Grignard 34 from iodobenzene

33 was employed.¹²Iodine−magnesium exchange was found to

occur within 30 min using isopropylmagnesium chloride in

THF at −20 °C. Subsequent addition of 10 mol % CuI

followed by enone 28 gave the crude ketone 35 after workup

with aqueous NH₄Cl and EtOAc, which was further puriﬁed by

vacuum distillation to give highly pure (>95 wt%) bromoke-

tone 35. The distillation was required to remove byproducts

and impurities which are shown in Scheme 7. Proto-quenched

structures 41 and 42 were detected. Byproduct 41 arises from

coupling of the enolate of 35 with the aryl bromide, and 42

could form from direct amination of 35 or, alternatively, by

initial hemiaminal formation followed by intramolecular C−N

bond formation.

Given the observed sensitivity of the free triﬂuoromethyl

ketone substrate, protection of the ketone as a dioxolane prior

to aminocarbonylation was subsequently explored. Under

standard conditions for acetal formation (catalytic acid,

ethylene glycol, azeotropic removal of water) no product was

formed, even after extended reaction times. This observation

was consistent with the literature on triﬂuoromethyl ketones,

which have been shown to be highly resistant to acid catalyzed

ketalization due to the electron withdrawing eﬀect of the

triﬂuoromethyl group, which disfavors adjacent oxonium ion

formation.¹⁷By switching to basic reaction conditions using 2-

chloroethanol, however, the desired dioxolane 43 was formed

in excellent yield (Scheme 9).¹⁸The aminocarbonylation of 43

with amine 40 proceeded smoothly to give the amide 44.

Deprotection of the dioxolane group proved to be challenging,

again due to the electronic eﬀect of the triﬂuoromethyl group.

Standard acid catalyzed hydrolysis was completely ineﬀective, as

was the use of strong Lewis acids such as TiCl₄. It is known that

triﬂuoromethyl ketone ketals may be deprotected under

dealkylative conditions with boron trihalides.¹⁶These con-

ditions were successful in partially deprotecting the ketal, but

concomitant demethylation of the aryl methoxy group to give

45 as well as other side reactions could not be avoided.

Due to the diﬃculties encountered in developing an

aminocarbonylation route to 22, a more direct alternative

route was explored. A process in which the amide would be

introduced by conversion of the aryl bromide to a Grignard

reagent and subsequent trapping with isocyanate 46 was

envisioned (Scheme 10).¹⁹To avoid deprotection of the

dioxolane group, an in situ “protection” of the triﬂuoromethyl

ketone as its enolate was proposed. The enolate should be

stable to reagents used for bromine/metal exchange, and the

ketone would be regenerated during aqueous quench of the

reaction mixture. Due to the low pK_aof triﬂuoromethyl

ketones, enolization should be facile, and furthermore, the

nucleophilicity of the enolate should be low, thus minimizing

competitive attack of the enolate on the isocyanate. The base to

be used for enolization needed to fulﬁll several requirements.

First, it must not generate a species after ketone deprotonation

which could be deprotonated on the subsequent addition of

organomagnesium reagents for bromine−magnesium exchange.

This excluded typical alkoxide or amide bases which would

generate alcohols or amines after deprotonation. Second, it

could not generate a species after ketone deprotonation which

could react with the isocyanate. This requirement also excluded

alkoxide or amide bases. Finally, it could not be a nucleophilic

Scheme 7. Structures and Quantities of Impurities

Generated in the Conjugate Addition Step

Grignard byproduct 36 and 1,2-addition byproduct 37 were the

major impurities. The dimer 38 and phenol 39 were generated

in small amounts, provided the reaction was maintained under a

rigorously oxygen-free atmosphere. In one kilo-lab batch, an

accidental introduction of air caused large amounts of dimer 38

(16%) and phenol 39 (3.4%) to be generated. The dimer

formation can be explained by the known dimerization of aryl

cuprates in the presence of oxygen,¹³while the phenol

formation likely resulted from reaction of the aryl Grignard

reagent with oxygen.¹⁴This conjugate addition reaction was

used to prepare 14 kg of 35.

With the bromoketone 35 in hand, we investigated methods

to convert the aryl bromide into the N-4-methoxyphenethyla-

mide.¹⁵Initial eﬀorts focused on aminocarbonylation to install

the amide.¹⁶The direct aminocarbonylation of 35 under all

conditions investigated gave a complex mixture of products,

from which only trace amounts of the desired product 22 were

detected (Scheme 8). The starting material 35 was rapidly

consumed under the reaction conditions. By LC-MS analysis of

the reaction mixture, byproducts with masses corresponding to

Scheme 8. Attempted Aminocarbonylation of 35

D

dx.doi.org/10.1021/jo400079z | J. Org. Chem. XXXX, XXX, XXX−XXX

The Journal of Organic Chemistry

Article

Scheme 9. Aminocarbonylation of a Protected Substrate

resulted in smooth generation of H₂gas and the formation of

sodium enolate 47, as conﬁrmed by React-IR measurements

(vide infra). Aging experiments showed enolate 47 to undergo

no decomposition on aging at rt for several days.

Scheme 10. Concept for One-step Conversion of 35 to 22 via

Enolization/Bromine Exchange/Isocyanate Quench

Bromine−metal exchange of enolate 47 was investigated with

several reagents (Table 1). The use of n-BuLi at −65 °C

resulted in fast and complete exchange. Subsequent trapping

with isocyanate 46 provided ketone 22 in low yield (25%),

accompanied by many byproducts (entry 1). Despite the low

yield, this result was encouraging in that it validated the ketone

“protection” as an enolate strategy. The use of 0.4 equiv of

lithium tributyl magnesiate at 0 °C resulted in a fast exchange,

with the bromide being consumed within 15 min (entry 2).

Quenching with isocyanate 46 and workup resulted in a 53%

isolated yield of 22 after chromatography. The use of either

isopropylmagnesium chloride (entry 3) or isopropylmagnesium

chloride lithium chloride complex (Turbo Grignard, entry 4)

resulted in low conversions after 24 h at rt.²⁰Knochel and co-

workers have described the use of Turbo Grignard in

combination with ∼2 equiv of 1,4-dioxane to generate the

highly active exchange reagent “i-Pr₂Mg-LiCl”.²¹The 1,4-

dioxane serves to drive the Schlenk equilibrium toward the

dialkylmagnesium reagent by precipitation of MgCl₂-dioxane

complex. By adding 2 equiv of 1,4-dioxane to the reaction with

base, since this would result in competitive addition to the

triﬂuoromethyl ketone instead of enolization. This requirement

excluded most organolithium and organomagnesium bases.

With these restrictions in mind, sodium hydride seemed a

logical choice since it generates only hydrogen gas as a

byproduct and it is not nucleophilic. Treatment of 35 with an

excess (1.2 equiv) of NaH (60 wt% in mineral oil) in THF at rt

Table 1. Screening of Reagents for Bromine−Metal Exchange

a

entry

conditions

conversion (%)

yield (%)

b

1

2

3

4

5

6

n-BuLi (1 equiv), −65 °C, 15 min

n-Bu₃MgLi (0.4 equiv), 0 °C, 15 min

i-PrMgCl (1.2 equiv), rt, 24 h

100

<5

25

b

53

i-PrMgCl-LiCl (1.2 equiv), rt, 24 h

10

i-PrMgCl (1.2 equiv), dioxane (2 equiv), rt, 24 h

i-PrMgCl-LiCl (1.1 equiv), dioxane (2 equiv), rt, 5 h

40−85

>98

c

85

a

b

Conversion of bromine−metal exchange of 47 to 48 as measured by HPLC analysis of reaction aliquots quenched into H₂O. Isolated yield of 22

c

after quenching with isocyanate 46, workup, and puriﬁcation by chromatography on SiO₂. Isolated yield of 22 after quenching with isocyanate 46,

workup, and puriﬁcation by crystallization.

E

dx.doi.org/10.1021/jo400079z | J. Org. Chem. XXXX, XXX, XXX−XXX

The Journal of Organic Chemistry

Article

Figure 2. Monitoring of the enolization of 35 with React IR.

Figure 3. Acceleration of the enolization of 35 on addition of water.

Scheme 11. Optimized Procedure for the Synthesis of Chiral Amide Ketone 22

Turbo Grignard (entry 6), a dramatic acceleration in the

exchange was observed, and complete consumption of the

bromide was achieved after 5 h at rt. In this case, a clean

reaction proﬁle was achieved, and the product 22 was isolated

in 85% yield after crystallization. Adding dioxane to the reaction

with isopropylmagnesium chloride (entry 5) was also eﬀective

in accelerating the exchange, although in this case the rate was

inconsistent from batch to batch, and did not reach complete

conversion as with Turbo Grignard/dioxane.

and new absorptions attributable to the enolate 47 appear

(Figure 2).

This technique was particularly beneﬁcial on scale-up of the

reaction, when a critical dependence of the enolization rate on

the water content of the reaction mixture was observed.

Laboratory batches typically used THF containing 100−500

ppm water. On scaling to a 4 kg reaction in a 50 L reactor,

however, the THF used from a 200 L drum was extremely dry

(∼10 ppm water). The enolization was found to be much

slower for this reaction. After 18 h at rt, only ∼50% enolization

had occurred, compared with laboratory scale batches in which

the enolization was always completed after 0.5−1 h. On

addition of a catalytic amount (5 mol %) of water to aliquots of

The enolization of 35 was eﬀectively monitored by React IR.

The carbonyl absorbance of 35 at ∼1750 cm⁻¹disappears on

addition of 1.2 equiv of NaH to a THF solution of 35 at 0 °C

F

dx.doi.org/10.1021/jo400079z | J. Org. Chem. XXXX, XXX, XXX−XXX

The Journal of Organic Chemistry

Article

Table 2. Screen of Propargylation Conditions

a

b

Conversion of 22 to 23 + 49 as measured by HPLC analysis of reaction aliquots quenched into H₂O. Diastereomeric ratio of 23 to 49 measured in

c

d

the reaction mixture by HPLC analysis prior to workup. HPLC assay yield. For entries 3−6, the crude product was ﬁrst treated with NaOMe/

MeOH to remove the alkynyl TMS group. Isolated yield of 23 after crystallization from hexanes/EtOAc. Diastereomeric ratio of 23 to 49 measured

in the crystallized, isolated 23 by HPLC analysis.

e

f

the reaction mixture, however, the enolization rapidly went to

completion (Figure 3). The acceleration of the enolization

could also be eﬀected by the addition of catalytic amounts of

alcohols, though with lesser eﬃciency.²²In order to ensure

reproducibility on scale-up, a speciﬁcation for water content of

the THF was set at 300−500 ppm water. With this speciﬁcation

as well as in process monitoring by React IR, the enolization

gave reproducible results.

The optimized conditions for conversion of bromoketone 35

to chiral amide ketone 22 are shown in Scheme 11. After

enolization and bromine/magnesium exchange, the reaction

mixture was cooled to 0 °C and quenched with isocyanate 46.

After 30 min, the reaction mixture was quenched with aqueous

HCl and toluene, and the aqueous layer separated. THF was

distilled and the product crystallized out on addition of heptane

and water. This process was employed to make the ﬁrst 20 kg

of 22.

With the chiral amide ketone in hand, investigations into the

propargylation reaction commenced. The reaction with

propargyl bromide and Al metal in the presence of catalytic

HgCl₂gave a clean and complete conversion to a 1:1 mixture of

the diastereomeric alcohols 23 and 49 (Table 2, entry 1). While

the lack of diastereoselectivity was disappointing, we were

gratiﬁed to ﬁnd that recrystallization of the 1:1 mixture of 23

and 49 from hexanes/EtOAc gave a 30% yield of diastereomer

23 with a diastereomeric purity of 99:1. This eutectic controlled

crystallization thus enabled us to obtain the challenging tertiary

alcohol stereocenter in pure form, despite the nonselective

propargylation reaction. We then screened further reagents and

conditions for propargylation in the hopes of increasing the

diastereoselectivity. The use of Zn metal gave access to the

product in similar yield and selectivity (entry 2). While this was

positive in that it removed toxic HgCl₂from the process, we

still faced a safety concern from the use of shock-sensitive

propargyl bromide. Corey reported the deprotonation of 1-

trimethylsilylpropyne with n-BuLi/TMEDA in Et₂O and the

subsequent addition of the propargyl lithium species to alkyl

halides and carbonyl compounds.²³The application of the

Corey conditions, only replacing Et₂O with THF, gave a low

conversion to product with a dr of 1:1 (entry 3). Extensive

variation of solvent and temperature led to no increase in the

conversion. We speculated that the strongly basic propargyl

lithium competitively enolized the triﬂuoromethyl ketone either

directly or by initial deprotonation of the amide followed by

intramolecular proton transfer and thereby shut down the

carbonyl addition process. We then investigated transmetala-

tion of the propargyl lithium reagent to magnesium (entry 4)

and zinc (entries 5 and 6) in the hopes of generating a less basic

species and avoiding deprotonation. While the use of MgBr₂

gave a low conversion as with the lithium reagent, the use of

ZnBr₂gave encouraging results, although they were not

reproducible.²⁴After extensive study of the reaction conditions,

we found that TMEDA was the culprit for irreproducibility.

Upon omission of TMEDA from the reaction, not only was the

deprotonation of TMS propyne still eﬀective, but the

propargylation gave a consistently high conversion and clean

reaction proﬁle. The reasons for TMEDA shutting down the

reaction are unclear. We speculate that the propargyl zinc

G

dx.doi.org/10.1021/jo400079z | J. Org. Chem. XXXX, XXX, XXX−XXX

The Journal of Organic Chemistry

Article

Scheme 12. Optimized Conditions for Nonselective Propargylation

Scheme 13. Deprotection of 23 and Proof of Structure and Stereochemistry

Scheme 14. One-Pot Cross Coupling/Cyclization of 23 to 24

species generated in the presence of TMEDA is likely

coordinated by TMEDA, rendering it more electron rich and

more basic, and thus more capable of causing enolization.

Marshall’s palladium catalyzed propargylation with propargyl

benzenesulfonate (entry 7) was also investigated, but gave a low

conversion with no diastereoselectivity.²⁵

After the propargylation, a 1:1 mixture of the diastereomeric

TMS alkynes 50 and 51 was produced (Scheme 12). Removal

of the trimethylsilyl group could be accomplished in the same

pot by quenching the propargylation reaction with MeOH and

adding sodium methoxide (25 wt% in MeOH). After 1 h at rt,

the proto-desilylation was complete, giving the diastereomeric

alkynes 23 and 49. Upon acidiﬁcation and workup, a solution of

the crude product in i-PrOAc was obtained, and after

concentration and addition of heptane, the desired diaster-

eomer 23 crystallized out in 31−33% yield with a dr of >98:2.

This propargylation reaction in combination with the powerful

crystallization enabled access to the key chiral intermediate in

pure form, and 6 kg of 23 was produced using this procedure.

We next investigated the deprotection of the 4-methox-

yphenethyl group. Literature procedures for removal of this

group from amides included the use of TFA, TsOH, or H₂SO₄.⁹

We found that heating a solution of 23 in neat TFA at 50 °C

for 16 h gave the amide 19 in 80% yield (Scheme 13). Because

amide 19 was also an intermediate in the Medicinal Chemistry

route to 1 (see Scheme 1), we could conﬁrm the structure of

this intermediate by comparison. Furthermore, conversion of

19 by the same cross coupling/cyclization sequence into 1

conﬁrmed the absolute conﬁguration and structure. With these

results, we had validated the chiral amide propargylation route.

There are two possible sequences of the last two conversions:

the sequence employed above, with deprotection of the amide

prior to heterocycle installation, or alternatively the heterocycle

installation followed by amide deprotection. We preferred the

latter sequence for two reasons. First, by moving the amide

deprotection to the last chemical step, we allowed an additional

step for removal of any residual Pd. If the cross coupling/

cyclization was left as the last chemical step, residual Pd control

might be more challenging. Second, by leaving the chiral amide

H

dx.doi.org/10.1021/jo400079z | J. Org. Chem. XXXX, XXX, XXX−XXX

The Journal of Organic Chemistry

Article

Scheme 15. Deprotection of 24 to 1·AcOH using HBr/AcOH

in place for an additional step, we would have another

opportunity to enrich chiral purity in the crystallization after

heterocycle installation. With this plan in mind, we investigated

the cross coupling/cyclization reaction of 23. The Medicinal

Chemistry route employed a Sonogashira cross coupling of 19

with iodopyridine 20 to give the alkynyl pyridine 21. This

compound was then treated with DBU in MeOH to eﬀect

cyclization with concomitant loss of the Boc group to give the

azaindole 1. We suspected that these two reactions could be

combined into a one-pot protocol. This proved possible by

running the Sonogashira reaction with MeOH as the solvent,

and adding DBU on completion of the cross coupling reaction.

It was also found that excluding CuI from the reaction resulted

in not only clean and complete conversion, but also eliminated

the formation of an impurity derived from alkyne homocou-

pling. A screen of numerous Pd catalysts and bases for the cross

coupling reaction was done. While numerous Pd catalysts were

eﬀective, Pd(OAc)₂was found to be the optimal catalyst in

terms of eﬃcacy and cost. Of the bases screened (NaOMe, N-

methylpyrrolidine, Et₃N, DBU, DABCO, quinuclidine, i-

Pr₂NEt, piperidine, tetramethylguanidine) DABCO gave the

best results. The use of DBU in the cross coupling reaction

resulted in no reaction. Interestingly, having MeOH as solvent

was critical to the success of the reaction. When other alcohols

(EtOH, n-PrOH, i-PrOH) were used, large amounts of

impurities were generated. The optimized procedure involved

heating a mixture of alkyne 23, iodide 20 (1.01 equiv), DABCO

(2.0 equiv), Pd(OAc)₂(0.5 mol %) and MeOH (2 volumes) at

65 °C for 3−5 h to eﬀect the cross coupling to give

intermediate 52 (Scheme 14). Addition of DBU (1.5 equiv)

followed by heating 2 h at 50 °C provided the azaindole 24.

The product was crystallized after addition of acetonitrile and

water and cooling to rt, and was isolated in 83% yield, with 97−

98 area% purity by HPLC, and 10−30 ppm of residual Pd.

Signiﬁcant optimization of the crystallization conditions was

done in order to minimize the amount of residual Pd in the

product, and also to achieve a fast ﬁltration of the solid. Initially,

MTBE/water was charged after the reaction for crystallization.

While this gave low levels of residual Pd in the product (30−70

ppm), the ﬁltration rate of the solid in the pilot plant was slow.

By switching to MeCN/water for the crystallization, a faster

ﬁltration rate was achieved, with even lower levels of residual

Pd (10−30 ppm).

temperatures (70−80 °C) was eﬀective in removing the 4-

methoxyphenethyl group, though with formation of several

impurities. Published conditions using p-toluenesulfonic acid in

toluene at 110 °C resulted in fast deprotection (<1 h) but with

formation of large amounts of impurities.^9bIt was found that

using HBr (48% aqueous) and AcOH (2 volumes and 4

volumes, respectively) at 80 °C for 8 h gave a relatively clean

conversion to 1. The reaction mixture was then treated with

toluene and enough aqueous NaOH to neutralize all HBr. The

toluene served to solubilize poly(4-methoxy)styrene generated

in the reaction as the byproduct of the 4-methoxyphenethyl

group. The product crystallized out as an AcOH solvate in 80−

85% yield with a purity of ∼96% (Scheme 15). The wet cake

could be recrystallized from toluene/AcOH to upgrade the

purity to 98−99%, and the product was obtained in an overall

yield of 67%. The product gave residue on ignition values of 0.2

up to 0.5%, indicating the presence of inorganic salts. This

process was used for production of 20 kg of 1·AcOH.

The ﬁnal form of the drug was a phosphoric acid cocrystal.

The cocrystal form exhibited desirable physicochemical proper-

ties, stability characteristics, and improved solubility and

bioavailability compared to the free form. The free base 1,

with a pK_avalue of 1.75, was unlikely to form a salt with

phosphoric acid (pK_a= 2.15), but it formed a stable and

consistent crystal form as a complex with phosphoric acid. The

single crystal X-ray structure analysis resolved the structure as a

cocrystal with 1:1 stoichiometry between phosphoric acid and

the API, with hydrogen bonding interactions between

phosphoric acid and the amide group of the free base (Figure

4).²⁶The process for formation of 1·H₃PO₄needed to not only

provide the API in high purity but also control the particle size

The deprotection of the 4-methoxyphenethyl group to

produce 1 was then investigated. The use of neat TFA at

50−60 °C, as was done for the deprotection of the alkynyl

substrate 23, was also eﬀective on 24, and after 16−20 h a clean

and complete conversion to 1 was achieved. The use of neat

TFA was not desirable for scale-up, however, and therefore a

further screen of acids was undertaken. The use of concentrated

HCl, 50% aqueous H₂SO₄or methanesulfonic acid at elevated

Figure 4. X-ray crystal structure of 1·H₃PO₄. Hydrogen bonding

interaction shown as dotted line.

I

dx.doi.org/10.1021/jo400079z | J. Org. Chem. XXXX, XXX, XXX−XXX

The Journal of Organic Chemistry

Article

Scheme 16. Phosphoric Acid Co-crystal Formation

Scheme 17. Nonselective Propargylation Route

at a d(90) of <15 μm. The latter requirement arose from the

inability to mill 1·H₃PO₄without introducing amorphous

content. A simple and robust process was developed as shown

in Scheme 16. The AcOH solvate 1·AcOH (or the anisole

solvate 1·PhOMe, vide infra) was dissolved in methyl ethyl

ketone (MEK) at 60 °C, and the resultant solution was clean-

ﬁltered to remove any particulates. A slight excess (1.05 equiv)

of 85% aqueous H₃PO₄was charged at 50 °C, followed by

heptane. The mixture was seeded, and additional heptane was

added. The batch was cooled, ﬁltered and the solid washed with

MEK/heptane and ﬁnally heptane. The API was obtained in

93% yield from the AcOH solvate 1·AcOH (or 97% yield from

the anisole solvate 1·PhOMe, vide infra) and in >99.5% purity

by HPLC and with a consistent particle size of <10 μm. This

process was employed to produce 94 kg of 1·H₃PO₄.

The ﬁrst generation route is summarized in Scheme 17. The

route enabled the synthesis of 1 in 6 chemical steps plus 1 step

for the ﬁnal cocrystal formation, compared with the discovery

route of 17 chemical steps. The overall yield was increased from

0.8 to 6.8%. To facilitate further scale-up some aspects of the

synthesis required improvement. First, the need for a tedious

distillation for puriﬁcation of enone 28 was not desirable for

large scale production. Second, isocyanate 46 required

phosgene for its preparation, which posed a safety concern.

In addition, the stability of isocyanate 46 was a concern, as a

diﬃcult to remove symmetrical urea impurity formed over time

on exposure to air and moisture. These issues prompted

investigations into a synthesis of chiral amide ketone 22 which

did not proceed through enone 28 or isocyanate 46. Next, the

lack of diastereoselectivity in the propargylation reaction and

the resultant low isolated yield of the product 23 greatly

decreased the overall yield and throughput of the synthesis. The

development of a diastereoselective propargylation was there-

fore critical. Finally, the modest yield obtained in the amide

deprotection step, the need for recrystallization to upgrade the

purity, the elevated salt content of the product, and the

formation of poly(4-methoxy)styrene warranted an improved

amide deprotection process which addressed these issues.

As a replacement of the enone 28, we examined the

chemistry of isopropylidene Meldrum’s acid 55 (Scheme 18).

This compound was prepared by Vogt and co-workers by the

condensation of Meldrum’s acid 54 with acetone in the

presence of 4 Å MS, catalytic NH₄OAc, and toluene as solvent

at rt.²⁷Although this procedure was eﬀective, the use of

J

dx.doi.org/10.1021/jo400079z | J. Org. Chem. XXXX, XXX, XXX−XXX

The Journal of Organic Chemistry

Article

Scheme 18. Preparation of Isopropylidene Meldrum’s Acid

Scheme 20. Conversion of Acid 57 to Ketone 35

55

powdered 4 Å MS was not desirable for large scale operations.

It was found that by using acetone as the reaction solvent, and

catalytic amounts of AcOH and morpholine, 55 could be

obtained in 79% yield. In this case the conversion of 54 to 55

did not reach completion, but rather ∼90% conversion. The

unreacted residual Meldrum’s acid could conveniently be

removed by a wash with aqueous NaOH. Importantly, 55

was a highly crystalline compound, and easily isolated in high

purity by crystallization from MTBE/cyclohexane.

The conjugate addition of aryl Grignard reagent 34, derived

as previously described by iodine−magnesium exchange of aryl

iodide 33 with i-PrMgCl, proceeded smoothly in the presence

of 5 mol % CuI to give the adduct 56 in 84% yield (Scheme

19). Subsequently, it was found that the conjugate addition

proceeded in the absence of CuI, and an even cleaner reaction

proﬁle was obtained in this case.²⁸Heating 56 in a mixture of

DMF and aqueous HCl promoted decarboxylative decom-

position of the Meldrum’s acid moiety and gave the crystalline

acid 57 in 95% yield. These two steps were conveniently

telescoped into a one-pot process. Thus, upon completion of

the conjugate addition, the reaction was quenched with

aqueous HCl, DMF was added, and the mixture was heated

to 100 °C with concomitant distillation of THF and other

volatiles. On completion of the hydrolysis/decarboxylation, the

cooled reaction mixture was treated with aqueous HCl to eﬀect

the crystallization of 57 in 80% yield from 55.²⁹

The conversion of acid 57 into triﬂuoromethyl ketone 35

was accomplished as shown in Scheme 20. We have previously

described the direct conversion of enolizable carboxylic acids to

triﬂuoromethyl ketones by heating with TFAA and pyridine in

toluene, followed by hydrolysis/decarboxylation on addition of

water.³⁰This procedure is a variation of that developed by Zard

and co-workers for the conversion of acid chlorides to

triﬂuoromethyl ketones.³¹Thus, treatment of acid 57 with

TFAA (3 equiv) and pyridine (4.5 equiv) in toluene at 65 °C

for 5 h followed by addition of water and heating an additional

1 h provided triﬂuoromethyl ketone 35 in 83% yield after

extractive workup. Importantly, 35 generated from this reaction

was free of the impurities generated by the previous CuI

catalyzed conjugate addition to enone 28, and consequently did

not require puriﬁcation by vacuum distillation. A concentrated

toluene solution was used directly in the next step. The

mechanism of the reaction likely proceeds via formation of

pyridinium enolate 58, which is triﬂuoroacetylated to give 59.

Subsequent addition of water eﬀects hydrolysis to give a β-

triﬂuoroacetyl carboxylic acid, which decarboxylates to give the

product 35.

The Meldrum’s acid based synthesis of 35 avoided the use of

triﬂuoromethyl enone 28, removed two product distillations

from the synthesis, and employed cheaper reagents. The next

challenge was to avoid the use of isocyanate 46 in the

preparation of amide ketone 22. To accomplish this, we

performed the same enolization/Grignard exchange reaction,

but quenched the intermediate aryl Grignard reagent with CO₂

instead of isocyanate 46 (Scheme 21). This produced the acid/

lactol 60a/60b in 78% yield. Compound 60 existed as a ∼64:36

mixture of open (60a) and closed (60b) forms by NMR in d-6

DMSO at rt. The acid/lactol 60 was converted to its acid

chloride by treatment with SOCl₂in PhMe at 50 °C, and the

acid chloride solution was then added to a mixture of 2,6-

lutidine and amine 40 to give amide ketone 22 in 75% yield.

With the cost, safety and scalability issues for the synthesis of

amide ketone 22 having been addressed in the above second

generation synthesis, the next challenge was the development of

an asymmetric propargylation of 22. This was accomplished via

the use of propargyl borolane 61,³²diethylzinc, and the chiral

ligand N-isopropyl-^L-proline 62 in THF at 20 °C to give a

reaction diastereoselectivity of approximately 4:1, from which

the desired diastereomer 23 was isolated in 75% yield with a

diastereomeric purity of 99:1 after crystallization (Scheme 22).

The complete story of the development and scope of this

asymmetric propargylation reaction is the subject of the

following paper in this journal.³³

The ﬁnal issue to be addressed for a large scale process was

the amide deprotection step. While the HBr/AcOH procedure

was eﬀective, several aspects required improvement. First, the

process resulted in the formation of a polymer, poly(4-

methoxy)styrene, from the chiral 4-methoxyphenethyl group.

This polymer was eﬀectively solubilized by the toluene added

during crystallization, but its presence still posed concerns for

Scheme 19. Synthesis of Acid 57

K

dx.doi.org/10.1021/jo400079z | J. Org. Chem. XXXX, XXX, XXX−XXX

The Journal of Organic Chemistry

Article

Scheme 21. Synthesis of Amide Ketone 22 via Carboxylation/Coupling

Scheme 22. Asymmetric Propargylation Reaction of 22

Scheme 23. Deprotection of 24 to 1·PhOMe using H₃PO₄/Anisole

quantiﬁcation and quality control. Second, the level of

impurities formed in the reaction was high, and the need for

a recrystallization of the wet cake to achieve adequate purity

added to cycle time and decreased the overall yield. Finally,

because HBr forms a salt with 1, addition of an equal amount of

NaOH was necessary to neutralize the HBr, and this resulted in

a large amount of salt in the reaction which in turn led to a high

inorganic content in the isolated product (ROI of up to 0.5%).

With these issues in mind, a more extensive screen of reaction

conditions was undertaken. The use of anisole as a cosolvent

was employed, with the aim of trapping the 4-methoxyphe-

nethyl carbocation and preventing polymer formation. It was

found that 85% aqueous H₃PO₄and anisole at 100 °C for 1 h

gave a clean and complete conversion to 1 (Scheme 23).

Importantly, the stoichiometric byproducts 1,1-di(4-anisyl)-

ethane 63 and the corresponding ortho isomer 64, formed by

trapping of the 4-methoxyphenethyl cation with anisole, were

observed in a ∼9:1 ratio. The isolation of the pure para isomer

63 from the reaction in 90% yield indicated that polymer

formation was minimal under the reaction conditions. The

isolation of the product was accomplished simply by addition of

MEK followed by water, which caused direct crystallization of

the product as a 1:1 solvate with anisole. Importantly, the

higher pK_aof H₃PO₄(2.15) meant that it did not form a salt

L

dx.doi.org/10.1021/jo400079z | J. Org. Chem. XXXX, XXX, XXX−XXX

The Journal of Organic Chemistry

Article

Scheme 24. Structures and Levels of Impurities Formed in the Deprotection of 24 to 1·H₃PO₄

a

HPLC area% of impurity peak at 220 nm in isolated product.

b

Levels in solid before recrystallization.

Scheme 25. Second Generation Synthesis of 1·H₃PO₄

with 1 (pK_a= 1.75), and therefore no neutralization was

necessary. Consequently, the product did not have any residual

inorganic content (ROI <0.1%). After ﬁltration, the product

was washed with water and isopropanol. The product

1·PhOMe was obtained in 96% yield and with a purity of

>98.5%. The high purity of the product obtained under these

conditions eliminated the need for a recrystallization to upgrade

purity.

The major impurities formed in the deprotection reaction are

shown in Scheme 24. The acid impurity 65 was formed at a

relatively low level with the HBr/AcOH procedure, but proved

very diﬃcult to remove by recrystallization. Fortunately, the

H₃PO₄/anisole procedure rendered this impurity undetectable.

The lactone impurity 66 and the alkylation impurity 67 were

formed in large amounts in the HBr/AcOH procedure, but

these were also greatly reduced in the H₃PO₄/anisole

procedure.

M

dx.doi.org/10.1021/jo400079z | J. Org. Chem. XXXX, XXX, XXX−XXX

The Journal of Organic Chemistry

Article

spectroscopy was performed on a TOF instrument with ESI and

positive ionization. Flash chromatography was performed on an

automated system with silica columns. Compounds 25−27, 30−33,

36, 39, 40, 46, and 54 are commercially available. The following

compounds have previously been described in the literature: 20,²

28,^1,2,1061,^31aand 62.³²

The complete second generation synthesis of 1·H₃PO₄is

shown in Scheme 25. The synthesis required 8 chemical steps

plus the ﬁnal cocrystal formation, and proceeded in an overall

yield of 17.6%. This was more than double the yield of the ﬁrst

generation synthesis (6.8%), primarily due to the development

and implementation of a diastereoselective propargylation

reaction. The synthesis addressed the key cost, safety and

scalability issues and enabled the production of 94 kg of

1·H₃PO₄to support development work.

1,1,1-Triﬂuoro-4-methylpent-3-en-2-one (28). A reactor was

charged with 2-methyl-1-propenylmagnesium bromide 27 (34.75 kg,

36.5 L, 18.25 mol, 1.1 equiv, 0.5 M in THF). The solution was cooled

to about −10 °C. N-triﬂuoroacetylmorpholine 26 (3.038 kg, 16.59

mol. 1.0 equiv) was charged at a rate to maintain the batch

temperature at not more than 5 °C. The batch was warmed to about

15 °C and held at this temperature for about 1h to complete the

addition reaction. The batch was then cooled to about −10 °C and

treated with concentrated HCl (4.56 L) at a rate to maintain the batch

temperature at not more than 20 °C. Water (13.7 L) and dodecane

(7.6 L) were charged and the batch was stirred for 10 min, and then

the layers were allowed to settle. The lower aqueous phase was

separated. The organic phase was then washed four times with a

mixture of water (14.8 L) and MeOH (3.6 L). The organic phase was

washed with water (15.2 L), and then was drained from the reactor to

yield 9.52 kg of crude product solution, which contained 1.77 kg of 28

by assay (70% yield). This was puriﬁed by thin ﬁlm distillation using a

Pope thin-ﬁlm distillation apparatus. Vacuum of 110−120 mmHg and

column set point of 155 °C was used for the ﬁrst pass. The heavy

material obtained after one pass was passed through a second time at a

column temperature of 190 °C to drive oﬀ the remaining 28. The

receiver ﬂask was cooled in a −50 °C bath to prevent loss of product

to the vacuum line. The feed rate of distillate was 3L/hour. A total of

3.35 kg of distillate was obtained which was 49.0 wt% 28, thus 1.64 kg

28 (65% yield) as a light orange liquid. ¹H NMR (400 MHz, CDCl₃) δ

6.29 (m, 1 H), 2.24 (d, J = 1.1 Hz, 3 H), 2.02 (d, J = 1.1 Hz, 3 H); ¹³C

NMR (100 MHz, CDCl₃) δ 179.3 (q, J = 33.6 Hz), 168.5, 116.1 (q, J

= 290 Hz), 115.5, 28.4, 21.9.

CONCLUSION

■

In summary, we have described development of a large scale

process for the synthesis of glucocorticoid agonist BI 653048

BS H₃PO₄(1·H₃PO₄). The key concept for the route was the

propargylation of a triﬂuoromethyl ketone (22) bearing a chiral

4-(methoxyphenyl)ethyl amide to set the tertiary alcohol

stereocenter. The bromo triﬂuoromethyl ketone 35 was

prepared initially by a copper catalyzed conjugate addition

reaction of an ortho-bromo aryl Grignard reagent to

triﬂuoromethyl enone 28. Subsequently, a synthesis based on

a copper-free conjugate addition of the same aryl Grignard

reagent to isopropylidene Meldrum’s acid deriviative 55 was

employed. The Meldrum’s adduct was in situ converted to the

corresponding carboxylic acid 57, which was subsequently

converted to triﬂuoromethyl ketone 35 by a modiﬁcation of the

Zard procedure.^30,31The amide ketone 22 was prepared by a

novel enolization/Grignard exchange/trapping reaction. The

highly electrophilic triﬂuoromethyl ketone was “protected” in

situ as its sodium enolate, which enabled the functionalization

of the aryl bromide into the requisite amide via Grignard

exchange and electrophilic trapping, either with an isocyanate,

or with carbon dioxide. The propargylation was accomplished

initially by deprotonation of trimethylsilylpropyne with n-BuLi,

transmetalation to ZnBr₂, and addition of the resultant allenyl

zinc reagent to ketone 22. While this procedure gave the

product with 1:1 diastereoselectivity, the desired diastereomer

23 could be crystallized from the 1:1 mixture in 31−33% yield

with a diastereomeric purity of 98:2. Subsequently, a novel

diastereoselective version of the propargylation reaction was

developed which formed the desired diastereomer in 4:1

selectivity. After crystallization, the desired diastereomer was

isolated in 70% yield with a diastereomeric purity of 99:1. The

azaindole moiety was introduced by a one-pot Pd-catalyzed

cross coupling of iodopyridine 20 with alkyne 23 and

subsequent indolization. Finally, the deprotection of the 4-

methoxyphenethyl group was eﬀected initially with aqueous

HBr and AcOH. Subsequently, an improved procedure using

H₃PO₄and anisole was developed. This process provided the

product 1·PhOMe in high yield and purity, and avoided the

formation of polymers from the 4-methoxyphenethyl cation by

trapping with anisole to form the discrete adducts 63 and 64.

The phosphoric acid cocrystal ﬁnal form 1·H₃PO₄was

generated in high yield with consistently small particle size.

The diastereoselective propargylation route to 1·H₃PO₄was

employed to prepare >94 kg of drug to support development

work and clinical trials.

(S)-2-Bromo-5-ﬂuoro-N-(1-phenylethyl)benzamide (29). A

solution of acid 31 (5.0 g, 22.8 mmol, 1 equiv) in THF (50 mL)

was treated portionwise at rt with CDI (4.07 g, 25.1 mmol, 1.1 equiv).

The mixture was stirred at rt for 30 min, and then treated with (S)-1-

phenylethylamine (3.48 mL, 27.36 mmol, 1.2 equiv). The reaction

mixture was stirred at rt for 1h. Water (100 mL) was added, and the

mixture was cooled in an ice−water bath. After 1 h, the resultant solid

was ﬁltered, washed with water and heptane, and dried under vacuum

at 40 °C to give 29 (4.98 g, 68% yield) as a white solid. mp 127−130

1

°C; H NMR (400 MHz, CDCl₃) δ 8.98 (d, J = 7.9 Hz, 1 H), 7.70

(dd, J = 5.2, 8.8 Hz, 1 H), 7.45−7.42 (m, 2 H), 7.37−7.23 (m, 5 H),

5.12 (p, J = 7.2 Hz, 1 H), 1.46 (d, J = 6.9 Hz, 1 H); ¹³C NMR (100

MHz, CDCl₃) δ 165.1 (d, J = 1.7 Hz), 162.2, 159.7, 144.1, 140.8 (d, J

= 6.9 Hz), 134.5 (d, J = 8.0 Hz), 128.2, 126.7, 126.1, 117.8 (d, J = 22.4

Hz), 115.9 (d, J = 24.2 Hz), 113.7 (d, J = 3.2 Hz), 48.5, 22.4; HRMS

calcd for C₁₅H₁₄BrFNO [M + H]: 322.0237. Found: 322.0214.

4-(2-Bromo-4-ﬂuorophenyl)-1,1,1-triﬂuoro-4-methylpentan-

2-one (35). A reactor was charged with THF (<500 ppm H₂O, 13.40

kg) and 2-bromo-4-ﬂuoro-1-iodobenzene 33 (4.34 kg, 14.4 mol) and

the system was ﬂushed with nitrogen. The solution was cooled to −30

°C. i-PrMgCl (2.0 M in THF, 7.38 kg, 15.1 mol) was charged to the

reactor at a rate to maintain the batch temperature at not more than

−25 °C. The reaction mixture was stirred for 30 min at −30 to −25

°C. CuI (274 g, 1.44 mol) was charged as a slurry in THF (1.3 kg) and

the reaction was stirred for 15 min at −30 to −25 °C. Then 28 (46.9

wt%, 4.67 kg, 14.4 mol) was charged to the reaction at a rate to

maintain the batch temperature at not more than −25 °C. The batch is

stirred at −30 to −25 °C for about 10 min, and then warmed to about

−20 °C and held at this temperature for about 4 h. A 23 wt% aqueous

NH₄Cl solution (26.28 kg) was charged followed by EtOAc (8.57 kg)

and the batch was warmed to about 20 °C and stirred at this

temperature for at least 4 h. The layers were separated, and the organic

phase was washed successively with 23 wt% aqueous NH₄Cl solution

(10.26 kg) and 10 wt% aqueous NaCl solution (10.16 kg). The

organic phase was distilled to a volume of ∼9 L, discharged, and

further distilled under vacuum (6−14 Torr, T_vap= 105−123 °C and

EXPERIMENTAL SECTION

■

General Information. All starting materials and reagents were

purchased from commercial sources and used as received unless

otherwise noted. Melting points are given for crystalline solids and are

uncorrected. All ¹H and ¹³C NMR data were referenced to the internal

deuterated solvent relative to TMS at 0 ppm. High resolution mass

N

dx.doi.org/10.1021/jo400079z | J. Org. Chem. XXXX, XXX, XXX−XXX

The Journal of Organic Chemistry

Article

T_bath= 140 − 155 °C) to provide 35 as an orange oil (4.17 kg, 85.6 wt

66.1, 65.8, 55.3, 48.8, 40.9, 36.7, 32.5, 32.0, 21.3; HRMS calcd for

C₂₄H₂₈F₄NO₄[M + H]: 470.1949. Found: 470.1922.

1

% purity, 75% yield). bp ∼250 °C at 760 Torr. H NMR (400 MHz,

(S)-5-Fluoro-N-(1-(4-methoxyphenyl)ethyl)-2-(5,5,5-tri-

ﬂuoro-2-methyl-4-oxopentan-2-yl)benzamide (22) from 35. A

reactor was charged with NaH (0.690 kg, 17.26 mol, 60 wt% in

mineral oil) and THF (18.5 L, 500 ppm water). The resultant slurry

was cooled to about 0 °C. A solution of 35 (5.25 kg, 14.38 mol, 89.6

wt%) in THF (2.6 L) was added at a rate to control the temperature at

not more than 10 °C. The batch was then warmed to about 25 °C and

stirred at this temperature for 18 h. Completion of the enolization was

conﬁrmed by analysis by IR. The batch was cooled to about 0 °C and

treated with i-PrMgCl-LiCl (11.66 kg, 1.33 M in THF, 15.82 mol) at a

rate that the internal temperature does not exceed 20 °C. 1,4-Dioxane

(4.0 L) was charged, and the batch was stirred at about 25 °C for 2−4

h until the bromine/magnesium exchange had reached >99%

conversion by GC analysis of aliquots. The batch was cooled to

about 0 °C and treated with a solution of isocyanate 46 (2.80 kg, 15.82

mol) in THF (2.6 L) at a rate to control the temperature at not more

than 15 °C. The batch was stirred at 5−15 °C for 30 min. A solution

of concentrated HCl (5.3 L) in water (15.8 L) was added at a rate to

control the temperature at not more than 30 °C. Toluene was charged

(10.6 L), the batch was stirred at about 25 °C for 15 min, and the

aqueous phase was separated. The organic phase was washed with a

solution of NaCl (2.64 kg) in water (21 L). The organic phase was

then distilled at about 75 °C under vacuum to a volume of ∼10 L.

Toluene (21 L) was charged, and the batch was again distilled at about

75 °C under vacuum to a volume of ∼10 L. Heptane (26 L) was

charged at 60−75 °C followed by water (5.3 L). The batch was stirred

at about 70−75 °C for 30 min, cooled linearly to 5 °C over 2h, and

held at 5 °C for 2h. The batch was ﬁltered, and the solid was washed

with heptane (7.5 L). The solid was dried under vacuum with a

nitrogen stream at about 50 °C for 12h. 22 was obtained as a white

solid (5.33 kg, 97.6 wt% purity, 85% yield). Chiral HPLC (Chiralpak

IA column, 225 nm detection, ﬂow rate 1 mL/min, isocratic 90:10 v/v

hexanes/i-PrOH, 10 min): 22 (7.1 min), 99.9%; enantiomer-22 (6.2

min), 0.1%; mp 83−86 °C; ¹H NMR (400 MHz, CDCl₃) δ 7.42 (dd, J

= 5.4, 8.9 Hz, 1 H), 7.30−7.24 (m, 2 H), 7.02 (m, 1 H), 6.91−6.86 (m,

3 H), 6.09 (d, J = 7.9 Hz, 1 H), 5.19 (p, J = 6.9 Hz, 1 H), 3.80 (s, 3 H),

3.43 (q, J = 16.6 Hz, 2 H), 1.55 (d, J = 6.9 Hz, 3 H), 1.48 (s, 3 H), 1.44

(s, 3 H); ¹³C NMR (100 MHz, CDCl₃) δ 190.4 (q, J = 34 Hz), 170.0

(d, J = 1.5 Hz), 161.8, 159.2 (d, J = 25 Hz), 139.7 (d, J = 3.8 Hz),

138.0 (d, J = 5.8 Hz), 134.4, 129.8 (d, J = 7.5 Hz), 127.5, 116.0 (d, J =

20 Hz), 115.5 (d, J = 22 Hz), 115.2 (d, J = 290 Hz), 114.2, 55.3, 49.0,

48.4, 37.4, 29.4, 29.3, 21.2; HRMS calcd for C₂₂H₂₄F₄NO₃[M + H]:

426.1687. Found: 426.1663.

5-Fluoro-2-((S)-4-hydroxy-2-methyl-4-(triﬂuoromethyl)hept-

6-yn-2-yl)-N-((S)-1-(4-methoxyphenyl)ethyl)benzamide (23). A

solution of 1-trimethylsilylpropyne (27.9 mL, 188.4 mmol, 1.6 equiv)

and THF (210 mL) was cooled to about −25 °C. n-BuLi (70.5 mL,

176.3 mmol, 1.5 equiv, 2.5 M/hexanes) was charged at a rate to

control the temperature at not more than −15 °C. The batch was

stirred at about −20 °C for about 1 h. A solution of ZnBr₂in THF

(116.8 g of 24.9 wt% ZnBr₂in THF, 129.2 mmol, 1.1 equiv) was

charged at a rate to control the temperature at not more than −15 °C.

The batch was stirred at about −20 °C for about 1 h. A solution of 22

(52.6 g, 95.0 wt%, 117.53 mmol, 1.0 equiv) in THF (105 mL) was

charged at a rate to control the temperature at not more than −15 °C.

The batch was stirred at about −20 °C for about 1 h. The reaction was

then quenched with 3N aqueous HCl (80 mL) at a rate to control the

temperature at not more than 20 °C. Additional water (30 mL) was

charged. After stirring for about 10 min, the aqueous phase was

separated. Water (110 mL) was charged, and after stirring for about 10

min, the aqueous phase was separated. Twenty-ﬁve wt% NaOMe in

MeOH (79.0 mL, 345.7 mmol, 2.94 equiv) was charged at a rate to

control the temperature between 20 and 30 °C. The batch was stirred

at about 25 °C for about 1 h. The reaction was then quenched with 3N

aqueous HCl (84 mL) at a rate to control the temperature at not more

than 30 °C. The pH of the reaction mixture was ∼7. The batch was

then distilled under vacuum at up to 70−75 °C to remove THF,

MeOH and hexanes. Approximately 415 mL of distillate was collected.

CDCl₃) δ 7.46 (dd, J = 6.1, 8.9 Hz, 1 H), 7.29 (dd, J = 2.8, 8.2 Hz, 1

H), 7.02−6.97 (m, 1 H), 3.63 (s, 2 H), 1.58 (s, 6 H); ¹³C NMR (100

MHz, CDCl₃) δ 189.4 (q, J = 34.1 Hz), 162.0, 159.5, 139.8 (d, J = 3.7

Hz), 130.2 (d, J = 7.8 Hz), 122.4 (d, J = 24 Hz), 121.1 (d, J = 8.4 Hz),

115.3 (q, J = 290 Hz), 114.3 (d, J = 20.0 Hz), 45.4, 37.8, 28.7. HRMS

calcd for C₁₂H₁₀BrF₄O [M − H]: 324.9857. Found: 324.9843.

The following 4 compounds were isolated by chromatography on

SiO₂of nonproduct fractions from the distillation of 35:

1-Bromo-3-ﬂuorobenzene (36). Yellow liquid; ¹H NMR (400

MHz, CDCl₃) δ 7.26−7.20 (m, 2 H), 7.19−7.13 (m, 1 H), 7.00−6.95

(m, 1 H); ¹³C NMR (100 MHz, CDCl₃) δ 162.8 (d, J = 250 Hz),

131.0 (d, J = 8.4 Hz), 127.4 (d, J = 3.1 Hz), 122.7 (d, J = 9.5 Hz),

119.2 (d, J = 24.3 Hz), 114.3 (d, J = 20.8 Hz).

2-(2-Bromo-4-ﬂuorophenyl)-1,1,1-triﬂuoro-4-methylpent-3-en-2-

ol (37). Colorless oil; ¹H NMR (400 MHz, CDCl₃) δ 7.99 (dd, J = 6.3,

9.0 Hz, 1 H), 7.35 (dd, J = 2.6, 8.2 Hz, 1 H), 7.10−7.06 (m, 1 H), 6.04

(s, 1 H), 2.94 (s, 1 H), 1.80 (d, J = 1.4 Hz, 3 H), 1.38 (d, J = 1.4 Hz, 3

H); ¹³C NMR (100 MHz, CDCl₃) δ 162.0 (d, J = 249 Hz), 141.6,

132.1 (d, J = 8.4 Hz), 122.6, 122.0 (d, J = 24 Hz), 114.2 (d, J = 20 Hz),

76.0 (q, J = 29 Hz), 26.1, 19.1; HRMS calcd for C₁₃H₁₂BrF₄O₃[M +

HCO₂]: 370.9911. Found: 370.9897.

2,2′-Dibromo-4,4′-diﬂuorobiphenyl (38). Waxy solid; ¹H NMR

(400 MHz, CDCl₃) δ 7.41 (dd, J = 2.6, 8.4 Hz, 2 H), 7.20 (dd, J = 5.9,

8.2 Hz, 2 H), 7.12−7.07 (m, 2 H); ¹³C NMR (100 MHz, CDCl₃) δ

163.3, 160.8, 137.2 (d, J = 3.7 Hz), 132.1 (d, J = 8.5 Hz), 124.0 (d, J =

9.5 Hz), 119.9 (d, J = 5.8 Hz), 114.5 (d, J = 21.3 Hz). HRMS calcd for

C₁₂H₆BrF₂[M − HBr + H]: 266.9615. Found: 266.9603.

1

2-Bromo-4-ﬂuorophenol (39). White solid; H NMR (400 MHz,

CDCl₃) δ 7.20−7.18 (m, 1 H), 6.98−6.91 (m, 2 H), 5.42 (br s, 1 H);

¹³C NMR (100 MHz, CDCl₃) δ 157.6, 155.2, 148.8 (d, J = 2.6 Hz),

118.7 (d, J = 25.9 Hz), 116.4 (d, J = 8.1 Hz), 115.9 (d, J = 22.7 Hz),

109.5 (d, J = 10.2 Hz).

2-(2-(2-Bromo-4-ﬂuorophenyl)-2-methylpropyl)-2-(triﬂuoro-

methyl)-1,3-dioxolane (43). To a solution of 35 (3.00 g, 9.17

mmol) in DMF (40 mL) was added 2-chloroethanol (1.84 mL, 27.51

mmol). The reaction mixture was treated with K₂CO₃(3.80 g, 27.51

mmol) and stirred at rt overnight. The reaction mixture was ﬁltered

and the solid washed with EtOAc. The ﬁltrate was concentrated,

diluted with EtOAc, and the organic phase washed with water (3 × 50

mL) and brine (50 mL), dried (MgSO₄), ﬁltered, and concentrated to

give 43 (3.30 g, 87.6 wt% purity, 95% yield) as a yellow oil. ¹H NMR

(400 MHz, CDCl₃) δ 7.40 (dd, J = 6.0, 9.0 Hz, 1 H), 7.33 (dd, J = 2.8,

8.3 Hz, 1 H), 6.96−6.91 (m, 1 H), 3.89−3.79 (m, 2 H), 3.31−3.22 (m,

2 H), 2.71 (br s, 2 H), 1.59 (s, 6 H); ¹³C NMR (100 MHz, CDCl₃) δ

161.5, 159.0, 141.2, 129.6 (d, J = 7.7 Hz), 123.6 (d, J = 8.7 Hz), 123.5

(q, J = 293 Hz), 122.3 (d, J = 23.7 Hz), 113.2 (d, J = 19.4 Hz), 107.0

(q, J = 30.4 Hz), 65.9, 38.0, 30.5; Repeated attempts to obtain HRMS

for this compound were unsuccessful.

(S)-5-Fluoro-N-(1-(4-methoxyphenyl)ethyl)-2-(2-methyl-1-

(2-(triﬂuoromethyl)-1,3-dioxolan-2-yl)propan-2-yl)benzamide

(44). A mixture of Pd(OAc)₂(121 mg, 0.54 mmol), BINAP (1.01 g,

1.62 mmol), K₂CO₃(2.98 g, 21.55 mmol), (S)-1-(4-methoxyphenyl)-

ethanamine 40 (3.26 g, 21.55 mmol), 43 (4.00 g, 10.78 mmol) and

toluene (40 mL) was heated at 110 °C under 250 psi of CO for 20 h.

The reaction mixture was ﬁltered through Celite, and the ﬁltrate was

washed with 1 M aqueous HCl (50 mL) and water (50 mL), dried

(MgSO₄), ﬁltered, and concentrated. The crude product was puriﬁed

by chromatography on SiO₂(CH₂Cl₂/EtOAc) to give 44 as a light-

1

brown oil (3.86 g, 76% yield). H NMR (400 MHz, CDCl₃) δ 7.40−

7.35 (m, 1 H), 7.33−7.29 (m, 2 H), 7.14−7.08 (m, 1 H), 6.98−6.93

(m, 2 H), 6.91−6.87 (m, 2 H), 5.32−5.25 (m, 1 H), 3.91−3.85 (m, 1

H), 3.83−3.77 (m, 1 H), 3.81 (s, 3 H), 3.27−3.21 (m, 1 H), 2.94−2.88

(m, 1 H), 2.46 (d, J = 15.8 Hz, 1 H), 2.24 (d, J = 15.8 Hz, 1 H), 1.56

(d, J = 6.8 Hz, 3 H), 1.48 (s, 3 H), 1.45 (s, 3 H); ¹³C NMR (100 MHz,

CDCl₃) δ 169.8, 161.7, 159.3, 159.0, 140.4 (d, J = 5.9 Hz), 138.2 (d, J

= 3.8 Hz), 134.8, 130.1 (d, J = 7.4 Hz), 127.6, 123.2 (q, J = 291 Hz),

115.9 (d, J = 22 Hz), 114.5 (d, J = 20 Hz), 114.1, 107.0 (q, J = 30 Hz),

O

dx.doi.org/10.1021/jo400079z | J. Org. Chem. XXXX, XXX, XXX−XXX

The Journal of Organic Chemistry

Article

i-PrOAc (300 mL) was charged and the batch was cooled to 20−25

°C. The batch was then treated with 3N aqueous HCl (60 mL). After

stirring for about 20 min, the aqueous phase was separated. Water

(130 mL) was charged, and after stirring for about 10 min, the

aqueous phase was separated. The batch was then distilled under

vacuum at up to 70−75 °C until 230 mL of distillate was collected.

The batch was assayed and adjusted to achieve a concentration of 23 +

49 in i-PrOAc of 1g/2.4g. The batch was cooled to about 20−25 °C.

Seed crystals of 23 (175 mg) were charged as a slurry in heptane (1

mL), and the batch was stirred for about 30 min at 20−25 °C.

Heptane (268 mL) was then added over 1h at 20−25 °C. The batch

was stirred for 15h at 20−25 °C and ﬁltered. The solid was washed

with 1:3 v/v i-PrOAc/heptane (2 × 16 mL), and the solid was dried at

25−35 °C. 23 was obtained as a oﬀ-white solid (18.6 g, 97.0 wt%

obtained as an oﬀ-white solid (36.725 kg, 92.9 wt% purity, 99.4 area%

purity by HPLC, 82% yield). Residual Pd: 8.2 ppm. mp 223−225 °C;

¹H NMR (400 MHz, d-6 DMSO) δ 11.63 (s, 1 H), 9.22 (d, J = 7.6 Hz,

1 H), 8.79 (s, 1 H), 8.17 (s, 1 H), 7.60−7.56 (m, 1 H), 7.31 (d, J = 7.6

Hz, 1 H), 7.21−7.15 (m, 1 H), 7.00 (d, J = 8.8 Hz, 1 H), 6.92 (d, J =

8.1 Hz, 2 H), 6.69 (s, 1 H), 6.47 (s, 1 H), 5.10−5.01 (m, 1 H), 3.75 (s,

3 H), 3.37−3.31 (m, 2 H), 3.04 (d, J = 15.6 Hz, 1 H), 2.88 (d, J = 15.6

Hz, 1 H), 2.46 (d, J = 15.6 Hz, 1 H), 2.22 (d, J = 15.6 Hz, 1 H), 1.55

(s, 3 H), 1.44 (d, J = 6.2 Hz, 3 H), 1.36 (s, 3 H); ¹³C NMR (100 MHz,

d-6 DMSO) δ 170.3, 160.9, 158.4, 158.1, 144.3, 141.3, 140.8 (d, J = 3.3

Hz), 138.3 (d, J = 5.8 Hz), 135.8, 134.0 (d, J = 6.7 Hz), 131.8, 130.3

(d, J = 7.5 Hz), 127.3, 126.1 (q, J = 289 Hz), 115.7 (d, J = 22 Hz),

115.4 (d, J = 20 Hz), 114.3, 113.6, 102.9, 75.7 (q, J = 25 Hz), 55.1,

48.2, 46.3, 44.8, 37.5, 33.3, 32.8, 30.6, 21.8, 7.0; HRMS calcd for

C₃₂H₃₆F₄N₃O₅S [M + H]: 650.2306. Found: 650.2271.

1

purity, dr = 98.7: 1.3, 33.0% yield). mp 168−169 °C; H NMR (500

MHz, CDCl₃) δ 7.49 (dd, J = 5.1, 8.7 Hz, 1 H), 7.31 (d, J = 8.5 Hz, 2

H), 7.05 (ddd, J = 3.1, 8.5, 8.5 Hz, 1 H), 6.92 (d, J = 8.5 Hz, 2 H), 6.84

(dd, J = 3.4, 8.7 Hz, 1 H), 6.10 (d, J = 8.3 Hz, 1 H), 6.05 (s, 1 H), 5.29

(dddd, J = 7.1, 7.1, 7.1, 7.1 Hz, 1 H), 3.82 (s, 3 H), 2.56 (d, J = 15.3

Hz, 1 H), 2.51 (d, J = 18.5 Hz, 1 H), 2.40 (d, J = 17.8 Hz, 1 H), 2.29

(d, J = 15.3 Hz, 1 H), 2.07 (t, J = 2.1 Hz, 1 H), 1.62 (s, 3 H), 1.59 (d, J

= 6.8 Hz, 3 H), 1.42 (s, 3 H); ¹³C NMR (125 MHz, CDCl₃) δ 171.8

(d, J = 1.8 Hz), 160.5 (d, J = 249 Hz), 159.3, 140.5 (d, J = 3.5 Hz),

137.3 (d, J = 5.8 Hz), 133.9, 130.1 (d, J = 7.6 Hz), 127.5, 125.7 (q, J =

289 Hz), 116.5 (d, J = 19.4 Hz), 115.4 (d, J = 22.6 Hz), 114.3, 78.8,

75.4 (q, J = 26.7 Hz), 71.7, 55.3, 49.1, 42.8, 37.0, 33.3, 33.2, 23.9,

20.64; HRMS calcd for C₂₅H₂₈F₄NO₃[M + H]: 466.2000. Found:

466.2001.

tert-Butyl-6-(ethylsulfonyl)-4-((S)-6-(4-ﬂuoro-2-((S)-1-(4-

methoxyphenyl)ethylcarbamoyl)-phenyl)-4-hydroxy-6-meth-

yl-4-(triﬂuoromethyl)hept-1-ynyl)pyridin-3-ylcarbamate (52).

This intermediate was isolated by extractive workup with EtOAc and

water of the above reaction prior to addition of DBU. The organic

phase was dried, and the product isolated by chromatography on SiO₂.

1

Tan solid; mp 131−134 °C; H NMR (400 MHz, d-6 DMSO) δ

9.14−9.12 (m, 2 H), 8.55 (s, 1 H), 7.84 (s, 1 H), 7.59 (dd, J = 5.6, 9.0

Hz, 1 H), 7.31−7.27 (m, 2 H), 7.14−7.09 (m, 1 H), 6.91−6.88 (m, 3

H), 6.58 (s, 1 H), 5.02 (p, J = 7.5 Hz, 1 H), 3.74 (s, 3 H), 3.40 (q, J =

7.4 Hz, 2 H), 2.78 (m, 2 H), 2.61 (d, J = 15.3 Hz, 1 H), 2.14 (d, J =

15.3 Hz, 1 H), 1.50 (s, 3 H), 1.46 (s, 9 H), 1.39−1.37 (m, 6 H), 1.13

(t, J = 7.4 Hz, 3 H); ¹³C NMR (100 MHz, d-6 DMSO) δ 170.1, 160.9,

158.4, 158.1, 151.9, 149.5, 142.0, 140.9 (d, J = 3.4 Hz), 138.8, 138.6

(d, J = 5.8 Hz), 135.8, 130.4 (d, J = 7.2 Hz), 127.3, 124.1, 122.2, 115.5

(d, J = 22 Hz), 115.2 (d, J = 19 Hz), 113.6, 98.5, 81.0, 75.8, 75.1 (q, J =

25 Hz), 55.0, 48.1, 46.1, 43.1, 37.5, 32.8, 30.5, 27.7, 26.8, 25.1, 21.9,

6.7; HRMS calcd for C₃₇H₄₄F₄N₃O₇S [M + H]: 750.2831. Found:

750.2823.

(S)-5-Fluoro-2-(4-hydroxy-2-methyl-4-(triﬂuoromethyl)hept-

6-yn-2-yl)benzamide (19) from 23. A solution of amide 23 (2.00 g,

4.30 mmol) in TFA (20 mL) was heated at 50 °C for 16h. The

reaction mixture was concentrated, and the residue was diluted with

EtOAc and washed with saturated aqueous NaHCO₃solution and

water. The organic solution was dried (Na₂SO₄), ﬁltered, and

concentrated. Flash column chromatography on SiO₂(80:20 to

50:50 hexanes/EtOAc) gave pure 19 as an oﬀ-white solid (1.14 g, 80%

yield). mp 149−151 °C; ¹H NMR (400 MHz, d-6 DMSO) δ 8.37 (s, 1

H), 8.03 (s, 1 H), 7.57 (dd, J = 5.5, 9.0 Hz, 1 H), 7.20 (ddd, J = 2.9,

8.5, 8.5 Hz, 1 H), 7.08 (dd, J = 2.9, 8.8 Hz, 1 H), 6.71 (s, 1 H), 2.86

(br, 1 H), 2.49−2.37 (m, 3 H), 2.32−2.28 (m, 1 H), 1.61 (s, 3 H),

1.46 (s, 3 H); ¹³C NMR (100 MHz, d-6 DMSO) δ 174.6, 160.9, 158.5,

140.2 (d, J = 3.2 Hz), 138.5 (d, J = 6.2 Hz), 130.0 (d, J = 7.4 Hz),

125.9 (q, J = 288 Hz), 115.4 (d, J = 5.5 Hz), 115.2 (d, J = 8.1 Hz),

78.8, 74.9 (q, J = 25 Hz), 73.7, 42.4, 36.9, 32.6, 32.3, 23.7; HRMS

calcd for C₁₆H₁₈F₄NO₂[M + H]: 332.1268. Found: 332.1251.

tert-Butyl 6-(ethylsulfonyl)-4-iodopyridin-3-ylcarbamate

(R)-2-(4-((5-(Ethylsulfonyl)-1H-pyrrolo[2,3-c]pyridin-2-yl)-

methyl)-5,5,5-triﬂuoro-4-hydroxy-2-methylpentan-2-yl)-5-ﬂu-

orobenzamide acetic acid solvate (1·AcOH). A reactor was

charged with 24 (15.42 kg, 97.19 wt% purity, 23.73 mol, 1 equiv),

AcOH (65.2 kg) and 48% aqueous HBr (46.29 kg). The batch was

heated at about 80 °C for about 7h. Toluene (64.9 kg) was charged at

about 80 °C. The batch was cooled to about 25 °C, and a solution of

NaOH (11.0 kg, 275.0 mol) in water (75.0 kg) was added at a rate to

control the temperature at not more than 70 °C. The batch was heated

to about 75 °C, and a slurry of 1·AcOH seed crystals (30.0 g) in

toluene (1.5 L) was charged. The batch was held at about 75 °C for 30

min, cooled to about 25 °C over 2 h, and held at about 25 °C for 3 h.

The batch was ﬁltered, and the solid was washed successively with

water (61.8 kg) and toluene (26.7 kg). The wet cake was recharged to

a clean reactor. AcOH (35.4 kg) was charged, and the mixture heated

to about 75 °C to obtain a solution. Toluene (80.0 kg) was added at

65−75 °C. A slurry of 1·AcOH seed crystals (15.0 g) in toluene (1.5

L) was charged. The batch was held at about 75 °C for 30 min, cooled

to about 25 °C over 2 h, and held at about 25 °C for 1 h. The batch

was ﬁltered, and the solid was washed with toluene (26.7 kg). The

solid was dried under vacuum at 70 °C with a nitrogen purge for 15 h.

1·AcOH was obtained as an oﬀ-white solid (9.16 kg, 86.6 wt% purity,

98.95 area% purity by HPLC, 67% yield). mp 133−135 °C; ¹H NMR

(400 MHz, d-6 DMSO) δ 11.97 (br s, 1 H), 11.59 (s, 1 H), 8.77 (s, 1

H), 8.34 (s, 1 H), 8.15 (s, 1 H), 7.99 (s, 1 H), 7.57 (dd, J = 5.6, 8.9 Hz,

1 H), 7.19−7.11 (m, 2 H), 6.83 (s, 1 H), 6.47 (s, 1 H), 3.34 (q, J = 7.5

Hz, 2 H), 3.05 (d, J = 15.1 Hz, 1 H), 2.93 (d, J = 15.1 Hz, 1 H), 2.49

(d, J = 15.1 Hz, 1 H), 2.34 (d, J = 15.1 Hz, 1 H), 1.92 (s, 3 H), 1.57 (s,

3 H), 1.55 (s, 3 H), 1.07 (t, J = 7.4 Hz, 3 H); ¹³C NMR (100 MHz, d-

6 DMSO) δ 174.2, 172.0, 160.9, 158.5, 144.4, 141.4, 140.2 (d, J = 3.4

Hz), 138.6 (d, J = 5.9 Hz), 134.0 (d, J = 6.2 Hz), 131.8, 130.2 (d, J =

7.4 Hz), 126.1 (q, J = 289 Hz), 115.4 (t, J = 22 Hz), 114.2, 102.8, 75.7

(q, J = 25 Hz), 46.2, 44.6, 37.5, 33.3, 32.6, 31.2, 21.0, 6.9; HRMS calcd

for C₂₃H₂₆F₄N₃O₄S [M − AcOH + H]: 516.1575. Found: 516.1547.

(R)-2-(4-((5-(Ethylsulfonyl)-1H-pyrrolo[2,3-c]pyridin-2-yl)-

methyl)-5,5,5-triﬂuoro-4-hydroxy-2-methylpentan-2-yl)-5-ﬂu-

1

(20). White solid; mp 114−115 °C; H NMR (400 MHz, CDCl₃) δ

9.35 (s, 1 H), 8.42 (s, 1 H), 7.02 (br, 1 H), 3.39 (q, J = 7.4 Hz, 2 H),

1.56 (s, 9 H), 1.30 (t, J = 7.4 Hz, 3 H); ¹³C NMR (100 MHz, CDCl₃)

δ 151.4, 149.8, 140.4, 139.9, 132.7, 98.3, 83.0, 46.8, 28.2, 7.0; HRMS

calcd for C₁₂H₁₈IN₂O₄S [M + H]: 413.0027. Found: 413.0003.

2-((R)-4-((5-(Ethylsulfonyl)-1H-pyrrolo[2,3-c]pyridin-2-yl)-

methyl)-5,5,5-triﬂuoro-4-hydroxy-2-methylpentan-2-yl)-5-ﬂu-

oro-N-((S)-1-(4-methoxyphenyl)ethyl)benzamide (24). A reactor

was charged with 23 (30.75 kg, 97. Six wt%, 64.45 mol, 1 equiv), 20

(27.6 kg, 97.2 wt%, 65.09 mol, 1.01 equiv), and DABCO (14.82 kg,

132.12 mol, 2 equiv). The vessel was sealed and inerted, and then

MeOH (47.5 kg) was charged. A suspension of Pd(OAc)₂(72.0 g,

0.321 mol, 0.005 equiv) in MeOH (4.0 kg) was charged, and the batch

was heated at about 65 °C for 3 h to eﬀect complete cross coupling to

intermediate 52. The batch was cooled to about 50 °C, and DBU

(14.7 kg, 96.56 mol, 1.5 equiv) was charged at 50−55 °C. The batch

was stirred at about 50 °C for about 2 h to eﬀect complete cyclization

of 52 to 24. Acetonitrile (61.3 kg) was charged to the batch at 45−55

°C. Then water (34.5 kg) was charged at 45−55 °C over 30 min. The

batch was seeded with 24 seeds crystals (156 g), and held at about 50

°C for 30 min. Water (72 kg) was charged over 1h at about 50 °C, and

the batch was cooled over 2 h to about 25 °C, held at this temperature

for 1 h, and ﬁltered. The solid was washed with MTBE (198.4 kg) and

dried under vacuum at 65 °C with a nitrogen purge for 12 h. 24 was

P

dx.doi.org/10.1021/jo400079z | J. Org. Chem. XXXX, XXX, XXX−XXX

The Journal of Organic Chemistry

Article

orobenzamide phosphoric acid cocrystal (1·H₃PO₄) from AcOH

solvate 1·AcO₃H. A reactor was charged with 1·AcOH (1.60 kg, 90.5

wt% free base, 2.81 mol, 1 equiv) and MEK (9.6 L). The mixture was

heated to about 60 °C to obtain a solution. The warm solution was

polish ﬁltered into a clean reactor, using additional warmed (≥40 °C)

MEK (3.2 L) to rinse the ﬁlter. After adjusting the batch temperature

to 50 °C, H₃PO₄(334.7 g, 2.95 mol, 86.2 wt%, 1.05 equiv) was added

at about 50 °C. The batch was stirred for 20 min, and then heptane

(2.13 L) was added over about 20 min at about 50 °C. Seed crystals of

1·H₃PO₄(1.80 g) were added as a slurry in heptane (110 mL). The

batch was stirred for 30 min at about 50 °C while a slurry developed.

Then heptane (4.27 L) was added over 1h at about 50 °C. The batch

was cooled linearly over 2 h to about 20 °C and held at about 20 °C

for 2 h. The batch was ﬁltered, and the solid was washed with MEK/

heptane 1:2 v/v (2 × 3.3 L) followed by heptane (3.3 L). The solid

was dried under vacuum at 70 °C for 24 h. 1·H₃PO₄was obtained as a

white solid (1.61 kg, 99.9 area% purity by HPLC, 93% yield). Chiral

HPLC (Chiralpak IA column, 235 nm detection, ﬂow rate 2 mL/min,

isocratic 70:30 v/v heptane/i-PrOH, 10 min): 1·H₃PO₄(4.30 min),

99.95%; enantiomer-22 (3.01 min), 0.05%; mp 204−207 °C; ¹H NMR

(400 MHz, d-6 DMSO) δ 11.61 (s, 1 H), 9.58 (br, 3 H), 8.78 (s, 1 H),

8.35 (s, 1 H), 8.16 (s, 1 H), 8.00 (s, 1 H), 7.58 (dd, J = 5.6, 9.0 Hz, 1

H), 7.20−7.12 (m, 2 H), 6.48 (s, 1 H), 3,35 (q, J = 7.6 Hz, 2 H), 3.05

(d, J = 15.0 Hz, 1 H), 2.93 (d, J = 15.0 Hz, 1 H), 2.51 (d, J = 15.0 Hz,

1 H), 2.35 (d, J = 15.0 Hz, 1 H), 1.58 (s, 3 H), 1.55 (s, 3 H), 1.08 (t, J

= 7.3 Hz, 3 H); ¹³C NMR (100 MHz, d-6 DMSO) δ 174.2, 160.9,

158.5, 144.3, 141.4, 140.2 (d, J = 3.7 Hz), 138.6 (d, J = 5.8 Hz), 134.0

(d, J = 5.1 Hz), 131.8, 130.2 (d, J = 7.4 Hz), 126.1 (q, J = 289 Hz),

115.4 (t, J = 22 Hz), 114.3, 102.9, 75.7 (q, J = 25 Hz), 46.2, 44.6, 37.5,

33.3, 32.6, 31.2, 6.9; HRMS calcd for C₂₃H₂₆F₄N₃O₄S [M − H₃PO₄+

H]: 516.1575. Found: 516.1569.

6.91 (m, 1 H), 3.10 (s, 2 H), 1.57 (s, 6 H); ¹³C NMR (100 MHz,

CDCl₃) δ 178.4, 160.6 (d, J = 248 Hz), 140.5 (d, J = 3.7 Hz), 129.9 (d,

J = 7.8 Hz), 122.6 (d, J = 24 Hz), 121.9 (d, J = 8.6 Hz), 114.0 (d, J =

20 Hz), 44.0, 38.2, 28.7; HRMS calcd for C₁₁H₁₆BrFNO₂[M + NH₄]:

292.0343. Found: 292.0333.

5-(2-(2-Bromo-4-ﬂuorophenyl)propan-2-yl)-2,2-dimethyl-

1,3-dioxane-4,6-dione (56). This compound was isolated from the

above procedure by extractive workup with EtOAc and saturated

aqueous NH₄Cl after completion of the conjugate addition, and

subsequent crystallization from MTBE/hexanes. White solid; mp

116−119 °C; ¹H NMR (400 MHz, CDCl₃) δ 7.54 (dd, J = 6.1, 9.0 Hz,

1 H), 7.30 (dd, J = 2.8, 8.1 Hz, 1 H), 7.07−7.02 (m, 1 H), 5.61 (s, 1

H), 1.86 (s, 3 H), 1.77 (s, 6 H), 1.73 (s, 3 H); ¹³C NMR (100 MHz,

CDCl₃) δ 163.2, 160.6 (d, J = 249 Hz), 141.1 (d, J = 3.7 Hz), 130.5 (d,

J = 7.8 Hz), 122.1 (d, J = 23 Hz), 120.0 (d, J = 8.5 Hz), 114.6 (d, J =

20 Hz), 104.1, 52.0, 40.9, 28.7, 26.4, 25.9, 25.1; HRMS calcd for

C₁₅H₁₇BrFO₄[M + H]: 359.0289. Found: 359.0273.

4-(2-Bromo-4-ﬂuorophenyl)-1,1,1-triﬂuoro-4-methylpentan-

2-one (35) from 57. A reactor was charged with 57 (83.6 kg, 303.9

mol, 1 equiv), toluene (295 kg) and TFAA (191 kg, 911.7 mol, 3

equiv). The reaction mixture was cooled to about 0 °C, and pyridine

(108 kg, 1367.6 mol, 4.5 equiv) was charged at a rate to control the

batch temperature at not more than 35 °C. The reaction mixture was

heated to 60−65 °C and held at this temperature for about 12 h. The

batch was then cooled to about 5 °C, and water (337 kg) was charged

at a rate to control the batch temperature at not more than 50 °C. The

reaction mixture is then heated at about 50 °C for about 1h. The

reaction mixture was cooled to about 25 °C, and heptane (229 kg) was

charged. The aqueous phase was separated, and the batch was washed

again with water (334 kg). The batch was then distilled under vacuum

at up to 70 °C to remove 434 kg of distillate. After cooling to about 25

°C, the batch was treated with heptane (736 kg) and SiO₂(50.0 kg),

and was agitated for about 30 min. The batch was then ﬁltered, and the

SiO₂cake was washed with heptane (78 kg). The combined ﬁltrates

were returned to the cleaned reactor, and were distilled under vacuum

at up to 70 °C to remove 876 kg of solvent. The resultant

concentrated orange solution of 35 was drained from the reactor and

assayed (105.0 kg, 78.7 wt% purity, 83% yield). Spectral data for 35

were consistent with material obtained from conjugate addition to 28.

5-Fluoro-2-(5,5,5-triﬂuoro-2-methyl-4-oxopentan-2-yl)-

benzoic Acid (60). A reactor was charged with sodium hydride (12.1

kg, 60 wt% in mineral oil, 303.1 mol, 1.2 equiv), THF (184 kg, 300−

500 ppm water), and 1,4-dioxane (57 kg), and the slurry was cooled to

about 5 °C. A solution of 35 (105.0 kg, 78.7 wt% purity, 252.6 mol, 1

equiv) in THF (63 kg) was charged at a rate to control the evolution

of hydrogen gas. The vessel was rinsed with an additional 21 kg of

THF. The batch was stirred at about 25 °C for about 2h, at which

point IR analysis indicated complete enolization. The reaction mixture

was then cooled to about 5 °C, and i-PrMgCl-LiCl (213.0 kg, 1.30 M,

290.5 mol, 1.15 equiv) was charged at a rate to control the batch

temperature at not more than 20 °C. The batch was stirred at 20−25

°C for 5h to eﬀect complete Br−Mg exchange (monitored by GC of

MeOH-quenched aliquots), and was then cooled to about −15 °C.

CO₂gas was bubbled into the reaction mixture (subsurface addition)

at a rate to control the batch temperature at not more than 20 °C. A

total of 27.5 kg of CO₂was charged. The reaction mixture was stirred

at about 5−15 °C for 30 min. The batch was then cooled to about 0

°C, and a solution of concentrated HCl (103.0 kg) in water (259.0 kg)

was charged at a rate to control the batch temperature at not more

than 30 °C. The batch was then distilled under vacuum at up to 35 °C

to remove 464 kg distillate. Water (207 kg) was then charged to the

batch at about 30−35 °C. The batch was cooled to about 0 °C over 2

h, seeded with 60 (70 g), and held at about 0 °C for at least 2 h. The

batch was ﬁltered and the solid washed with water (274 kg). The solid

was dried under vacuum with a nitrogen stream at 25−30 °C until KF

≤ 0.5%. 60 is obtained as a tan solid (69.6 kg, 82.8 wt% purity, 78%

yield). Note: Compound 60 existed as a ∼65:35 mixture of keto-acid

60a and lactol 60b, derived from the acid ketalizing with the CF₃

2,2-Dimethyl-5-(propan-2-ylidene)-1,3-dioxane-4,6-dione

(55). A reactor was charged with Meldrum’s acid 54 (100.0 kg, 693.8

mol, 1 equiv) and acetone (500.0 kg). To the resultant solution was

added morpholine (1.05 kg, 12.1 mol, 0.018 equiv) followed by AcOH

(0.83 kg, 13.8 mol, 0.02 equiv). The reaction mixture was stirred at

about 25 °C for about 48 h. Methylcyclohexane (385 kg) was added,

and acetone was distilled out under vacuum at a temperature of not

more than 40 °C. MTBE (370 kg) was charged, and the organic phase

was washed quickly with 5% aqueous NaOH (2 × 50.0 kg). The

organic phase was then distilled under vacuum at not more than 45 °C

until the MTBE content was less than 5%. The slurry was cooled to

about 0 °C and held at this temperature for about 1 h. The batch was

ﬁltered, and the solid washed with cold (0 °C) methylcyclohexane

(30.0 kg) and dried under vacuum with a nitrogen stream at 25 °C to

1

give 55 as a white solid (101.0 kg, 79% yield). mp 122−125 °C; H

NMR (400 MHz, CDCl₃) δ 2.42 (s, 6 H), 1.63 (s, 6 H); ¹³C NMR

(100 MHz, CDCl₃) δ 177.2, 161.0, 115.9, 103.5, 27.1, 26.7.

3-(2-Bromo-4-ﬂuorophenyl)-3-methylbutanoic Acid (57). A

reactor was charged with 2-bromo-4-ﬂuoro-1-iodobenzene 33 (42.8

kg, 142.2 mol, 1.05 equiv) and THF (50 kg). The solution was cooled

to −20 °C and i-PrMgCl solution in THF (79.7 kg, 163.5 mol, 1.15

equiv, 2.0 M) was charged at a rate to control the batch temperature at

not more than 0 °C. The batch was stirred for 30 min at about −15

°C. A solution of 55 (25.0 kg, 135.7 mol, 1 equiv) in THF (39 kg) was

charged at a rate to control the batch temperature at not more than 5

°C. The batch was stirred for 2 h at about 0−10 °C to eﬀect complete

conversion to intermediate 56. The reaction mixture was quenched

with a solution of concentrated HCl (25 kg) in water (50 kg) at a rate

to control the batch temperature at not more than 25 °C. DMF (55

kg) was charged, and the batch was distilled at atmospheric pressure

up to 105 °C to remove THF and toluene. The batch was held at

about 105 °C for 15 h. The batch was cooled to about 25 °C and was

treated with a solution of concentrated HCl (25 kg) in water (50 kg).

The batch was cooled to about 0 °C and held at this temperature for 2

h. The solid was ﬁltered, washed with water (100 kg), and dried under

vacuum at about 50 °C until KF ≤ 0.2%. 57 is obtained as an oﬀ-white

solid (31.0 kg, 96.0 wt% purity, 98.3 area% purity by HPLC, 80%

yield). mp 88−90 °C; ¹H NMR (400 MHz, CDCl₃) δ 11.53 (br, 1 H),

7.36 (dd, J = 6.0, 9.0 Hz, 1 H), 7.30 (dd, J = 2.8, 8.3 Hz, 1 H), 6.95−

1

ketone. mp 97−101 °C; H NMR (400 MHz, d-6 DMSO) δ 13.93−

13.13 (br s, 0.65 H), 8.51−7.88 (br s, 0.35 H), 7.56−7.47 (m, 1 H),

Q

dx.doi.org/10.1021/jo400079z | J. Org. Chem. XXXX, XXX, XXX−XXX

The Journal of Organic Chemistry

Article

7.41−7.14 (m, 2 H), 3.60 (s, 1.3 H), 2.53−2.49 (m, 0.35 H), 2.18−

2.14 (m, 0.35 H), 1.48 (s, 4.95 H), 1.34 (s, 1.05 H); ¹³C NMR (100

MHz, d-6 DMSO) δ 189.6 (q, J = 33 Hz), 171.6, 166.2, 161.7, 161.0,

159.2, 158.6, 140.9, 139.5 (d, J = 3.4 Hz), 135.5 (d, J = 6.5 Hz), 133.8

(d, J = 6.7 Hz), 129.8 (d, J = 7.2 Hz), 127.7 (d, J = 7.0 Hz), 122.0 (q, J

= 287 Hz), 118.6 (d, J = 21 Hz), 117.0 (d, J = 24 Hz), 115.9 (d, J = 21

Hz), 115.0 (d, J = 24 Hz), 114.9 (q, J = 291 Hz), 97.9 (q, J = 32 Hz),

47.5, 45.7, 36.5, 35.3, 31.2, 30.2, 28.9; HRMS calcd for C₁₃H₁₆F₄NO₃

[M + NH₄]: 310.1061. Found: 310.1052.

4,4′-(Ethane-1,1-diyl)bis(methoxybenzene) (63). White solid;

1

mp 67−69 °C; H NMR (400 MHz, CDCl₃) δ 7.13−7.09 (m, 4 H),

6.83−6.79 (m, 4 H), 4.04 (q, J = 7.2 Hz, 1 H), 3.75 (s, 6 H), 1.57 (d, J

= 7.3 Hz, 3 H); ¹³C NMR (100 MHz, CDCl₃) δ 157.9, 139.1, 128.5,

113.8, 55.3, 43.2, 22.3; HRMS calcd for C₁₆H₂₂NO₂[M + NH₄]:

260.1645. Found: 260.1639.

1-Methoxy-2-(1-(4-methoxyphenyl)ethyl)benzene (64). Col-

orless oil; ¹H NMR (400 MHz, CDCl₃) δ 7.16−7.09 (m, 4 H), 6.89−

6.84 (m, 1 H), 6.82−6.77 (m, 3 H), 4.52 (q, J = 7.1 Hz, 1 H), 3.73 (s,

3 H), 3.72 (s, 3 H), 1.54 (d, J = 7.3 Hz, 3 H); ¹³C NMR (100 MHz,

CDCl₃) δ 157.7, 156.9, 138.6, 135.4, 128.7, 127.7, 127.1, 120.6, 113.6,

110.7, 55.5, 55.3, 36.7, 21.2; HRMS calcd for C₁₆H₂₂NO₂[M + NH₄]:

260.1645. Found: 260.1637.

(S)-5-Fluoro-N-(1-(4-methoxyphenyl)ethyl)-2-(5,5,5-tri-

ﬂuoro-2-methyl-4-oxopentan-2-yl)benzamide (22) from 60. A

reactor was charged with 60 (64.0 kg, 219.0 mol, 1 equiv) and toluene

(258 kg). N,N-dimethylacetamide (0.14 kg) was added. The slurry was

stirred at about 25 °C. Thionyl chloride (28.7 kg, 241.2 mol, 1.1

equiv) was charged and the batch was stirred for 30 min at 25 °C. The

reaction mixture was heated at about 50 °C for 3h to eﬀect complete

formation of acid chloride. The reaction mixture was then cooled to

about 10 °C. In a separate reactor was charged successively S-1-(4-

methoxy-phenyl)ethylamine (33.2 kg, 175.0 mol, 1.0 equiv), THF

(97.4 kg), and 2,6-lutidine (47 kg, 439.3 mol, 2.0 equiv), and the

solution was cooled to about 0−5 °C. The acid chloride solution from

the ﬁrst reactor was charged to the second reactor at a rate to control

the batch temperature at not more than 10 °C. The reaction mixture

was then stirred at about 15−20 °C for 2 h. The batch was cooled to

0−5 °C and quenched with a solution of concentrated HCl (83 kg) in

water (280 kg) at a rate to control the batch temperature at not more

than 30 °C. The batch temperature was adjusted to about 25 °C, and

the aqueous phase was separated. The organic phase was washed with

a solution of concentrated HCl (32 kg) and water (64 kg), then with

water (2 × 280 kg), and then the organic phase was distilled under

vacuum at up to 55 °C until no more distillate comes over. The batch

temperature was adjusted to 60−65 °C. Heptane (384 kg) was

charged at 60−65 °C followed by water (64 kg), also at 60−65 °C.

The batch was cooled linearly over 2 h to 5 °C and held at 5 °C for 1

h. The batch was ﬁltered, and the solid was washed with heptane (10

kg). The solid was dried under vacuum with a nitrogen stream at about

50 °C for 12 h. 22 was obtained as a white solid (70.0 kg, 99.7 wt%

purity, 99.7 area% purity, > 99.5% ee, 75.1% yield). Spectral data for

22 were consistent with material obtained from the isocyanate route.

(R)-2-(4-((5-(Ethylsulfonyl)-1H-pyrrolo[2,3-c]pyridin-2-yl)-

methyl)-5,5,5-triﬂuoro-4-hydroxy-2-methylpentan-2-yl)-5-ﬂu-

orobenzamide anisole solvate (1·PhOMe). A reactor was charged

with 24 (30.00 kg, 92.68 wt% purity, 42.79 mol, 1 equiv), anisole (89.4

kg) and 85% aqueous H₃PO₄(151.8 kg). The batch was heated at

about 100 °C for about 75 min. The batch was then cooled to about

80 °C. MEK (2-butanone, 48.3 kg) was charged at 75−83 °C. Then

water (150 kg) was charged over about 1h, while maintaining the

batch at 75−83 °C. The batch was held at about 80 °C for 30 min,

cooled over about 1h to 20−25 °C, and held at this temperature for

about 1h. The batch was then ﬁltered, and the solid was washed with

water (90 kg) followed by isopropanol (70.8 kg). The solid was dried

under vacuum at about 70 °C with a nitrogen sweep for 16 h, until KF

< 0.5%. 1·PhOMe was obtained as a white solid (26.17 kg, 80.98 wt%

free base (excluding anisole weight), 98.7 area% purity by HPLC, 96%

yield). mp 194−197 °C; ¹H NMR (400 MHz, d-6 DMSO) δ 11.59 (s,

1 H), 8.77 (s, 1 H), 8.34 (s, 1 H), 8.15 (s, 1 H), 7.99 (s, 1 H), 7.57

(dd, J = 5.6, 8.9 Hz, 1 H), 7.31−7.27 (m, 1 H), 7.19−7.11 (m, 2 H),

6.94−6.91 (m, 3 H), 6.82 (br, 1 H), 6.47 (s, 1 H), 3.75 (s, 3 H), 3.34

(q, J = 7.3 Hz, 2 H), 3.04 (d, J = 15.0 Hz, 1 H), 2.92 (d, J = 15.0 Hz, 1

H), 2.49 (d, J = 15.0 Hz, 1 H), 2.33 (d, J = 15.0 Hz, 1 H), 1.57 (s, 3

H), 1.55 (s, 3 H), 1.07 (t, J = 7.4 Hz, 3 H); ¹³C NMR (100 MHz, d-6

DMSO) δ 174.2, 160.9, 159.2, 158.5, 144.3, 141.4, 140.2 (d, J = 3.4

Hz), 138.7 (d, J = 5.9 Hz), 134.0 (d, J = 5.8 Hz), 131.8, 130.2 (d, J =

7.4 Hz), 129.4, 126.1 (d, J = 289 Hz), 120.4, 115.4 (t, J = 22 Hz),

114.2, 113.8, 102.8, 75.7 (q, J = 25 Hz), 54.9, 46.2, 44.6, 37.5, 33.2,

32.6, 31.2, 7.0; HRMS calcd for C₂₃H₂₆F₄N₃O₄S [M − PhOMe + H]:

516.1575. Found: 516.1565.

The following 3 impurities were isolated by preparative HPLC of a

portion of the concentrated organic layer of the mother liquors from

the crystallization of 1·PhOMe.

(R)-2-(4-((5-(Ethylsulfonyl)-1H-pyrrolo[2,3-c]pyridin-2-yl)-

methyl)-5,5,5-triﬂuoro-4-hydroxy-2-methylpentan-2-yl)-5-ﬂu-

orobenzoic acid (65). White solid; ¹H NMR (600 MHz, d-6

DMSO) δ 13.58 (br, 1 H), 11.64 (s, 1 H), 8.76 (s, 1 H), 8.12 (s, 1 H),

7.58 (dd, J = 5.6, 9.1 Hz, 1 H), 7.22−7.19 (m, 1 H), 7.18−7.16 (m, 1

H), 6.42 (s, 1 H), 6.17 (br, 1 H), 3.35−3.31 (m, 2 H), 2.87 (d, J = 15.2

Hz, 1 H), 2.71 (d, J = 15.6 Hz, 1 H), 2.64 (d, J = 15.2 Hz, 1 H), 2.18

(d, J = 15.2 Hz, 1 H), 1.63 (s, 3 H), 1.44 (s, 3 H), 1.06 (t, J = 7.3 Hz, 3

H); ¹³C NMR (150 MHz, d-6 DMSO) δ 171.8, 170.2, 160.5, 158.8,

144.4, 141.0, 140.9, 136.1, 133.9, 131.8, 130.1 (d, J = 7.7 Hz), 126.1

(q, J = 287 Hz), 115.9 (d, J = 20 Hz), 115.0 (d, J = 22 Hz), 114.2,

102.8, 75.8 (q, J = 25 Hz), 59.7, 46.2, 44.1, 37.9, 33.0, 32.6, 29.3, 20.7,

14.0, 6.9; HRMS calcd for C₂₃H₂₅F₄N₂O₅S [M + H]: 517.1420.

Found: 517.1411.

(R)-3-((5-(Ethylsulfonyl)-1H-pyrrolo[2,3-c]pyridin-2-yl)-

methyl)-8-ﬂuoro-5,5-dimethyl-3-(triﬂuoromethyl)-4,5-

dihydrobenzo[c]oxepin-1(3H)-one (66). White solid; ¹H NMR

(600 MHz, d-6 DMSO) δ 12.06 (s, 1 H), 8.75 (s, 1 H), 8.19 (s, 1 H),

7.49 (dd, J = 3.5, 6.0 Hz, 1 H), 7.37 (ddd, J = 2.9, 2.9, 8.3 Hz, 1 H),

7.20 (dd, J = 2.9, 8.9 Hz, 1 H), 6.53 (s, 1 H), 3.35 (q, J = 7.5 Hz, 2 H),

3.27 (d, J = 15.5 Hz, 1 H), 3.14 (d, J = 15.8 Hz, 1 H), 2.56−2.50 (m, 2

H), 1.41 (s, 3 H), 1.35 (s, 3 H), 1.10 (t, J = 7.3 Hz, 3 H); ¹³C NMR

(150 MHz, d-6 DMSO) δ 165.9 (d, J = 2.2 Hz), 160.3 (d, J = 245 Hz),

144.8, 141.0 (d, J = 3.3 Hz), 136.9, 134.2, 133.7, 132.0, 131.5 (d, J =

7.2 Hz), 127.7 (d, J = 7.7 Hz), 124.5 (q, J = 283 Hz), 119.8 (d, J = 21

Hz), 117.9 (d, J = 23 Hz), 114.7, 104.5, 81.7 (q, J = 26 Hz), 46.3, 44.1,

36.4, 34.1, 30.6, 30.0, 29.9, 6.8; HRMS calcd for C₂₃H₂₃F₄N₂O₄S [M +

H]: 499.1315. Found: 499.1322.

2-((4R)-4-((5-(Ethylsulfonyl)-3-(1-(4-methoxyphenyl)ethyl)-

1H-pyrrolo[2,3-c]pyridin-2-yl)meth-yl)-5,5,5-triﬂuoro-4-hy-

droxy-2-methylpentan-2-yl)-5-ﬂuorobenzamide (67). This com-

pound was formed as an inseparable ∼58:42 mixture of diastereomers.

White solid; ¹H NMR (600 MHz, d-6 DMSO) δ 11.34 (s, 1 H), 8.72−

8.71 (m, 1 H), 8.32 (s, 0.58 H, major diast.), 8.26 (s, 0.42 H, minor

diast.), 7.98 (s, 0.58 H, major diast.), 7.92 (s, 0.42 H, minor diast.),

7.75 (s, 0.42 H, minor diast.), 7.63 (s, 0.58 H, major diast.), 7.56 (dd, J

= 5.6, 8.9 Hz, 0.42 H, minor diast.), 7.52 (dd, J = 5.5, 8.8 Hz, 0.58 H,

major diast.), 7.19−6.88 (m, 5 H), 6.84−6.78 (m, 2H), 4.18−4.11 (m,

1 H), 3.71 (s, 1.74 H, major diast.), 3.69 (s, 1.26 H, minor diast.),

3.32−3.20 (m, 2 H), 3.06−2.91 (m, 2 H), 2.58−2.28 (m, 2 H), 1.57−

1.47 (m, 9 H), 1.02−0.98 (m, 3 H); ¹³C NMR (150 MHz, d-6

DMSO) δ 174.2, 174.0, 160.6, 160.5, 158.94, 158.90, 157.25, 157.20,

143.5, 139.9, 139.8, 138.8 (d, J = 6.4 Hz), 138.6 (d, J = 5.5 Hz), 137.2,

137.0, 136.6, 134.22, 134.17, 133.8, 133.7, 130.2, 130.1, 130.0, 129.6,

129.4, 127.8, 127.7, 126.1 (q, J = 287 Hz), 119.0, 118.9, 115.5, 115.4,

115.3, 115.1, 113.5, 113.43, 113.40, 76.0, 54.9, 48.5, 46.1, 45.5, 45.4,

37.4, 37.3, 33.5, 33.2, 33.0, 31.5, 31.0, 30.4, 30.3, 20.1, 6.7; HRMS

calcd for C₃₂H₃₆F₄N₃O₅S [M + H]: 650.2312. Found: 650.2319.

(R)-2-(4-((5-(Ethylsulfonyl)-1H-pyrrolo[2,3-c]pyridin-2-yl)-

methyl)-5,5,5-triﬂuoro-4-hydroxy-2-methylpentan-2-yl)-5-ﬂu-

orobenzamide phosphoric acid cocrystal (1•H₃PO₄) from

anisole solvate 1·PhOMe. A reactor was charged with 1·PhOMe

(26.10 kg, 80.98 wt% free base, 41.01 mol, 1 equiv) and MEK (138.6

kg). The mixture was heated to about 60 °C to obtain a solution. The

warm solution was polish ﬁltered into a clean reactor, using additional

The following 2 byproducts were isolated by chromatography on

SiO₂of a portion of the concentrated organic layer of the mother

liquors from the crystallization of 1·PhOMe.

R

dx.doi.org/10.1021/jo400079z | J. Org. Chem. XXXX, XXX, XXX−XXX

The Journal of Organic Chemistry

Article

warmed (≥40 °C) MEK (46.2 kg) to rinse the ﬁlter. After adjusting

the batch temperature to 50 °C, H₃PO₄(4.91 kg, 43.06 mol, 86.0 wt%,

1.05 equiv) was added at about 50 °C. The batch was treated with

heptane (26.1 kg) was added over about 30 min at about 50 °C. Seed

crystals of 1·H₃PO₄(26.0 g) were added as a slurry in heptane (1 L).

The batch was stirred for 30 min at about 50 °C while a slurry

developed. Then heptane (52.2 kg) was added over 1h at about 50 °C.

The batch was cooled linearly over 2 h to about 20 °C and held at

about 20 °C for 3 h. The batch was ﬁltered, and the solid was washed

with MEK/heptane 1:2 v/v (37.8 kg) followed by heptane (35.8 kg).

The solid was dried under vacuum at 80 °C for 24 h. 1·H₃PO₄was

obtained as a white solid (24.4 kg, 99.7 area% purity by HPLC, 99.9%

ee, 97% yield). Spectral data for 1·H₃PO₄were consistent with

material obtained starting from 1·AcOH.

olanes with Triﬂuoromethyl Ketones. PCT. Int. Appl. WO/2010/

141328, 2010.

(4) Lee, T. W.; Proudfoot, J. R.; Thomson, D. S. Bioorg. Med. Chem.

Lett. 2006, 16, 654.

(5) Harcken, C.; Ward, Y.; Thomson, D.; Riether, D. Synlett 2005,

3121.

(6) For alternative synthetic approaches to structurally related

molecules see: (a) Song, J. J.; Tan, Z.; Xu, J.; Yee, N. K.; Senanayake,

C. H. Process for Stereoselective Synthesis of Certain Triﬂuoromethyl-

Substituted Alcohols. PCT. Int. Appl. WO/2005/040145, 2005;

(b) Song, J. J.; Tan, Z.; Xu, J.; Yee, N. K.; Senanayake, C. H.; Xu,

J.; Gallou, F. Stereoselective Synthesis of Certain Triﬂuoromethyl-

Substituted Alcohols. U.S. Pat. Appl. Publ. US/2005/0209488, 2005;

(c) Song, J. J.; Reeves, J.; Roschangar, F.; Tan, Z.; Yee, N. K. Process

for Stereoselective Synthesis of Triﬂuoromethyl-Substituted Alcohols.

PCT. Int. Appl. WO/2007/040959, 2007; (d) Reeves, J. T.; Song, J. J.;

Tan, Z. Stereoselective Synthesis of Certain Triﬂuoromethyl-

Substituted Alcohols. PCT. Int. Appl. WO/2009/134737, 2009.

(7) As of January 9, 2013, 1 gram of (S)-(−)-methyl p-tolyl sulfoxide

(7), catalog # 339997, cost $397.00 from Sigma-Aldrich (http://www.

sigmaaldrich.com/catalog/product/aldrich/339997).

ASSOCIATED CONTENT

■

S

* Supporting Information

1

Copies of H and ¹³C NMR spectra and X-ray crystallographic

data for 1·H₃PO₄. This material is available free of charge via

the Internet at http://pubs.acs.org.

(8) Mauragis, M. A.; Veley, M. F.; Lipton, M. F. Org. Process Res. Dev.

1997, 1, 39.

AUTHOR INFORMATION

■

(9) (a) Paik, S.; Lee, J. Y. Tetrahedron Lett. 2006, 47, 1813.

(b) Chern, C.-Y.; Huang, Y.-P.; Kan, W. M. Tetrahedron Lett. 2003, 44,

1039. (c) Ishihara, K.; Hiraiwa, Y.; Yamamoto, H. Synlett 2000, 80.

(d) James, I. W. Tetrahedron 1999, 55, 4855.

(10) Henne, A. L.; Hinkamp, P. E. J. Am. Chem. Soc. 1954, 76, 5147.

(11) Abouabdellah, A.; Bonnet-Delpon, D. Tetrahedron 1994, 50,

11921.

(12) (a) Boymond, L.; Rottlander, M.; Cahiez, G.; Knochel, P.

Angew. Chem., Int. Ed. 1998, 37, 1701. (b) Knochel, P.; Dohle, W.;

Gommermann, N.; Kneisel, F. F.; Kopp, F.; Korn, T.; Sapountzis, I.;

Vu, V. A. Angew. Chem., Int. Ed. 2003, 42, 4302.

(13) (a) Liu, W.; Lei, A. Tetrahedron Lett. 2008, 49, 610. (b) Surry,

D. S.; Su, X.; Fox, D. J.; Franckevicius, V.; Macdonald, S. J. F.; Spring,

J. R. Angew. Chem., Int. Ed. 2005, 44, 1870.

Corresponding Author

*E-mail: jonathan.reeves@boehringer-ingelheim.com.

Notes

The authors declare no competing ﬁnancial interest.

ACKNOWLEDGMENTS

■

The authors acknowledge Professors Erick Carreira, Scott

Denmark, Barry Trost, and Peter Wipf for insightful

discussions. Dr. Herbert Nar is thanked for performing the

X-ray crystal structure analysis of 1·H₃PO₄. Analytical support

provided by Dr. Uwe Zimmermann, Alexandra Backes, Carina

Mayer, Irina Peters and Heike Roth is gratefully acknowledged.

(14) Walling, C.; Buckler, S. A. J. Am. Chem. Soc. 1955, 77, 6032.

(15) Initial studies targeted the phenethyl amide, but it was found

that the phenethyl group could only be removed using concentrated

H₂SO₄, and under these conditions the substrates examined were

destroyed. We therefore switched to the 4-methoxyphenethyl group,

which can be removed under considerably milder condtions from

amides.

(16) Beccalli, E. M.; Broggini, G.; Martinelli, M.; Sottocornola, S.

Chem. Rev. 2007, 107, 5318.

́ ́

(17) Begue, J. P.; Bonnet-Delpon, D. Tetrahedron 1991, 47, 3207.

(18) (a) Simmons, H. E., Jr.; Wiley, D. W. J. Am. Chem. Soc. 1960, 82,

2288. (b) Aubert, C.; Begue, J. P.; Charpentier-Morize, M.; Nee, G.;

Langlois, B. J. Fluorine Chem. 1989, 44, 377. (c) Barbasiewicz, M.;

Makosza, M. Org. Lett. 2006, 8, 3745.

(19) Sigurdsson, S. T.; Seeger, B.; Kutzke, U.; Eckstein, F. J. Org.

Chem. 1996, 61, 3883.

(20) Krasovskiy, A.; Knochel, P. Angew. Chem., Int. Ed. 2004, 43,

3333.

(21) Krasovskiy, A.; Straub, B. F.; Knochel, P. Angew. Chem., Int. Ed.

2006, 45, 159.

REFERENCES

■

(1) (a) Betageri, R.; Gilmore, T.; Kuzmich, D.; Kirrane, T. M.;

Bentzien, J.; Wiedenmayer, D.; Bekkali, Y.; Regan, J.; Berry, A.; Latli,

B.; Kukulka, A. J.; Fadra, T. N.; Nelson, R. M.; Goldrick, S.; Zuvela-

Jelaska, L.; Souza, D.; Pelletier, J.; Dinallo, R.; Panzenbeck, M.;

Torcellini, C.; Lee, H.; Pack, E.; Harcken, C.; Nabozny, G.; Thomson,

D. S. Bioorg. Med. Chem. Lett. 2011, 21, 6842. (b) Riether, D.;

Harcken, C.; Razavi, H.; Kuzmich, D.; Gilmore, T.; Bentzien, J.; Pack,

E. J., Jr.; Souza, D.; Nelson, R. M.; Kukulka, A.; Fadra, T. N.; Zuvela-

Jelaska, L.; Pelletier, J.; Dinallo, R.; Panzenbeck, M.; Torcellini, C.;

Nabozny, G. H.; Thomson, D. S. J. Med. Chem. 2010, 53, 6681.

(c) Regan, J.; Lee, T. W.; Zindell, R. M.; Bekkali, Y.; Bentzien, J.;

Gilmore, T.; Hammach, A.; Kirrane, T. M.; Kukulka, A. J.; Kuzmich,

D.; Nelson, R. M.; Proudfoot, J. R.; Ralph, M.; Pelletier, J.; Souza, D.;

Zuvela-Jelaska, L.; Nabozny, G.; Thomson, D. S. J. Med. Chem. 2006,

49, 7887.

(2) Betageri, R.; Bosanac, T.; Burke, M. J.; Harcken, C.; Kim, S.;

Kuzmich, D.; Lee, T. W.-H.; Li, Z.; Liu, P.; Lord, J.; Razavi, H.; Reeves,

J. T.; Thomson, D. Azaindole Derivatives as Glucocorticoid Mimetics,

Their Preparation, Pharmaceutical Compositions, and Uses Thereof.

PCT. Int. Appl. WO/2009/149139, 2009.

(3) The general synthetic strategy towards 1·H₃PO₄has been

covered by the following provisional patent applications: (a) Fandrick,

D. R.; Reeves, J. T.; Song, J. J.; Tan, Z.; Qu, B.; Yee, N.; Rodriguez, S.

Process for Preparation of Triﬂuoromethyl-Substituted Alcohols by

Stereoselective Synthesis. PCT. Int. Appl. WO/2010/141331, 2010;

(b) Fandrick, D. R.; Reeves, J. T.; Song, J. J.; Yee, N. Process for

Preparation of Aralkyl Triﬂuoromethyl Ketones from Meldrum’s Acid

and Ketones. PCT. Int. Appl. WO/2010/141334, 2010; (c) Fandrick,

D. R.; Reeves, J. T.; Song, J. J. Process for Stereoselective Synthesis of

Alkynyl Aryl Triﬂuoromethyl Alcohols via Reaction of Silylalkynylbor-

(22) Ribe, S.; Wipf, P. Chem. Commun. 2001, 299.

(23) Corey, E. J.; Kirst, H. A. Tetrahedron Lett. 1968, 48, 5041.

(24) ZnBr₂was preferred over ZnCl₂due to the much lower

exothermicity of its dissolution in THF. A 26 wt% solution of ZnBr₂in

THF was employed.

(25) (a) Marshall, J. A.; Adams, N. D. J. Org. Chem. 1998, 63, 3812.

(b) Marshall, J. A.; Adams, N. D. J. Org. Chem. 1999, 64, 5201.

(26) Kim, S.; Li, Z.; Tseng, Y.-C.; Spinelli, E.; Nar, H.; Varsolona, R.;

Reeves, J. T.; Lee, H.; Song, J. J.; Smoliga, J.; Yee, N.; Senanayake, C.

Org. Process Res. Dev. 2013, 17, 540−548.

(27) Vogt, P. F.; Molino, B. F.; Robichaud, A. J. Synth. Commun.

2001, 31, 679.

S

dx.doi.org/10.1021/jo400079z | J. Org. Chem. XXXX, XXX, XXX−XXX

The Journal of Organic Chemistry

Article

(28) Haslego, M. L.; Smith, F. X. Synth. Commun. 1980, 10, 421.

(29) A more direct approach to 57, by Friedel−Crafts reaction of 1-

bromo-3-ﬂuorobenzene with 3,3-dimethylacrylic acid in the presence

of AlCl₃gave predominantly the wrong regioisomer (9:1 regiose-

lectivity favoring alkylation at the position ortho to ﬂuorine and para to

bromine). The same 9:1 regioselectivity was observed in the Friedel−

Crafts reaction of 1-bromo-3-ﬂuorobenzene with triﬂuoromethyl

enone 28.

(30) (a) Reeves, J. T.; Gallou, F.; Song, J. J.; Tan, Z.; Lee, H.; Yee, N.

K.; Senanayake, C. H. Tetrahedron Lett. 2007, 48, 189. (b) Reeves, J.

T.; Tan, Z.; Fandrick, D. R.; Song, J. J.; Yee, N. K.; Senanayake, C. H.

Org. Synth. 2012, 89, 210.

(31) (a) Boivin, J.; El Kaim, L.; Zard, S. Z. Tetrahedron Lett. 1992, 33,

1285. (b) Boivin, J.; El Kaim, L.; Zard, S. Z. Tetrahedron 1995, 51,

2573. (c) Boivin, J.; El Kaim, L.; Zard, S. Z. Tetrahedron 1995, 51,

2585.

(32) (a) Fandrick, D. R.; Roschangar, F.; Kim, C.; Hahm, B. J.; Cha,

M. H.; Kim, H. Y.; Yoo, G.; Kim, T.; Reeves, J. T.; Song, J. J.; Tan, Z.;

Qu, B.; Haddad, N.; Shen, S.; Grinberg, N.; Lee, H.; Yee, N. K.;

Senanayake, C. H. Org. Process Res. Dev. 2012, 16, 1131. (b) Fandrick,

K. R.; Fandrick, D. R.; Reeves, J. T.; Gao, J.; Ma, S.; Li, W.; Lee, H.;

Grinberg, N.; Lu, B. Z.; Senanayake, C. H. J. Am. Chem. Soc. 2011, 133,

10332. (c) Fandrick, D. R.; Fandrick, K. R.; Reeves, J. T.; Tan, Z.;

Tang, W.; Capacci, A. G.; Rodriguez, S.; Song, J. J.; Lee, H.; Yee, N. K.;

Senanayake, C. H. J. Am. Chem. Soc. 2010, 132, 7600.

(33) Fandrick, D. R.; Reeves, J. T.; Bakonyi, J. M.; Nyalapatla, P. R.;

Tan, Z.; Niemeier, O.; Akalay, D.; Ollenberger, S.; Fandrick, K. R.;

Song, J. J.; Sun, X.; Qu, B.; Haddad, N.; Shen, S.; Ma, S.; Byrne, D.;

Chitroda, A.; Fuchs, V.; Narayanan, B. A.; Grinberg, N.; Lee, H.; Yee,

N.; Wohlleben, W.; Brenner, M.; Senanayake, C. H. J. Org. Chem.

2013, DOI: jo400080y.

T

dx.doi.org/10.1021/jo400079z | J. Org. Chem. XXXX, XXX, XXX−XXX

Article Doi

Development of a large scale asymmetric synthesis of the glucocorticoid agonist BI 653048 BS H3PO4

DOI: 10.1021/jo400079z

Source and publish data:

Authors:

Article abstract of DOI:10.1021/jo400079z

Full text of DOI:10.1021/jo400079z

Products guided by the article

R&D Labs maybe for 1257648-83-1

Relevant to this article

Hot Product