Route selection and process development of a multikilogram route to the inhaled A2a agonist UK-432,097

Article

pubs.acs.org/OPRD

Route Selection and Process Development of a Multikilogram Route

to the Inhaled A_2aAgonist UK-432,097

Christopher P. Ashcroft, Yann Dessi, David A. Entwistle,* Lynsey C. Hesmondhalgh, Adrian Longstaff,

and Julian D Smith*

Chemical R&D, Pfizer Global Research and Development, Ramsgate Road, Sandwich, Kent CT13 9NJ, United Kingdom

ABSTRACT: This article describes the selection, process development, and scale-up of a synthetic route to a complex

nucleoside analogue, the A_2aagonist UK-432,097 (1), that culminated in the manufacture of over 25 kg of the API. The key steps

in the process were (1) a stereoselective glycosidation reaction; (2) a scalable bleach−TEMPO oxidation; and (3) an unusual

elevated temperature crystallization process for the final API. The problems that were encountered with the scale-up of the route

together with how they were overcome are also presented.

glycosidation reaction between the uronamide 2 and the

INTRODUCTION

■

adenine core 3 under Vorbruggen conditions⁷to give 4

̈

Chronic obstructive pulmonary disease (COPD) is currently

the fourth leading cause of death in the United States,¹affects

between 10 to 24 million of the U.S. population, and is

predicted to become the fourth leading cause of death

worldwide by 2030.²COPD is a progressive degenerative

disease that is described as reduced capacity to breathe due to

inflammation and damage of the lung tissues.³It is a severely

debilitating disease that reduces quality of life, and the high

level of patient aftercare results in a high economic burden.

Treatments are limited, and as such there is a growing need for

alternative therapies.⁴

followed by elaboration of the methyl ester on the heteroaryl

group into the complete aminopyridine-containing side chain

via the union of 5 and 6.^8a

From the perspective of developing a multikilo synthesis of 1,

we wished to identify a robust and efficient synthetic route that

would be amenable to scale-up. We investigated the key bond-

forming steps in the medicinal chemistry route, and identified

major problems with the uronamide 2 that precluded it being

used in a large scale manufacturing process. The synthesis of

the uronamide 2 started from ^D-ribose as shown in Scheme 2

and utilised classical methods, including a ruthenium-catalysed

oxidation^10aof 7 to give 8^10bprior to installing the amide in 9

together with a protection−deprotection-reprotection sequence

and the intermediacy of oils (7, 10) and diastereomeric

mixtures.

UK-432,097 (1) (Figure 1) is a potent adenosyl subtype 2a

agonist (A_2aagonist)⁵that offers a potential inhaled treatment

While the reagents used in this route were not particularly

attractive for the development of a large scale process, it was

the stability of the dibenzoate 2 that caused most concern. This

material existed in its pure form as a gum despite extensive

efforts to find a crystalline form, and required chromatographic

purification. However, even when pure, the dibenzoate 2 was

unstable and degraded quite quickly via what was presumed to

be a β-elimination mechanism that ultimately led to what we

believed to be furan-derivatives. Degradation of 2 even

occurred during solvent strip-down of solutions, and also

upon storage of the material as a gum. These factors led us to

conclude that 2 was not a viable intermediate in a scalable

synthesis of 1.

The lack of stability of 2 highlighted to us the importance of

selecting the right protecting group strategy when designing a

new route. The amide oxidation state at C-5 of the sugar

moiety in 1 renders the methine hydrogen at C-4 relatively

acidic and prone to the observed β-elimination of the protected

oxygen atom at C-3, as well as epimerization at C-4. For the β-

elimination pathway, having a good leaving group (e.g., a

Figure 1. Structure of UK-432,097.

for the inflammatory symptoms of COPD.^4,6To support

development of 1, a series of demands for active pharmaceutical

ingredient (API) were received. Campaign 1 (300 g),

Campaign 2 (2.5 kg), and Campaign 3 (25 kg) would support

the project through preclinical toxicology testing, drug product

development in a solution and dry powder inhaler, and in

phases 1 and 2 clinical programs.

In this article, we describe the selection of the route used for

scale-up, together with how it was initially developed and

implemented in Campaigns 1 and 2 and further developed and

refined for the manufacture of 25 kg of 1 in Campaign 3.

ROUTE SELECTION

The final stages of the route used in the medicinal chemistry

synthesis of 1 (Scheme 1) involved the stereoselective

■

Received: December 11, 2011

Published: February 10, 2012

470

dx.doi.org/10.1021/op200365n | Org. Process Res. Dev. 2012, 16, 470−483

Organic Process Research & Development

Article

a

Scheme 1. Final stages of the medicinal chemistry route to 1

a

Reagents and conditions: (a) 3, N,O-bis(trimethylsilyl)acetamide, PhMe then 2, reflux; (b) (i) Na₂CO₃, MeOH, RT; (ii) ethylenediamine, 105 °C;

(c) PhMe/ⁱPrOH, reflux.

a

Scheme 2. Synthesis of uronamide 2

present in 1 as late in the synthesis as possible via an amide

bond-forming reaction. Taking these factors into account

together with the strategy for the sugar unit, led us to the

retrosynthetic analysis for the new route to 1 as depicted in

Figure 2. Here, the nucleoside core 11 and the amine side chain

12 are coupled together followed by a final deprotection

reaction to give 1. Given the earlier discussion, the nucleoside

11 would logically be prepared via oxidation of the primary

alcohol 13 followed by ethylamide formation.

However, stereoselective installation of the preferred nucleo-

side anomer 13 in the presence of the acetonide would be

challenging. It is known that the presence of an acyl-protected

hydroxyl group at C-2 is necessary for anchimeric assistance to

impart high levels of stereoselectivity in glycosidation

reactions;^7,8bthus, having the acetonide present during the

reaction would most likely lead to poor diastereoselectivity.

Additionally it was unlikely that the acetonide would be

compatible with the reaction conditions used in glycosidation

reactions (e.g., TMS triflate or other Lewis acids). It therefore

seemed logical to prepare the fully protected nucleoside 14 via

a glycosidation reaction between cheap and commercially

available 1,2,3,5-tetra-O-acetyl-β-^D-ribofuranose (15) and a

suitably functionalized adenine unit 16. In our previous

contribution we described the synthesis of the related

nucleoside analogue UK-371,104 which utilized a similar

glycosidation reaction to unify the sugar and adenine units.^8b

a

Reagents and conditions: (a) HCl, MeOH, acetone; (b) KIO₄, RuCl₃,

acetone (aq); (c) CDI, THF, EtNH₂; (d) PPTS, MeOH, reflux; (e)

(i) Bz₂O, TEA, DMAP, DCM; (ii) AcOH, Ac₂O, H₂SO₄, DCM.

benzoate ester as in 2) would promote this pathway in

comparison with a poorer leaving group (e.g. alkoxy), although

epimerization could still be a problem. These considerations led

us to identify an alkoxy-like protecting group, specifically

acetonide, as this has the additional advantage of also protecting

the hydroxyl at C-2. To maintain as much convergency as

possible, we wished to install the complete amide side chain

Figure 2. Retrosynthetic analysis for process chemistry route to 1.

471

dx.doi.org/10.1021/op200365n | Org. Process Res. Dev. 2012, 16, 470−483

Organic Process Research & Development

Article

We anticipated that similar reaction conditions would be

applicable for the glycosidation involved in the new route to 1.

The synthesis of the aminopyridine containing side-chain

unit 12 from cheap and readily available N-benzyl-4-amino-

piperidine (17) is shown in Scheme 3. Protection of the

The adenine unit 16 is related to an ester of the carboxylic

acid used in the synthesis of UK-371,104 reported previously.^8b

However, the first synthesis of the carboxylic acid via palladium-

catalysed cyanation of the respective chloride 24 followed by

nitrile hydrolysis proved to be quite troublesome, and therefore

an alternative method was developed that utilized palladium-

catalysed ethoxycarbonylation of the chloride 24 to give the

ethyl ester 25 (Scheme 4). The chloride substrate 24 was

a

Scheme 3. Synthesis of amine 12

a

Scheme 4. Ethoxycarbonylation route to 16

a

Reagents and conditions: (a) 0.4 mol % Pd(OAc)₂, 0.8 mol % dppf,

Na₂CO₃(aq), EtOH, 300 psi CO, 105 °C, 76%; (b) TFA, EtOH, 50

°C, 99%.

prepared using the published procedure,^5c,8band reaction with

ethanol and carbon monoxide in the presence of Pd(OAc)₂/

dppf and aqueous sodium carbonate afforded the desired ester

25 in 76% yield. Occasional venting and repressurization of the

autoclave with carbon monoxide was necessary to remove the

carbon dioxide generated in the reaction, as this diluted the

carbon monoxide and slowed the reaction rate. Finally, removal

of the THP protecting group was accomplished in almost

quantitative yield using trifluoroacetic acid (TFA) in ethanol to

provide 16. This route was used by external vendors to prepare

a total of 11 kg of 16 to use in Campaigns 1 and 2.

a

Reagents and conditions: (a) (i) AcCl, TEA, THF; (ii) EtOAc, 77%;

(b) (i) H₂/Pd/C, EtOH; (ii) EtOAc, 83%; (c) AmOH, Na₂CO₃, 2-

bromopyridine, 82%; (d) HCl (6 N), EtOH, 88%; (e) (i) CDI,

MeCN, DIPEA; (ii) tert-butyl (2-aminoethyl)carbamate; (iii) EtOAc,

94%; (f) AcCl, EtOH, 94%.

primary amine in 17 as an acetamide to give 18 was followed by

hydrogenolysis of the benzyl group to give 4-acetamidopiper-

idine (19) as a crystalline solid. Thermally induced reaction of

19 with 2-bromopyridine gave 20 and hydrolysis of the

acetamide group afforded the aminopiperidine 21 that was

isolated as its dihydrochloride salt.^5c,9Formation of the urea

using 1,1′-carbonyldiimidazole (CDI) initially posed some

problems due to the formation of the symmetrical urea

derivative via reaction of the carbamoylimidazolide derivative

22 with 21. This problem was overcome by slowly adding

diisopropylethylamine (DIPEA) to a cooled suspension of 21

dihydrochloride and CDI in acetonitrile (in which DIPEA·HCl

is soluble), resulting in slow release of 21 free base to generate

the required carbamoyl imidazolide 22 with minimal formation

of the undesired symmetrical urea side product. Subsequent

addition of BOC-protected ethylenediamine and reaction at

reflux furnished the desired urea 23 that, following workup, was

crystallized from ethyl acetate. Deprotection of 23 to provide

12 was accomplished with anhydrous HCl in ethanol generated

from acetyl chloride addition to the ethanol solution of 23.

Interestingly, the only crystalline form of 12 dihydrochloride

that could be isolated was a solvate containing 0.4 mol of

ethanol per mole of 12. Over 35 kg of 12 dihydrochloride

ethanolate was prepared at external vendors using this route.

ASSEMBLY OF BUILDING BLOCKS AND

CONVERSION TO 1

■

Glycosidation of 16 with 1,2,3,5-tetra-O-acetyl-β-D-ribofuranose

(15) was accomplished, using conditions similar to those

published^8bfor UK-371,104, and afforded the noncrystalline

product 14 (Scheme 5).

When the published conditions were repeated exactly with

15 and 16, with no external base, a much poorer reaction

profile was obtained compared to the UK-371,104 substrate

which contained a basic piperidine moiety. In the presence of

an amine such as N-methylmorpholine (NMM), the reaction of

15 and 16 was much cleaner, and it was found that a slight

excess of TMS triflate over NMM was necessary to achieve

good reaction conversions. We suspect that this observation

was due to the reaction of TMS triflate with NMM to give the

silylated quaternary ammonium triflate salt 26 (Figure 3)

which, while being reactive enough to silylate the adenine unit

16 to give 27 as required for the reaction, is presumably not a

sufficiently reactive enough silyl transfer reagent to promote

ionization of 15 to give the stabilized oxonium ion 28. The

slight excess of TMS triflate (1.3 equiv) relative to NMM (1.1

472

dx.doi.org/10.1021/op200365n | Org. Process Res. Dev. 2012, 16, 470−483

Organic Process Research & Development

Article

the use of saturated solutions resulted in no purge of this

impurity, probably because the increased ionic strength of the

aqueous phase prevented the sodium salt of 30 from

partitioning into that phase. The triol 29 was finally crystallized

and isolated in good yield over the two steps. The

corresponding oppositely configured α-anomer was not

detected in either crude mixtures of 14 and 29 or in isolated

solid 29, the anomeric centre essentially being perfectly stereo

controlled by the neighboring 2-acetate group in cation 28.

This process was run internally within Pfizer lab and kilo lab

facilities for Campaigns 1 and 2, with totals of 976 g and 11.0

kg of 29 being made respectively. For Campaign 3, the process

to 29 was outsourced, and 52 kg was received for subsequent

processing.

Scheme 5. Glycosidation and conversion to triol 29

(Campaigns 1 and 2)

a

Having coupled the sugar and adenine units together, the

next phase of the synthesis required elaboration of the primary

hydroxyl in 29 to an ethyl amide. This required selective

protection of the two secondary alcohol groups in 29 to enable

oxidation of the primary hydroxyl group. Formation of the

acetonide 13 was accomplished with sulphuric acid in a mixture

of acetone and 2 equiv of 2,2-dimethoxypropane (Scheme 6).

a

Scheme 6. Acetonide formation Campaign 3

a

Reagents and conditions: (a) TMS-OTf, DME, NMM, 60 °C; (b)

EtOH, DBU. 90% over two steps.

equiv) provides sufficient reactive catalyst to enable the

glycosidation reaction to proceed.

The product 14 from the glycosidation reaction was not

crystalline; thus, following workup the solution of 14 was taken

directly into the next step where the acetate groups were

removed in the presence of the ethyl ester attached to the

adenine unit to give the triol 29 (Scheme 5). This was initially

accomplished using transesterification in ethanol with 0.1 equiv

of sodium ethoxide as the catalyst. However, these conditions

invariably led to the formation of several percent of the

carboxylic acid 30, presumably due to the presence of traces of

water and sodium hydroxide in the reaction system.

Efforts to find conditions that minimized the formation of 30

culminated in using a stoichiometric amount of DBU in

ethanol, together with minimization of water to below 0.1 w/w

% in the reaction medium. Even with these optimal conditions

it was not possible to completely prevent formation of 30 and

to avoid this impurity from interfering in the downstream

chemistry, it was removed at this stage by washing the product

solution with aqueous 0.1 M NaHCO₃solution. Interestingly,

a

Reagents and conditions: (a) Acetone, 2,2-DMP, H₂SO₄(0.95

equiv), 88%.

Successful reactions required the use of 1.0 equiv, rather than a

catalytic amount, of concentrated sulphuric acid, and the

product 13 was isolated in a direct drop process from the

reaction mixture as the hydrogen sulphate salt. It is likely that

the driving force for this equilibrium reaction was the

crystallization of the product. Although the free base of 13

was also crystalline, we elected to isolate and use the hydrogen

sulphate salt directly in the subsequent step. For Campaigns 1

and 2, this process afforded 1.1 and 10 kg of the acetonide

hydrogen sulphate salt 13, respectively.

Figure 3. Rationalization of the stoichiometries of NMM and TMS triflate.

473

dx.doi.org/10.1021/op200365n | Org. Process Res. Dev. 2012, 16, 470−483

Organic Process Research & Development

Article

Further scale-up for Campaign 3 required a minor develop-

ment of the acetonide-forming step. During Campaigns 1 and

2, variations in the colour and long-term stability of the

acetonide hydrogen sulphate salt 13 were noted, and it was

rationalised that traces of residual sulphuric acid could

potentially be entrained within the isolated product giving

rise to the instability and colouring on storage. Reducing the

charge of sulphuric acid to 0.95 equiv gave a reproducibly stable

hydrogen sulphate salt 13 that was again isolated by direct drop

crystallization from the reaction medium in 88% yield. In

Campaign 3, a total of 55 kg of hydrogen sulphate salt 13 was

isolated in two lots (Scheme 6).

TEMPO, the starting material, and product was carried out

prior to scale-up as was full reaction calorimetry and TSU

testing of intermediate solutions. No significant barriers were

found to progression of this step or any other step described

herein. Details of the testing for this and subsequent Campaign

3 steps are given in the Experimental Section.

For Campaign 3 further optimization of the reaction and

workup was investigated as the phase separations during the

workup operations were slow and thought to be due to TBAB

acting as a surfactant. Tetraethylammonium bromide (TEAB)

is approximately 4.5 times more water soluble than TBAB,¹⁶

and TEAB gave an acceptable rate of reaction and a moderately

improved rate of phase separation during workup compared to

TBAB. Conversely, but as expected, trials of NaBr and KBr

gave better workup phase separations but drastically reduced

reaction rates due to negligible bromide levels in the organic

phase. In Campaign 3, TBAB was replaced with TEAB, and the

DCM solvent volume was reduced by 33%, giving approx-

imately 48 kg of acid 31 in two batches as a DCM solution

(Scheme 7).

The next stage of the synthesis was the oxidation of the

primary hydroxyl group in 13 (Scheme 7). Oxoammonium salts

Scheme 7. Oxidation, amide formation, and saponification

a

sequence, Campaign 3

In Campaigns 1 and 2, the DCM solutions of 31 obtained

after workup were exchanged into THF by distillation, and the

solution was azeotropically dried in readiness for the

subsequent step. Formation of the ethylamide 32 proved to

be relatively straightforward and was accomplished by

activation of the carboxylic acid in THF using CDI followed

by reaction with a commercial solution of ethylamine in THF.

It was surprising to find that commercially available solutions of

ethylamine in THF contained large amounts of water and

hydrolysis of the acylimidazolide back to 31 was a significant

competing reaction when the ethylamine solution was added to

the acylimidazolide. We were unable to secure a source of

anhydrous ethylamine in THF, and it would have been

challenging to accurately and safely charge controlled amounts

of anhydrous ethylamine (bp 17 °C). In the event we found

that a reverse addition protocol, carried out by adding the

solution of the acylimidazolide derived from 31 to a solution of

1.2 equiv of 70% w/w aqueous ethylamine in THF at −10 °C,

suppressed hydrolysis, back to 31, to an acceptable few percent.

After quenching the reaction with aqueous citric acid and

switching solvents to ethyl acetate, residual 31 was removed by

washing with 1% w/w aqueous triethanolamine. Once again,

the product was not crystalline; thus, the solution of 32 in ethyl

acetate was exchanged into methanol in readiness for the next

step.

a

Reagents and conditions: (a) (i) NaClO (aq), TEAB, Na₂CO₃,

TEMPO, DCM; (ii) Na₂SO₃quench; (iii) HCl (aq); (b) (i) CDI,

DCM, −10 °C; (ii) EtNH₂·HCl, DCM (reverse addition); (iii) Swap

to MeOH; (c) [to 11a] (i) NaOH (aq), MeOH; (ii) Swap to ⁱPrOAc,

t

i

then 1 N HCl (aq); (iii) 2 N HCl, water; (iv) BuNH₂, PrOAc; (v)

Azeo dry; (vi) ^tBuNH₂, ⁱPrOAc, 83% from 13; (d) [to 11b] (i) NaOH

i

(aq), MeOH; (ii) Swap to PrOAc, then citric (aq); (iii) MEK,

dicyclohexylmethylamine, 68% from 13.

The use of aqueous ethylamine in Campaign 2 not only

imparted operational complexity of solvent exchanges but also

carried a high risk that competing imidazolide hydrolysis might

become a significant factor as the process was scaled up. The

solvent exchanges were reduced in Campaign 3 by the

discovery that the amide formation could be performed in

DCM with a base and solid ethylamine hydrochloride, and as

such, a solvent swap from that used in the oxidation step was

not required. In laboratory trials the use of TEA as the

liberating base gave poorly stirred mixtures as TEA hydro-

chloride formed a gum, whereas DIPEA gave good stirring as it

gave a soluble hydrochloride salt. Critically, when using amine

bases to liberate the ethylamine, an impurity, poorly resolved by

are known for their ability to oxidise functional groups

including alcohols,¹¹and their in situ preparation from stable

radical precursors with various oxidants is well documented.¹²

In Campaigns 1 and 2, oxidation of the primary hydroxyl group

in 13 was carried out using 1,1,5,5-tetramethylpiperidinyloxyl

(TEMPO) and bleach in a two-phase system^13a,14,15utilizing

DCM and aqueous sodium bicarbonate in the presence of

tetrabutylammonium bromide (TBAB).^13dWe also evaluated

the sodium chlorite conditions that have been successfully

scaled up^13b,cbut concluded that the bleach−TEMPO−TBAB

conditions were safer to scale since the use and handling of

hazardous sodium chlorite was avoided, and accordingly these

conditions were used for the oxidation of 1 kg and 9.9 kg of 13

hydrogen sulphate salt during Campaigns 1 and 2, respectively.

The product 31 was not crystalline, so solutions of 31 in DCM

were obtained from this stage, and were used as such for the

next step. Full safety testing consisting of DSC testing of

1

HPLC, was seen at up to 15% by H NMR spectroscopy and

was identified as the C-4 epimer, 33 (Figure 4).

Retrospectively, when using a modified analytical HPLC

method, the level of 33 in the original Campaign 2 intermediate

32 was found to be 5%, so this gave confidence that 33 could

474

dx.doi.org/10.1021/op200365n | Org. Process Res. Dev. 2012, 16, 470−483

Organic Process Research & Development

Article

phase. At this stage, the salt 11a was still soluble, but

crystallization was achieved by azeotropically drying the

solution. Whilst this process worked well on 300-g scale in

the lab for Campaign 1, on scaling this process to 4 kg into a

kilo lab for Campaign 2 we observed a substoichiometric

amount of tert-butylamine in the salt present in the slurry which

we attributed to instability of the salt at the distillation

temperature, resulting in some decomposition back to volatile

tert-butylamine (bp 46 °C) and the free acid. This issue was

overcome by addition of further tert-butylamine to the slurry

after cooling, but was a concern for further scale-up in

Campaign 3. Additionally, it proved impossible to remove all of

the isopropyl acetate from the isolated and dried salt 11a,

although we did not determine whether this was due to a

solvate being formed. In Campaigns 1 and 2, 630 g and 7.1 kg

of 11a were prepared, respectively.

Although the bond forming chemistry had worked well we

wished to streamline the processing unit operations involved in

this step for Campaign 3 notably, removing a recharge of amine

to fully form the salt, and remove a break back to the free acid

before the penultimate amide bond formation (vide infra). A

salt screen was performed on 11 that focused initially on both

tertiary amines and metal salts with the intent of directly using

the salt in the following amide-forming step. The dicyclohex-

ylmethylamine salt 11b and the lithium salt were identified as

suitable candidates, but during laboratory trials the dicyclohex-

ylmethylamine salt 11b gave the cleanest downstream

processing and was selected for scale up in Campaign 3

(Scheme 7). After the ester was hydrolysed in NaOH/

methanol, as described above, the mixture was worked up

with citric acid and isopropyl acetate. After brine washing, the

isopropyl acetate was distilled and replaced into the

crystallization solvent methyl ethyl ketone (MEK), and

dicyclohexylmethylamine was added to yield, after filtration

and drying, a single batch of 44 kg of 11b equating to a yield of

Figure 4. Structure of C-4 epimer 33.

also be purged downstream in Campaign 3 if the initial level

could be minimized. This was achieved by using no additional

base, a −10 °C operating temperature, and an inverse addition

of the imidazolide solution to a slurry of ethylamine

hydrochloride in DCM. Use of these conditions resulted in a

reproducible 7% of epimer 33 being obtained in the product 32

solution. Approximately 50 kg of 32 was produced in this way

as a methanol solution in a single batch for Campaign 3

(Scheme 7).

Hydrolysis of the ethyl ester in 32 was accomplished using

sodium hydroxide. It was found that this reaction was also quite

sensitive to epimerization of the acidic chiral centre α to the

amide, and also β-elimination. Indeed, an inadvertent over-

charge of sodium hydroxide in Campaign 2 resulted in

significant levels of impurity formation that, although gave a

diminished yield, fortunately did not impact the downstream

chemistry greatly. We found that 11, whilst not being as

crystalline as the free acid, formed a crystalline salt 11a with

tert-butylamine, and thus an isolation process was developed.

The reaction mixture was acidified with hydrochloric acid at 40

°C, needing the slightly elevated temperature to avoid the

deposition of an immobile gum of the free-acid form of 11 in

the reactor. After removing most of the methanol by

distillation, the free acid was then extracted into isopropyl

acetate, and tert-butylamine was added to the wet organic

a

Scheme 8. Penultimate step, amide solvate preparations, Campaign 3

a

Reagents and conditions: (a) 11b, HCl (aq), DCM; (b) (i) CDI, DCM, −10 °C; (ii) 12, TEA, DCM, −10 °C; (iii) MeOH, water, 81%; (c)

MeCN, water, 95%; (d) (i) MeCN, 65%; (ii) Second crop MeCN, water, 29%.

475

dx.doi.org/10.1021/op200365n | Org. Process Res. Dev. 2012, 16, 470−483

Organic Process Research & Development

Article

68% over the three steps from the acetonide hydrogen sulphate

salt 13. Importantly, the level of the analogous C-4 epimer had

reduced to 4% in salt 11b from 7% in the ethyl ester 32.

Although the changes made to the transformation from 32 to

salt 11b meant that no extra charge of amine for salt formation

was required, it did not reduce the number of solvent swaps in

that part of the process. The main reduction in complexity was

designed to come from using the new salt 11b directly in the

amide formation (vide infra), removing the need for a salt break

and azeotropic drying of the resultant solution of the free acid

11.

single-crystal X-ray analysis (Figure 5). Gratifyingly, the

crystallization of the DCM solvate monohydrate 34a resulted

With the key units 11 and 12 in hand, the final stages of the

synthesis were to couple them together via an amide formation

(Scheme 8). In developing the coupling process for Campaigns

1 and 2, using the tert-butylamine salt 11a, the main challenge

was finding solvents that enabled sufficient solubility of all

components, avoiding the reformation of the ethyl ester 32 by

reaction of an activated derivative of 11 with the ethanol

present in the solid form of 12. Initially, the reaction was

carried out in DCM, but it was found that the salt break of the

tert-butylamine salt 11a using aqueous hydrochloric acid gave

poor phase separations that were attributed to the presence of

low levels of insoluble impurities. This problem was overcome

by carrying out the salt break using ethyl acetate as the organic

phase, and this had the added advantage that the solution of the

free acid could be easily dried by azeotropic distillation prior to

the amide-formation stage. Unfortunately, cooling the dry ethyl

acetate solution resulted in the deposition of 11 as a viscous

gum in the vessel. This was avoided by the addition of some

DCM to the solution once it had cooled to 35 °C. For the

dihydrochloride 12 the situation was more complicated since

the free base of 12 is highly soluble in aqueous systems even at

high ionic strengths, and is also quite insoluble in organic

solvents apart from DCM in which it is only modestly soluble.

Therefore, it was necessary to use the dihydrochloride salt 12

directly in the amide-forming step using an organic solvent and

an amine base to release the free base form of 12 in situ. This

was achieved by combining 12 with DIPEA in DCM, and then

slowly adding the ethyl acetate solution of the acyl imidazolide

derived from 11 at −10 °C. These conditions also suppressed

the formation of the ethyl ester from ethanol present in 12, due

to the slow reaction rate of ethanol with the acylimidazolide at

the low temperature used.

At the outset of development work for this step, the coupled

product 34 was not crystalline and we had planned to purify the

product using chromatography during Campaign 1. However,

we were unable to find a suitable and practical chromatography

system for purification of 34 prior to the final stage of the

synthesis. At the time the initial development work was carried

out, a crystalline form of the final API had also not been found;

thus, given our inability to find a suitable purification process

for the final intermediate 34, we planned to purify (by

chromatography) the API 1 itself and then isolate it as an

amorphous solid via precipitation in Campaign 1.

During the first lab scale-up, the coupling of 11 and 12 was

carried out in two batches using DCM alone. After aqueous

workup, the wet DCM solution containing the first batch of 34

was stored for a few days in a bottle whilst the second batch

was brought through. When we returned to the stored material,

we found that a large amount of solid had crystallized from the

wet DCM solution. A sample was collected, dried and found to

be a crystalline form of 34 that had crystallized as a DCM-

solvate monohydrate 34a, and the structure was confirmed by

Figure 5. Single-crystal X-ray structures for 34 crystallized from (a)

DCM giving 34a and (b) methanol giving 34b. Solvents and water

molecules are boxed in yellow. Pyridyl substituent is boxed in red.

in an excellent purge of impurities, and enabled us to complete

the Campaign 1 lab synthesis of 620 g of 34a without the need

to resort to chromatography. Furthermore, we found that

recrystallization of 34a using different solvents afforded two

different crystalline families of solvated forms, either mono-

solvate monohydrates (from DCM, isopropanol, acetone,

acetonitrile) or disolvates (from EtOH, MeOH) as determined

by single-crystal X-ray crystallography. Examination of the

structures showed that the conformation and internal H-

bonding arrays were similar for each family except for the

pyridine-bearing piperidine moiety. In both series, the

piperidine moiety adopted a pseudochair conformation, but

in the monosolvate monohydrate series, the pyridyl substituent

took an equatorial position (Figure 5a) whereas in the disolvate

family the piperidine adopted an axial position (Figure 5b).

Within each family of solvates the location of solvent molecules

was consistent, but the location did differ between each family.

Whilst the isolation of the DCM monosolvate monohydrate

34a proved to be advantageous for Campaign 1, the DCM

solutions of 34 generated during the workup of the coupling

process would have been supersaturated and potentially could

undergo undesirable crystallization from process solutions

during the workup stages if used again. For further scale-up

in Campaign 2, our attention focused on forming and

crystallizing the methanol disolvate 34b since this offered the

potential to significantly simplify the work up of the coupling

reaction as all the reaction byproduct (imidazole, DIPEA free

base and hydrochloride), residual starting materials, solvents

and many process-related impurities were soluble in methanol

and would potentially be purged. Once the coupling reaction

had reached completion, a simple exchange of solvents present

476

dx.doi.org/10.1021/op200365n | Org. Process Res. Dev. 2012, 16, 470−483

Organic Process Research & Development

Article

in the reaction mixture into methanol by atmospheric pressure

distillation resulted in the crystallization of the methanol

disolvate 34b which was then isolated by filtration. During

Campaign 2, this process worked effectively to deliver a total of

4.2 kg of 34b, but we found that some process-related

impurities were not purged to the extent needed, and thus it

was necessary to recrystallize this material twice from methanol

to remove these impurities, giving 3.7 kg of 34b ready for the

final step.

isolation, and acetonitrile−water recrystallizations to give 36.5

kg of 34c in 78% overall yield from acid salt 11b (Scheme 8).

The final stage of the synthesis was to remove the acetonide

protecting group in 34 and then to isolate the final API 1

(Scheme 9). The development of the workup stage was

a

Scheme 9. Isolation of amorphous API 1, Campaign 1

The presence of methanol in 34b was a potential problem

since the conditions ultimately selected for the final step used

methanesulphonic acid (vide infra), and we were concerned

about the possible formation of the potentially genotoxic

methyl methanesulphonate.¹⁷We found the methanol present

in the isolated 34b to be relatively loosely bound and could be

removed during the drying process without adversely impacting

the physical form of the solid, although the drying conditions

were prolonged and required high humidity in the drying oven.

This suggests that the methanol is replaced in the crystal lattice

by water, although this was not proven. Whilst using the

described process enabled the synthesis of 3.7 kg of methanol-

free 34 ready for the final step of Campaign 2, the multiple

recrystallizations and prolonged drying process would be

impractical on larger scales and required further development.

For Campaign 3, direct amide-coupling reactions of the

dicyclohexylmethylamine salt 11b with 12 on laboratory scale

all indicated that this would be a viable process; however, when

a sample of the bulk 11b was piloted in the laboratory, a good

conversion to amide 34 was observed, but crucially the

crystallization from methanol was impaired, giving a poor

recovery. In the time available the root cause of this failure was

not identified, and a pragmatic way forward had to be found. As

is very often the case, the supply of API was on critical path for

the program; thus, for expediency the salt break was

reintroduced as shown in Scheme 8, with all the attendant

solvent exchanges and drying stages. The salt 11b was broken

with aqueous 1 N HCl and extracted into DCM for direct use

in the CDI activation step. The side-chain hydrochloride salt 12

and TEA were then added to the acylimidazolide at 20 °C. A

final solvent exchange of the DCM to methanol and seeding

gave 40 kg of methanol disolvate 34b isolated in a single batch

in 81% yield.

In Campaign 2 the methanol disolvate 34b required

desolvation before being taken into the methanesulphonic

acid-mediated acetonide hydrolysis. The desolvation was

potentially an operational area of concern on the 10-fold

scale up for Campaign 3, and thus a recrystallization of 34b

from acetonitrile and water was introduced to give the product

as an acetonitrile monosolvate monohydrate 34c in 95%

recovery (Scheme 8).

To reach the desired purity specification a second

recrystallization from acetonitrile was required during which

extra water was not added. This recrystallization gave a poor

65% recovery, and a second crop was obtained, this time by

reintroducing a small amount of water to the mother liquor,

thereby increasing the water activity, to crystallize a further 29%

material. In total 34c was isolated in 94% yield over the two

crops (see Experimental Section for details). Importantly, at

this stage the analogous C-4 ribose epimer was not detectable,

having been purged as anticipated during the downstream

processing from 11. Approximately 44 kg of acid salt 11b was

taken through the amide formation, methanol disolvate 34b

a

Reagents and conditions: (a) (i) TFA, H₂O, butan-2-ol, 70 °C; (ii)

NaOH, H₂O; (iii) TBME, 53%.

constrained somewhat by the low solubility of 1 in all relatively

volatile anhydrous organic solvents that might be useful for

phase separations during an aqueous workup. Fortunately,

polar, water-immiscible organic solvents that had the capability

of retaining significant amounts of water in the organic phase

(such as butan-2-ol, MEK, and methyl acetate) proved to be

effective solvents for 1, and thus development work focused

around these systems. Initially, deprotection of 34 was achieved

using hydrochloric acid in ethanol at 60 °C, but these

conditions resulted in significant cleavage of the nucleoside

bond and also had the risk of generating ethyl chloride, a

potential genotoxic impurity. We therefore screened different

conditions and found that aqueous TFA was an effective

reagent that avoided the formation of potentially genotoxic side

products when used with alcoholic solvents. These conditions

were selected for use in Campaign 1. Heating the DCM solvate

monohydrate 34a in a mixture of 5.3 equiv of TFA in a mixture

of water and butan-2-ol at 70 °C resulted in the deprotection of

the acetonide to give 1 (Scheme 9), however we observed the

formation of several impurities as the reaction proceeded

(products from hydrolysis of the nucleoside bond, hydrolysis of

both amide bonds, and solvolysis of the ethylamide with butan-

2-ol being the main degradants) and purity became

progressively worse with time. In addition, unsurprisingly, the

formation of but-2-yl trifluoroacetate in significant quantity was

observed. Following workup, which involved pH adjustment,

multiple water washes, and azeotropic drying of the solution, a

relatively pure solution of 1 in butan-2-ol was produced in

Campaign 1. However, a method to isolate amorphous material

from the butan-2-ol solution was required since a crystalline

form of the product had not been found at this point. To this

end, we screened the addition of the butan-2-ol solution to

various antisolvents to determine whether precipitation and

isolation of an amorphous solid was viable. By using tert-butyl

methyl ether (TBME) as the antisolvent, a filterable precipitate

was formed on gram scale that on drying afforded the product

in reasonable yield and purity. When this method was applied

to the combined solutions from Campaign 1, the precipitation,

filtration, and tert-butyl methyl ether washing stages worked as

expected, but when the material was dried in a vacuum oven,

the solid collapsed to a gum that subsequently expanded to a

voluminous foam. This was due to inefficient removal of butan-

2-ol from the solid during the washing process. Indeed, it

proved impossible to remove all traces of residual butan-2-ol

from the product even in a very prolonged drying process, and

477

dx.doi.org/10.1021/op200365n | Org. Process Res. Dev. 2012, 16, 470−483

Organic Process Research & Development

Article

a

Scheme 10. Isolation of crystalline API 1, Campaign 3

a

Reagents and conditions: (a) (i) MSA (aq), 70 °C; (ii) Na₂HPO₄(aq), MeOAc; (iii) Azeotropically dry, seed and crystallize at reflux, 78%.

the overall yield from 34a was low (53%). Nevertheless, 313 g

of 1 of suitable quality was prepared in Campaign 1, but it was

recognized that the conditions used for the deprotection and

isolation were far from ideal and required further development

for Campaigns 2 and 3.

azeotropic distillation with the aim of reducing the water level

to 2%. However, when a sample of the dried solution was taken

and examined by H spectroscopy, new impurities, strongly

1

suspected to be MEK self-condensation products, were evident.

Moreover, it proved impossible to crystallize 1 from this

solution, and MEK was abandoned and effort was focussed on

other solvents. From previous screens it was known that 1 only

dissolved in water-immiscible solvents (e.g., butan-2-ol, MEK)

saturated with several percent of water. In addition, a solvent

that could be azeotropically dried to adjust the water content to

the correct level was needed. On this basis methyl acetate

(dissolves approximately 8% w/w water when saturated,

azeotrope contains 5% water at 56 °C) and ethyl acetate

(dissolves approximately 3.3% w/w water when saturated,

azeotrope contains 8% water at 71 °C) were selected for further

investigation.¹⁸Wet ethyl acetate was immediately ruled out as

1 was insoluble, even in 50 volumes. However, water-saturated

methyl acetate easily dissolved 1 in 10 volumes. Methyl acetate,

given its low boiling point (57 °C) and flash point (−9 °C) is

not a preferred solvent for large-scale use, but the significant

advantages it offered to the workup and crystallization process

outweighed the potential disadvantages, and so it was cautiously

selected for further development.

Initial investigations for the use of methyl acetate in the

workup of the deprotection reaction were hampered by poor

phase separations, but we found that using 10% w/w disodium

hydrogen phosphate for the salt break and extraction with

methyl acetate followed by a wash of the organic phase using a

2% w/w disodium hydrogen phosphate solution provided

sufficient ionic strength to enable reasonable phase separations

to occur. The organic phase was then diluted with methyl

acetate and azeotropically dried until the water content was

approximately 2% w/w. Heating the resulting solution at reflux

over 20 h then resulted in crystallization of 1 in good yield.

The final stages of the synthesis were scaled up in our kilo lab

facility for Campaign 2, with the deprotection stage being run

in two batches to give solutions of 1 in methyl acetate that were

combined for the crystallization stage. The deprotection step

worked largely as expected but some problems were

encountered with the crystallization process with oiling

occurring. Adjustment of the water content to 2.2% and the

addition of seed crystals of 1 and heating the mixture for 20 h

eliminated the oiling altogether, and following filtration,

washing, and drying, Campaign 2 afforded a total of 2.52 kg

of 1 in 72% yield from the desolvated acetonide 34b.

Fortunately, sometime after the completion of Campaign 1, a

crystalline form of 1 was discovered,^5band we were able to

develop a crystallization process for Campaign 2 as well as

alternative conditions for the deprotection stage. To find

alternative conditions for the acetonide deprotection, we were

keen to avoid the use of organic solvents in the reaction stage

since we were worried about reactions occurring between the

acid promoter and a solvent, leading to potentially genotoxic

impurities. We envisioned that, by utilising the basic 2-

aminopyridine group in the molecule, it might be possible to

combine superstoichiometric amounts of an acid with 34 to

form a water-soluble salt that would enable the reaction to be

carried out in water without the addition of an organic solvent.

When this was tried, some acids (TFA, p-toluenesulphonic

acid) resulted in the formation of gums, whereas reacting 1.1

equiv of aqueous methanesulphonic acid (MSA) with the

acetonide 34 at 70 °C gave a clear solution that underwent

deprotection over 24 h. To minimise degradation reactions

occurring once the conversion was complete, the reaction was

stopped by neutralization of the 0.1 M excess methanesul-

phonic acid by the addition of 0.12 equiv of aqueous solution of

disodium hydrogen phosphate at 70 °C. On cooling the

reaction mixture, the methanesulphonate salt of 1 remained in

solution and enabled the workup to be undertaken without

formation of unwanted gums or solids.

The extraction solvent used during the workup would

necessarily depend on the solvent to be used for the

crystallization of 1. The discovery of a crystalline form of 1

was a major breakthrough for the project, but it took a

considerable time to develop a workable crystallization process.

After much experimentation, it emerged that the key features

needed to crystallize 1 were a polar nonhydroxylic solvent

containing a few percent of water, and conducting the

crystallization at elevated temperature. The latter requirement

is quite unusual, but without heating, supersaturated solutions

would either remain as clear solutions or would precipitate

amorphous gums. The first set of reliable conditions that were

identified was to prepare a solution of 1 in 10 volumes of a 2%

v/v solution of water in MEK and to heat this solution at 70 °C

over several hours over which time 1 crystallized in reasonable

yield. The use of MEK was tested for the workup of the

deprotection step by taking the aqueous solution of the

methanesulphonate salt of 1, adding MEK, and neutralising the

mixture with sodium hydrogen carbonate. The resulting wet

MEK solution of 1, after washing with water, was dried by

For Campaign 3, the only significant change to the last

chemical step and API crystallization was the use of 34c as an

input material in preference to desolvated 34b. The mixed

acetonitrile water solvate 34c was taken up in aqueous MSA

and heated at 70 °C for 10 h and worked up as before. The API

478

dx.doi.org/10.1021/op200365n | Org. Process Res. Dev. 2012, 16, 470−483

Organic Process Research & Development

Article

(1) was crystallized by azeotropically drying the methyl acetate

solution to 2.4% w/w water, seeding, and holding at reflux for

at least 24 h. The final API, UK-432,097, 1 of appropriate

purity and polymorphic form was isolated as a 25.3-kg single

batch in 78% from 34c (Scheme 10).

collected by filtration. The filtrate was used as a secondary cold

wash of the reaction vessel and refiltered on the original cake

cold. The cake was washed with chilled ethyl acetate (42 kg).

The product was dried on the filter at 61 °C for 45 h under s

stream of nitrogen, giving 23 as a white solid 30.2 kg (83.0 mol,

90%).

N-(2-Aminoethyl)-N′-[1-(2-pyridyl)-4-piperidyl]urea Hy-

drochloride Ethanol Solvate (12). Acetyl chloride (38.9 kg,

496 mol) was added to a solution of BOC protected amine 23

(30.0 kg, 82.5 mol) in absolute ethanol (236 kg) at 45 °C. The

reaction mixture was heated to 45 °C for 2 h and was cooled to

room temperature over 2 h during which time a suspension was

formed. The mixture was stirred for 2 h, the solid was collected

by filtration, and the filter cake was washed with cold (0−5 °C)

absolute ethanol (2 × 47 kg). The solid was dried on the filter

for 3 h and then under a stream of nitrogen for 96 h to give the

ethanol solvate 12 as a colourless hygroscopic solid, (27.6 kg,

85%). ¹H NMR (400 MHz, DMSO-d₆, 70 °C): δ 8.02 (3H, br

s,), 7.94 (2H, m), 7.35 (1H, d, J = 11.8 Hz), 6.89 (1H, t, J = 6.4

Hz), 4.33 (br s), 4.14 (3H, br d, J = 13.9 Hz), 3.7 (1H, m), 3.40

(2H, t, J = 11.1 Hz), 3.23 (H, t, J = 6.1 Hz), 2.81 (2H, q, J = 5.9

Hz), 1.89 (2H, m), 1.46 (2H, m). ¹³C NMR (100 MHz,

DMSO-d₆, 30 °C): δ 158.5, 151.8, 144.5, 137.3, 113.3, 112.8,

45.9, 45.8, 37.9, 31.8.

SUMMARY

■

We have identified and developed a new synthetic route to the

structurally complex A_2aagonist UK-432,097 (1) that was used

to prepare over 25 kg of API to support early phase

development. The new route addressed the issues associated

with the medicinal chemistry route by oxidizing the ribose C-5

hydroxyl position and elaborating it to the ethylamide, post

nucleoside bond formation, and also resulted in the discovery

of a robust crystalline form of the API. The key steps involved

in the process were (1) a stereoselective glycosidation reaction;

(2) a scalable bleach−TEMPO oxidation; and (3) an unusual

elevated-temperature crystallization process for the final API.

EXPERIMENTAL SECTION

■

All materials obtained from commercial suppliers were used

without further purification unless otherwise stated. All

reactions were performed under an inert nitrogen atmosphere.

Melting points are uncorrected. NMR spectra were recorded on

a Jeol 400 MHz instrument: ¹H NMR at 70 °C in DMSO-d₆as

the solvent unless otherwise indicated. Chemical shifts are

reported in ppm (δ) relative to residual protons in the

deuterated solvent. HPLC chromatograms were recorded on

Aligent 1100 instruments, and LCMS data were recorded on an

Agilent 1100 series running a Waters Micromass ZQ mass

spectrometer. Two HPLC methods were used.

Acidic method: Zorbax CB C18 column, 50 mm × 3 mm,

1.8 μm column. Flow rate = 1.2 mL/minute, wavelength 225

nm, temperature 70 °C. Eluent A = MeCN. Eluent B = 0.05%

TFA in water. Eluent ramp (time, A:B ratio), 0 min, 5:95; 3.5

min, 100:0; 4.5 min, 100:0; 4.6 min, 5:95; 5 min, stop.

Basic method: Zorbax Extend CB C18 column, 50 mm × 3

mm, 1.8 μm column. Flow rate = 1.2 mL/minute, wavelength

225 nm, temperature 70 °C. Eluent A = MeCN. Eluent B =

0.1% NH₄OH in water. Eluent ramp (time, A:B ratio), 0 min,

5:95; 3.5 min, 100:0; 4.5 min, 100:0; 4.6 min, 5:95; 5 min, stop.

Representative Examples of Campaign 3 Outsourced

Chemistry. tert-Butyl 2-{N′-[1-(2-pyridyl)-4-piperidyl]-

ureido}ethylcarbamate (23). Amine salt 21^5c,9(23.0 kg, 91.9

mol) and 1,1′-carbonyldiimidazole (16.4 kg, 101 mol) were

suspended in acetonitrile (118 kg) and cooled to 5 °C. DIPEA

(24.4 kg, 189 mol) was added, with a final acetonitrile (9 kg)

line rinse, over 6 h with maximum stirring, keeping the

temperature below 10 °C, giving a solution. Upon full

consumption of 21, the mixture was warmed to room

temperature, and a solution of N-BOC-ethylene diamine

(15.5 kg, 96.7 mol) in acetonitrile (9 kg) was added and

then heated to reflux for 2 h. Upon reaction completion

acetonitrile (110 kg) was removed in vacuo and replaced with

DCM (230 kg). The organic phase was washed with water (2 ×

92 L). The combined DCM phases were concentrated (145 kg

solvent removed), further distilled, and replaced with ethyl

acetate (5 × 139 kg) with a further 558 kg of solvent being

removed. The ethyl acetate solution was cooled to 55 °C, and

the resulting slurry was granulated at 55 °C for 6.5 h. The slurry

was cooled to 5 °C over 6 h upon which the thick slurry was

stirred at that temperature for 2 h, and the product was

2-Chloro-N-(2,2-diphenylethyl)-9-(tetrahydro-2H-pyran-2-

yl)-9H-purin-6-amine (24). For representative experimental

details of conversion of dichloropurine to chloropurine

derivative 24 see reference 8b experimental section, compounds

13 and 14.

Ethyl 6-[(2,2-Diphenylethyl)amino]-9-(tetrahydro-2H-

pyran-2-yl)-9H-purine-2-carboxylate (25). 1,1′-Bis-

(diphenylphosphino)ferrocene (7.00 g, 12.6 mmol) was

added to a suspension of palladium acetate (1.50 g, 6.68

mmol) in absolute ethanol (1 L), and the resultant suspension

was stirred under an atmosphere of nitrogen for 18 h to give

the catalyst mixture. Anhydrous sodium carbonate (94 g, 887

mmol) was added to a solution of chloropurine 24 (700 g, 1.61

mol) in ethanol (4.5 L) in an autoclave, and the catalyst

mixture was added under nitrogen. The autoclave was flushed

twice with carbon monoxide and then pressurised to 20 bar

with carbon monoxide. The mixture was then heated at 105 °C

with stirring for 10 h, and the autoclave was then vented,

flushed with carbon monoxide, and then repressurised to 20 bar

of carbon monoxide. Heating at 105 °C was continued for a

further 14 h upon which the mixture was cooled to 60 °C and

filtered through a bed of warm Celite. The resultant filtrate was

allowed to cool to ambient temperature, whereupon crystal-

lization occurred, and after stirring at this temperature for 7 h,

the resultant suspension was filtered. The filter cake was

washed with cold ethanol (500 mL), and the solid was dried in

vacuo for 24 h at 55 °C to give ethyl ester 25 as a cream-

1

coloured solid (575 g, 76%), mp 138−140 °C. H NMR (300

MHz, CDCl₃, 30 °C) δ: 8.05 (1H, s), 7.45−7.15 (10H, m),

5.95−5.80 (2H, m), 4.60−4.30 (5H, m), 4.15 (1H, br d), 3.80

(1H, br t), 2.20−1.60 (6H, m), 1.50 (3H, t).

Ethyl 6-[(2,2-Diphenylethyl)amino]-9H-purine-2-carboxy-

late (16). Trifluoroacetic acid (73.5 g, 0.645 mol) was added

to a suspension of ethyl ester 25 (250 g, 0.530 mol) in absolute

ethanol (1.25 L), and the resultant mixture was heated at 50 °C

for 20 h. The mixture was then cooled, and the solid was

collected by filtration. The filter cake was washed with absolute

ethanol (350 mL) and was dried in vacuo at 50 °C overnight to

give purine derivative 16 (206.5 g, 99%) as a cream-coloured

479

dx.doi.org/10.1021/op200365n | Org. Process Res. Dev. 2012, 16, 470−483

Organic Process Research & Development

Article

fine powder, mp >200 °C. ¹H NMR (300 MHz, DMSO-d₆, 30

°C) δ: 8.36 (1H, br s), 8.00 (1H, br t), 7.48−7.12 7 (10H, m),

4.80−4.00 (5H, br m), 1.48−1.22 (3H, br m).

74.2, 71.2, 62.1, 61.8, 50.1, 45.2, 14.7, (Residual ⁱPrOAc signals,

67.5, 22.13, 21.6). HPLC (basic method) retention time 2.72′.

LCMS m/z: 520.2 (MH⁺).

N-(2,2-Diphenylethyl)-2-(ethoxycarbonyl)adenosine (29).

N-Methylmorpholine (1.44 kg, 14.2 mol) was charged to a

suspension of 16 (5.00 kg, 12.9 mol) in anhydrous DME (15

L) at 22 °C. The mixture was then heated to 50 °C, and a

solution of trimethylsilyl trifluoromethanesulphonate (3.73 kg,

16.8 mol) in DME (6.25 L) was then added at such a rate as to

maintain the reaction temperature between 50 and 60 °C. The

temperature of the reaction mixture was then adjusted to 65 °C,

and a solution of 1,2,3,5-tetra-O-acetyl-β-^D-ribofuranose (15)

(4.52 kg, 14.2 mol) in DME (17.5 L) was added over a period

of 15 min. The reaction mixture was stirred at 65 °C for 90 min

and upon reaction completion was quenched at 65 °C over 15

min with an aqueous solution of trisodium citrate monohydrate

(1.90 kg, 6.46 mol) and citric acid monohydrate (1.31 kg) in

water (12.5 L). The DME was mostly removed by distillation

(initially under vacuum) and replaced with isopropyl acetate

(25 L). After cooling to 25 °C the phases were separated, and

the upper organic phase was washed with water (2 × 12.5 L)

and saturated aqueous sodium bicarbonate solution (12.5 L).

The organic phase was then diluted with isopropyl acetate (45

L), and the solution was azeotropically dried by distillation at

atmospheric pressure until approximately 45 L of distillate had

been collected, and the organic solution contained no more

than 0.03% w/w water by Karl Fischer analysis. The mixture

was then concentrated to a volume of approximately 18 L by

distillation at atmospheric pressure, and absolute ethanol (60

L) was added. The mixture was then further concentrated by

distillation at atmospheric pressure until the volume was

approximately 43 L. The resultant solution of triacetate 14 was

then cooled to 20 °C, and a solution of 1,8-diazabicyclo[5.4.0]-

undec-7-ene (1.96 kg, 12.9 mol) in absolute ethanol (7 L) was

added over 15 min and was stirred for 3 h upon which the

mixture was slowly added to a solution of citric acid (2.71 kg,

12.9 mol) in water (37.5 L), and the equipment was rinsed

through with ethanol (2 L). The resulting mixture was then

concentrated to a volume of approximately 55 L by distillation

(initially under vacuum). The mixture was then cooled to 25

°C, it was then extracted with isopropyl acetate (75 L), and the

organic phase was washed with 0.1 M aqueous sodium

bicarbonate solution (2 × 37.5 L). The solution was then

diluted with isopropyl acetate (25 L) and then azeotropically

dried by distillation at atmospheric pressure until the final

volume was 32 L. The solution was then cooled to 20 °C over 3

h, seeded with crystals of 29 (10 g), aged at 20 °C for 18 h, and

was then cooled to 2 °C for 2 h. The solids were collected by

filtration, washed with cold isopropyl acetate (2 × 10 L), and

were then dried in vacuo at 50 °C for 48 h to give 29 (6.05 kg,

11.64 mol, 90%) as an off-white solid that contained

approximately 4% w/w of entrained residual isopropyl acetate.

¹H NMR (400 MHz, DMSO-d₆, 70 °C) δ: 8.20 (1H, br s), 7.84

(1H, br s), 7.32 (4H, br d, J = 7.1 Hz), 7.23 (4H, br t, J = 7.3

Hz), 7.14 (2H, br t, J = 7.3 Hz), 5.89 (1H, d, J = 7.1 Hz), 5.22

(1H, d, J = 5.9 Hz), 4.94 (1H, d, J = 4.6 Hz), 4.81 (1H, t, J =

5.7 Hz), 4.62 (1H, br s), 4.56 (1H, dd, J = 11.1, 6.1 Hz), 4.32

(2H, m), 4.15 (1H, dd, J = 8.1, 4.9 Hz), 3.94 (1H, dd, J = 6.9,

3.7 Hz), 3.66 (1H, ddd, J = 12.0, 4.9, 4.3 Hz), 3.57 (1H, ddd, J

= 12.0, 6.6, 4.2 Hz), 1.33 (3H, br t, J = 6.9 Hz). (Residual

ⁱPrOAc signals, 4.84 (hept, J = 6.3 Hz), 1.92 (s), 1.14 (d, J = 6.3

Hz)). ¹³C NMR (100 MHz, DMSO-d₆, 30 °C): δ 164.4, 154.7,

151.1, 149.1, 143.1, 142.0, 128.9, 128.7, 126.9, 120.8, 87.9, 86.6,

Campaign 3 In-House Preparations. N-(2,2-Dipheny-

lethyl)-2-(ethoxycarbonyl)-2′,3′-O-isopropylidene Adenosine

Hydrogen Sulphate (13). Sulphuric acid (98% w/w; 5.73 kg,

58.5 mol, 0.95 equiv) was added over a period of 10 min to a

suspension of triol 29 (32.0 kg, 61.6 mol, 1 equiv) in acetone

(149 L) and 2,2-dimethoxypropane (12.83 kg, 123 mol, 2

equiv). The sulphuric acid addition line was rinsed through

with acetone (10 L), and the resultant solution was stirred at 20

°C for 50 min after which a thick slurry had formed. The

mixture was filtered and the cake washed with acetone (32 L).

The solid was dried in vacuo at 50 °C overnight to yield

1

sulphate salt 13 as a tan solid (35.6 kg, 54.1 mol, 88%). H

NMR (400 MHz, DMSO-d₆, 70 °C): 8.41 (1H, br s), 7.88 (1H,

br s), 7.31 (4H, m), 7.24 (4H, m), 7.14 (2H, m), 6.14 (1H, d, J

= 2.4 Hz), 5.28 (1H, dd, J = 6.1, 2.4 Hz), 4.99 (10H, br s), 4.62

(1H, br s), 4.32 (2H, m), 4.17 (1H, m), 3.57 (1H, dd, J = 11.7,

5.4 Hz), 3.52 (1H, dd, J = 11.8, 5.0 Hz), 1.52 (3H, s), 1.33 (3H,

br t, J = 7.1 Hz), 1.30 (3H, s). ¹³C NMR (100 MHz, DMSO-d₆,

30 °C): δ 164.3, 154.6, 151.2, 148.6, 143.1, 142.1, 128.9, 128.1,

126.9, 120.4, 113.6, 89.7, 87.8, 84.3, 81.9, 62.2, 61.8, 50.1, 45.2,

27.6, 25.8, 14.6 HPLC (basic method) retention time = 3.25′.

LCMS m/z: 560.1 (Parent MH⁺).

Ethyl 6-[(2,2-Diphenylethyl)amino]-9-(2,3-O-isopropyli-

dene-β-^D-ribofuranuronosyl)-9H-purine-2-carboxylate (31).

Safety testing data. DSC testing of TEMPO showed strong

exothermic decomposition from 196 °C, whereas sulphate salt

13 showed only a melting endotherm from 142 °C. Reaction

calorimetry showed no accumulation at any stage, but the

addition of bleach solution showed a dose-controlled exotherm

of −275 kJ/mol. TSU screening of the reaction mixture after

addition of bleach showed no thermal events up to 176 °C, and

screening of the final DCM solution before azeotropic drying

showed no thermal events up until 200 °C.

An aqueous sodium hydrogen carbonate solution (13.7 kg

dissolved in 164 L water) was added to a stirred slurry of

sulphate salt 13 (35.6 kg, 54.06 mol, 1 equiv), TEAB (11.4 kg,

54.06 mol, 1 equiv), and TEMPO (144 g, 0.922 mol, 1.7 mol

%) in DCM (355 L). The mixture was cooled to 2 °C, and

aqueous sodium hypochlorite (13% w/w; 71.8 kg, 127 mol,

2.35 equiv) was added over 60 min at such a rate that the

internal temperature did not rise above 5 °C, and the mixture

was stirred for 2 h. Upon reaction completion the mixture was

warmed to 20 °C, and aqueous sodium sulphite (35.6 kg in 356

L of water) was added and stirred overnight. Upon a negative

test for active oxidant the phases were separated, and the

retained organic phase was washed with 1 M hydrochloric acid

(360 L). A subsequent wash with aqueous sodium chloride (71

kg NaCl in 356 L water) gave an emulsion that required

overnight separation and settling. The aqueous sodium chloride

phase was back extracted with DCM (178 L), requiring another

slow separation. The DCM extracts were combined and

azeotropically distilled until the water content was not more

than 0.05% w/w and the product held as a solution. A small

sample of the solution of 31 was evaporated to a gum for data

collection. ¹H NMR (400 MHz, DMSO-d₆, 90 °C): δ 8.30 (1H,

s), 7.63 (1H, br s), 7.31 (4H, br d, J = 7.6 Hz), 7.25 (4H, m),

7.15 (2H, m), 6.32 (1H, s), 5.66 (1H, dd, J = 6.1, 2.2 Hz), 5.46

(1H, br d, J = 6.1 Hz), 4.61 (2H, m), 4.31 (2H, br q, J = 7.1

Hz), 4.23 (2H, br m), 3.00 (br s), 1.51 (3H, s), 1.34 (6H, m).

¹³C NMR (100 MHz, DMSO-d₆, 30 °C): δ 171.1, 164.0, 154.7,

480

dx.doi.org/10.1021/op200365n | Org. Process Res. Dev. 2012, 16, 470−483

Organic Process Research & Development

Article

150.4, 148.9, 143.1, 143.0, 128.9, 128.7, 126.9, 119.4, 113.0,

89.9, 86.6, 84.6, 84.3, 61.6, 50.0, 45.3, 27.0, 25.4, 14.6. HPLC

(acidic method) retention time = 3.28′;. LCMS m/z: 574.4

(MH⁺).

added over 15 min and washed through with isopropyl acetate

(11 L). The solution was dried and ambient pressure distilled,

and solvent was replaced; 16.5 L was removed and replaced

with fresh isopropyl acetate (11 L), and then a further 9 L of

solvent was removed. At this point a check of the reaction

mixture showed tert-butylamine had been lost during the

distillation, so a further charge of tert-butylamine (320 mL) in

isopropyl acetate (1 L) was made. The mixture was cooled to

20 °C over 3 h, then to 0 °C over 2 h to give a stirrable slurry.

The product was filtered, washed on the filter with chilled

isopropyl acetate (4.4 L), and dried in vacuo at 50 °C for 48 h

Ethyl 6-[(2,2-Diphenylethyl)amino]-9-(N-ethyl-2,3-O-iso-

propylidene-β-^D-ribofuranosyluronamide)-9H-purine-2-car-

boxylate (32). Safety testing data. DSC testing of ethylamine

hydrochloride showed a melting endotherm from 77 °C and a

further endotherm from 313 °C, whereas acid 31 showed an

endotherm from 56 °C followed by a strong exotherm from

185 °C. Reaction calorimetry showed no accumulation at any

stage and no significant energy release at any stage. TSU

screening of the reaction mixtures: after addition of CDI, after

addition of the imidazolide to the amine salt, and after the final

solution was concentrated, all showed no thermal events up to

185 °C.

1,1′-Carbonyldiimidazole (16.28 kg, 100.4 mol, 1.2 equiv)

was added to a stirred solution of acid 31 (∼48.0 kg, 83.6 mol,

1 equiv) in DCM (480 L) at room temperature. Upon

completion of acylimidazolide formation (approximately 3 h)

the mixture was cooled to 5 °C. In a separate vessel a slurry of

ethylamine hydrochloride (8.2 kg, 100 mol, 1.2 equiv) in DCM

(144 L) was prepared and cooled to −15 °C. The

acylimidazolide solution was then added, followed by DCM

(30 L) line rise to the cooled salt slurry at such a rate to

maintain the internal temperature below 0 °C. The reaction

mixture was then stirred at 0 °C for 16 h, whereupon the

mixture was warmed to 20 °C and diluted with methanol (360

L). The mixture was distilled at ambient pressure to leave

approximately 480 L vessel volume, and the solvent was

replaced with fresh methanol (360 L). The vessel contents were

again distilled to leave approximately 480 L volume reaching a

final internal temperature of 63 °C. The methanol solution was

held in drums in readiness for the preparation of salts of acid 11

(vide infra). A small sample was evaporated to a gum for data

collection. ¹H NMR (400 MHz, DMSO-d₆, 70 °C): δ 8.33 (1H,

br s), 7.86 (1H, br s), 7.30 (4H, d, J = 7.6 Hz), 7.24 (4H, m),

7.14 (2H, m), 6.33 (1H, br s), 5.48 (1H, br d, J = 6.1 Hz), 5.37

(1H, br d, J = 6.1 Hz), 4.62 (1H, br s), 4.48 (1H, s), 4.30 (2H,

m), 4.15 (2H, br m), 2.74 (2H, m), 1.51 (3H, s), 1.32 (6H, m),

0.57 (3H, br t, J = 7.1 Hz). ¹³C NMR (100 MHz, DMSO-d₆, 30

°C): δ 168.6, 164.2, 154.7, 150.9, 148.7, 143.1, 142.6, 128.9,

128.7, 126.8, 120.5, 113.2, 89.6, 87.2, 84.2, 83.9, 61.6, 50.0,

45.2, 33.5, 27.2 25.6, 14.6, 14.3. HPLC (acidic method)

retention time = 3.36′; (basic method) retention time = 3.26′.

LCMS m/z: 601.4 (MH⁺).

6-[(2,2-Diphenylethyl)amino]-9-(N-ethyl-2,3-O-isopropyli-

dene-β-^D-ribofuranosyluronamide)-9H-purine-2-carboxylic

Acid tert-Butylamine Salt (11a). Sodium hydroxide, 1 M

solution (8.4 L), was charged over 30 min to a 5 °C solution of

ester 32 (4.57 kg, 7.61 mol) in methanol (∼23 L) and stirred

for 1 h. Upon reaction completion the reaction was warmed to

40 °C and slowly neutralized to pH 6−7 with 1 M hydrochloric

acid (2 L) and solvent removed (20 L) by distillation at

ambient pressure. Whilst at reflux isopropyl acetate (23 L) was

added, and a further 23 L of solvent removed by distillation.

Fresh isopropyl acetate (23 L) was added whilst the mixture

was above 50 °C; the mixture was cooled to 20 °C, and water

(9 L) was added. The mixture was acidified to pH 1−2 with 2

M hydrochloric acid (3 L). The phases were separated, and the

organic phase was washed with water (14 L). The organic

phase was returned to the reaction vessel, and a solution of tert-

butylamine (670 g, 9.3 mol) in isopropyl acetate (4.5 L) was

1

to give salt 11a as a white solid (4.1 kg, 6.35 mol, 83%). H

NMR (400 MHz, DMSO-d₆, 70 °C): δ 8.18 (1H, br s), 7.75

(1H, br s), 7.30 (4H, br d, J = 7.6 Hz), 7.24 (4H, br t, J = 7.6

Hz), 7.14 (2H, br t, J = 7.1 Hz), 6.23 (1H, d, J = 2.5 Hz), 5.29

(1H, dd, J = 6.0, 2.8 Hz), 5.25 (1H, dd, J = 6.0, 2.8 Hz), 4.56

(1H, br t, J = 7.6 Hz), 4.47 (1H, d, J = 2.4 Hz), 4.21 (2H, br s),

2.95, (2H, br quin, J = 7.1 Hz), 1.52 (3H, s), 1.31 (3H, s), 1.18

(9H, s), 0.76 (3H, t, J = 7.1 Hz). ¹³C NMR (100 MHz, DMSO-

d₆, 30 °C): δ 170.0, 168.7, 161.5, 154.7, 149.1, 143.4, 140.0,

128.9, 128.6, 126.8, 114.0, 89.5, 85.4, 83.5, 83.3, 50.8, 50.1,

44.7, 33.9, 28.0, 27.4, 25.7, 14.9. HPLC (basic method)

retention time 2.08′; LCMS m/z: 573.3 (Parent MH⁺).

6-[(2,2-Diphenylethyl)amino]-9-(N-ethyl-2,3-O-isopropyli-

dene-β-^D-ribofuranosyluronamide)-9H-purine-2-carboxylic

Acid Dicyclohexylmethylamine Salt (11b). Safety testing

data. DSC testing of the solution post brine washing showed

only endothermic activity past 313 °C. Product salt 11b

showed an endothermic melt from 166 °C and further

endothermic activity past 319 °C. Reaction calorimetry showed

no accumulation at any stage and the only significant energy

release of −246 kJ/mol on neutralization of the sodium salt

with aqueous citric acid, but the addition of bleach solution

showed a dose-controlled exotherm of −275 kJ/mol. TSU

screening of the organic phase after addition of aqueous citric

acid showed no thermal events up to 179 °C, and screening of

the organic solution before distillative removal of methanol

showed no thermal events up until 199 °C.

A solution of aqueous sodium hydroxide (20.9 kg, 208.9 mol,

2.5 equiv) in water (213 L) was cooled to 10 °C and added

slowly to a solution of ester 32 (∼50.2 kg, 83.6 mol, 1.0 equiv)

in methanol (∼502 L), maintaining the temperature at or

below 0 °C. The solution was warmed to 20 °C and stirred for

3 h. The reaction mixture was diluted with isopropyl acetate

(251 L), and aqueous citric acid (10% w/vol, 250 L) was

added, whereupon the mixture was heated to reflux and 540 L

of distillate was removed. Isopropyl acetate (250 L) was added

to the remaining solution which was cooled to 20 °C, and the

phases were separated. The organic phase was washed with

sodium chloride solution (20% w/vol, 250 L). The combined

aqueous washes were back extracted with isopropyl acetate

(250 L), and subsequently washed with citric acid solution

(10% w/vol, 250 L). The isopropyl extracts were combined and

distilled under reduced pressure to low volume (∼100 L), and

methyl ethyl ketone (251 L) was added. This distillation and

replacement was repeated with the final volume made up to

approximately 275 L. The solution was held at 50 °C and

dicyclohexylmethylamine (13.1 kg, 66.9 mol, 0.8 equiv) was

added and the mixture cooled to 18 °C and stirred for 16 h.

The mixture was filtered and the solid washed with methyl

ethyl ketone (100 L) and dried at 50 °C in vacuo to give salt

1

11b as a white solid (43.9 kg, 57.2 mol, 68%). H NMR (400

MHz, DMSO-d₆, 70 °C): δ 8.29 (1H, br s), 7.63 (1H, br m),

481

dx.doi.org/10.1021/op200365n | Org. Process Res. Dev. 2012, 16, 470−483

Organic Process Research & Development

Article

7.50 (1H, br m), 7.30 (4H, d, J = 7.2 Hz), 7.24 (4H, t, J = 7.2

Hz), 7.13 (2H, t, J = 7.2 Hz), 6.30 (1H, br s), 5.37 (1H, br d, J

= 6.1 Hz), 5.33 (1H, br d, J = 6.1 Hz), 4.58, (1H, br m), 4.50

(br s), 4.21 (2H, br s), 2.83 (2H, m), 2.61 (2H, m), 2.23 (3H,

s), 1.71 (8H, m), 1.54 (2H, br s), 1.52 (3H, s), 1.32 (3H, s),

1.21 (8H, m), 1.05 (2H, m), 0.63 (3H, s). ¹³C NMR (100

MHz, DMSO-d₆, 30 °C): δ 168.8, 168.3, 157.5, 154.7, 148.9,

143.3, 141.0, 128.9, 128.6, 126.8, 119.3, 113.7, 89.6, 86.0, 83.7,

83.5, 60.3, 50.1, 44.8, 33.8, 32.4, 29.2, 28.6, 27.3, 25.7, 25.5,

24.7, 14.7. HPLC (basic method) retention time 2.07′. LCMS

m/z: 573.4 (Parent MH⁺).

6-[(2,2-Diphenylethyl)amino]-9-(N-ethyl-2,3-O-isopropyli-

dene-β-^D-ribofuranosyluronamide)-N-(2-{N′-[1-(2-pyridyl)-4-

piperidyl]ureido}ethyl)-9H-purine-2-carboxamide Acetoni-

trile Monosolvate Monohydrate (34c). Methanol disolvate

34b (40.7 kg, 46.1 mol) was dissolved in acetonitrile (204 L) at

reflux and water (2 L, 99.5 mol). The mixture was cooled to 2

°C over 2 h to give a yellow slurry which was stirred for 48 h

and then filtered; the cake was washed with chilled acetonitrile

(41 L) and pulled dry. The solid was dried in vacuo overnight

at 50 °C to yield acetonitrile monosolvate monohydrate 34c as

a white solid (38.8 kg, 44.2 mol, 96%).

6-[(2,2-Diphenylethyl)amino]-9-(N-ethyl-2,3-O-isopropyli-

dene-β-^D-ribofuranosyluronamide)-N-(2-{N′-[1-(2-pyridyl)-4-

piperidyl]ureido}ethyl)-9H-purine-2-carboxamide Methanol

Disolvate (34b). Safety testing data. DSC testing of the

starting material salt 11b showed endothermic melt from 188

°C and a small broad exotherm from 306 °C. Testing of salt 12

showed an endothermic melt from 127 °C and two small

exotherms from 176 °C. Product solvate 34b showed an

endothermic melt from 119 °C and a small exotherm past 286

°C. Reaction calorimetry showed no accumulation at any stage;

however, peak gas evolution of 3.4 L/min/mol was seen during

CDI addition, and a further peak flow of 1.8 L/min/mol was

seen on addition of salt 12. A dose-controlled exotherm of −42

kJ/mol was seen upon addition of TEA. TSU screening of

reaction after addition of CDI showed a small exotherm from

140 °C. TSU testing post TEA addition showed no thermal

events up to 200 °C.

Recrystallization of 34c. Safety Testing Data. DSC testing

of the starting material salt 34c showed endothermic melt from

120 °C and a small broad exotherm from 286 °C. Reaction

calorimetry showed an exotherm of −67.5 kJ/mol.

Acetonitrile monosolvate monhydrate 34c (38.8 kg, 44.2

mol) was recrystallized from acetonitrile (194 L) in the absence

of water to give an inferior yield (25.45 kg, 29.0 mol, 65%). The

filtrate was distilled to a total volume of approximately 65 L,

water (0.6 L, 33.2 mol) was added, and the mixture was cooled

to 70 °C and seeded with 34c. The mixture was cooled to 2 °C

to give a yellow slurry that was stirred for 14 h and filtered, and

the cake was washed with chilled acetonitrile (14 L) and pulled

dry. The solid was dried in vacuo overnight at 50 °C to yield

acetonitrile monosolvate monohydrate 34c as a white solid

(11.25 kg, 12.83 mol, 29%). Total yield of 36.7 kg, 94% over

1

the two crops. H NMR (400 MHz, DMSO-d₆, 70 °C): δ 8.35

(1H, br s), 8.29 (1H, br s), 8.04 (1H, ddd, J = 4.9, 2.1, 1.1 Hz),

7.70 (1H, br s), 7.43 (2H, m), 7.33 (4H, m), 7.23 (4H, m),

7.13 (2H, m), 6.71 (1H, d, J = 8.5 Hz), 6.53 (1H, dd, J = 6.1,

5.3), 6.36 (1H, br s), 5.87 (1H, br m), 5.78 (1H, d, J = 7.5 Hz),

5.53 (1H, dd, J = 6.1, 1.9 Hz), 5.35 (1H, d, J = 6.1 Hz), 4.58

(1H, m), 4.55 (1H, d, J = 2.3 Hz), 4.25 (2H, br s), 4.01 (2H,

m), 3.60 (1H, m), 3.34 (2H, m), 3.27 (2H, m), 2.88 (2H, m),

2.73 (2H, m), 2.01 (MeCN, s), 1.75 (2H, m), 1.50 (3H, s),

1.32 (3H, s), 1.26 (2H, m), 0.51 (3H, t, J = 7.1 Hz). ¹³C NMR

(100 MHz, DMSO-d₆, 30 °C): δ 169.5, 164.1, 159.3, 158.3,

154.6, 153.4, 148.5, 148.1, 143.4, 143.3, 142.6, 138.0, 128.8,

128.7, 126.8, 120.2, 118.5 (MeCN), 113.3, 112.9, 107.5, 90.2,

87.2, 84.2, 83.8, 50.4, 47.2, 44.9, 44.1, 41.0, 33.7, 32.2, 27.1,

25.6, 14.2, 1.6 (MeCN). HPLC (basic method) retention time

3.02′. LCMS m/z: 818.5 (nonsolvated MH⁺).

Dicyclohexylmethylamine salt 11b (43.9 kg, 57.2 mol, 1.0

equiv) was dissolved in DCM (176 L) at 20 °C, and

hydrochloric acid (1M, 132 L) was added. The layers were

separated, and the aqueous phase was extracted with DCM (44

L). The initial organic phase was washed with water (132 L),

and the two DCM extracts of acid 20 were combined and

azeotropically dried until below 0.1% w/w water by Karl

Fischer analysis to give a dry DCM solution of free acid 11.

1,1′-Carbonyldiimidazole (11.13 kg, 68.6 mol, 1.2 equiv) was

added portionwise to the free acid 11 solution in DCM (∼220

L). The reaction was stirred for 2 h at 20 °C, and then amine

hydrochloride salt 12 (23.2 kg, 68.6 mol, 1.2 equiv) was added

as a solid followed by triethylamine (6.9 kg, 68.6 mol, 1.2

equiv). The reaction mixture was stirred overnight upon which

the DCM was distilled and replaced with methanol at ambient

pressure. The solution was cooled to 20 °C and seeded with

product amide 34b, stirred for 8 h at 20 °C, and then cooled to

0 °C and held overnight. The mixture was filtered, washed with

chilled methanol (33 L), and dried at 50 °C in vacuo to give

amide methanol disolvate 34b as a white powder (40.7 kg, 46.2

6-[(2,2-Diphenylethyl)amino]-9-(N-ethyl-β-D-ribofuranosy-

luronamide)-N-(2-{N′-[1-(2-pyridyl)-4-piperidyl]ureido}ethyl)-

9H-purine-2-carboxamide (1). Methanesulphonic acid (4.72

kg, 49.1 mol) was added to a stirred slurry of acetonide 34c

(36.5 kg, 41.6 mol) in water (183 L), and the solution was

heated at 70 °C for 10 h. Upon reaction completion, 1.5%

residual starting material as monitored by acidic HPLC method,

the reaction mixture was cooled to 20 °C and diluted with

methyl acetate (355 L). The reaction mixture was washed with

5% w/w aqueous disodium hydrogen phosphate solution. The

phases were separated, and the organic phase was washed once

more with 0.5% w/w aqueous disodium hydrogen phosphate

solution. The resultant water-wet methyl acetate solution of 1

was transferred to a speck-free vessel via a 1 μm suprapore

polypropylene inline filter and was then azeotropically dried by

distillation and replacement of the solvent (815 L in total) with

portions of fresh methyl acetate until the residual water level

was 2.2% w/w by Karl Fisher analysis. The crystallization

solution was seeded and heated at reflux for a further 24 h

whereupon crystallization occurred. The slurry was cooled to

22 °C and filtered, and the cake was washed with 2.4% water-

1

mol, 81%). H NMR (400 MHz, DMSO-d₆, 70 °C): δ 8.35

(1H, br s), 8.29 (1H, br s), 8.04 (1H, dd, J = 4.9, 1.9 Hz), 7.70

(1H, br s), 7.43 (2H, m), 7.33 (4H, m), 7.23 (4H, m), 7.13

(2H, m), 6.71 (1H, d, J = 8.6 Hz), 6.53 (1H, dd, J = 6.9, 4.9),

6.36 (1H, br s), 5.87 (1H, br m), 5.78 (1H, d, J = 7.9 Hz), 5.53

(1H, dd, J = 6.1, 2.1 Hz), 5.35 (1H, d, J = 5.8 Hz), 4.58 (1H,

m), 4.55 (1H, d, J = 2.0 Hz), 4.24 2H, (br s), 4.01 (2H, m),

3.59 (1H, m), 3.34 (2H, m), 3.27 (2H, m), 2.88 (2H, m), 2.73

(2H, m), 1.75 (2H, m), 1.50 (3H, s), 1.32 (3H, s), 1.27 (2H,

m), 0.51 (3H, t, J = 7.0 Hz). ¹³C NMR (100 MHz, DMSO-d₆,

30 °C): δ 169.4, 164.1, 159.3, 158.3, 154.6, 153.3, 148.5, 148.1,

143.4, 143.3, 142.6, 138.0, 128.8, 128.7, 126.8, 120.2, 113.3,

112.9, 107.5, 90.1, 87.2, 84.2, 83.8, 50.4, 47.2, 44.9, 44.1, 41.0,

33.6, 32.2, 27.1, 25.6, 14.2. HPLC (basic method) retention

time 3.02′. LCMS m/z: 818.5 (nonsolvated MH⁺).

482

dx.doi.org/10.1021/op200365n | Org. Process Res. Dev. 2012, 16, 470−483

Organic Process Research & Development

Article

(c) Akkari, R.; Burbiel, J. C.; Hockemeyer, J.; Muller, C. E. Curr. Top.

Med. Chem. 2006, 6, 1375.

wet methyl acetate (35 L) and pulled dry. The solid was

transferred to a vacuum oven and dried in vacuo overnight at

50 °C to yield 1 as a white solid (25.3 kg, 32.5 mol, 78%), mp

(7) (a) Vorbruggen, H; Ruh-Pohlenz, C. Org. React. 2000, 55, 1.

̈

1

(b) Vorbruggen, H. Acta Biochim. Pol. 1996, 43 (1), 25.

̈

190−192 °C; H NMR (400 MHz, DMSO-d₆, 70 °C): δ 8.58

(c) Vorbruggen, H.; Hofle, G. Chem. Ber. 1981, 114, 1256.

̈

(1H, br s), 8.43 (1H, br s), 8.10 (1H, br t, J = 5.9 Hz), 8.04

(1H, dd, J = 5.1, 2.0 Hz), 7.73 (1H, br s), 7.43 (1H, ddd, J =

8.8, 6.9, 2.0 Hz), 7.37 (4H, d, J = 7.5 Hz), 7.24 (4H, t, J = 7.3

Hz), 7.12 (2H, t, J = 7.3 Hz), 6.71 (1H, d, J = 8.5 Hz), 6.53

(1H, dd, J = 7.1, 5.1 Hz), 6.00 (1H, d, J = 6.8 Hz), 5.85 (1H, br

m), 5.81 (1H, d, J = 7.5 Hz), 5.41 (1H, d, J = 4.6 hz), 5.34 (1H,

d, J = 5.9 Hz), 4.64 (1H, m), 4.57 (1H, t, J = 7.4 Hz), 4.29 (2H,

m), 4.21 (1H, m), 4.01 (2H, br dt, J = 13.4, 3.4 Hz), 3.60 (1H,

m), 3.33 (2H, br m), 3.24 (2H, br m), 3.19 (2H, m), 2.86 (2H,

br t, J = 13.6 Hz), 1.74 (2H, m), 1.24 (2H, br ddq, J = 10.8, 3.4

Hz), 0.97 (3H, t, J = 7.1 Hz). ¹³C NMR (100 MHz, DMSO-d₆,

30 °C): δ 169.7, 163.9, 159.3, 158.5, 154.3, 153.1, 149.5, 148.1,

143.5, 142.1, 138.0, 128.8, 128.8, 126.8, 120.6, 112.9, 107.5,

87.9, 85.0, 73.5, 73.4, 50.8, 47.2, 44.8, 44.1, 41.5, 39.5, 34.0,

32.2, 15.2. HPLC (acidic method) retention time 2.36′. High

res MS, ES m/z: 778.3776 obs (calc 778.3783) (MH⁺).

(8) (a) Mantell, S. J.; Stephenson, P. T.; Monaghan, S. M.; Maw, G.

N.; Trevethick, M. A.; Yeadon, M.; Walker, D. K.; Selby, M. D.;

Batchelor, D. V.; Rozze, S.; Chavaroche, H.; Lemaitre, A.; Wright, K.

N.; Whitlock, L.; Start, E. F.; Wright, P. A.; Macintyre, F. Bioorg. Med.

Chem. Lett. 2009, 19, 4471. (b) Challenger, S.; Dessi, Y.; Fox, D. E.;

Hesmondhalgh, L. C.; Pascal, P.; Pettman, A.; Smith, J. D. Org. Process

Res. Dev. 2008, 12, 575.

(9) Nisato, D. Carminati, P. EP 0021973, 1981.

(10) (a) Singh, A. K.; Chaurasia, N.; Rahmani, S.; Srivastava, J.;

Singh, B. Catal. Lett. 2004, 95, 135. (b) Walton, E.; Rodin, J. R.;

Stammer, C. H.; Holly, F. W.; Folkers., K. J. Am. Chem. Soc. 1958, 80

(19), 5168.

(11) Golubev, D.; Rozantsev, E. G.; Neiman, B. M. Izv. Akad. Nauk

SSSR, Ser. Khim. 1965, 1927.

(12) de Nooy, A. E. J.; De; Besemer, A. C.; van Bekkum, H. Synthesis

1996, 1153.

(13) (a) Anelli, P. L.; Biffi, C.; Montanari, F.; Quici, S. J. Org. Chem.

1987, 52 (12), 2559. (b) Zhao, M.; Li, J.; Mano, E.; Song, Z.; Tschaen,

D. M.; Grabowski, E. J. J.; Reider, P. J. J. Org. Chem. 1999, 64 (7),

2564. (c) Song, Z. J.; Zhao, M.; Desmond, R.; Devine, P.; Tschaen, D.

M.; Tillyer, R.; Frey, L.; Heid, R.; Xu, F.; Foster, B.; Li, J.; Reamer, R.;

Volante, R.; Grabowski, E. J. J.; Dolling, U. H.; Reider, P. J.; Okada, S.;

Kato, Y.; Mano, E. J. Org. Chem. 1999, 64 (26), 9658. (d) Davis, N. J.;

Flitsch, S. L. Tetrahedron Lett. 1993, 34, 1181.

AUTHOR INFORMATION

Corresponding Author

*david.entwistle@me.com; julian.smith@pfizer.com

■

Notes

The authors declare no competing financial interest.

(14) Ciriminna, R.; Pagliaro, M. Org. Process Res. Dev. 2010, 14, 245.

(15) Ahman, J.; Birch, M.; Haycock-Lewandowski, S. J.; Long, J.;

Wilder, A. Org. Process Res. Dev. 2008, 12, 1104.

(16) Aqueous solubilities. TEAB (48.35 g/100 g of water), TBAB

(12.35 g per 100 g water). Gopal, R.; Chandra, D. J. Ind. Chem. Soc.

1970, 47 (2), 156.

(17) (a) Teasdale, A.; Eyley, S. C.; Delaney, E.; Jacq, K.; Taylor-

Worth, K.; Lipczynski, A.; Reif, V.; Elder, D. P.; Facchine, K. L.; Golec,

S.; Oestrich, R. S.; Sandra, P.; David, F. Org. Process Res. Dev. 2009, 13,

429 and the references therein. (b) Elder, D. P; Snodin, D. J. J. Pharm.

Pharmacol. 2010, 61 (3), 269.

ACKNOWLEDGMENTS

■

We thank Sam Farenden and Mahlet Negussie for analytical

support, John Williams, David Clifford, David Martyn, Fabrice

Salingue, Paul Glynn and the operators for crucial support in

the Pfizer scale-up facilities. We also thank Darrell Fox (Pfizer)

and Jim Ramsden (ChiroTech) for their contributions to the

development of the ethoxycarbonylation approach to 16, Geoff

Fuller Chemicals for the supply of 16, 25 and 21, Peakdale

Technologies and BASF for the supply of 12, and Degussa and

SEAC for the supply of 29.

(18) McConville, F. X. The Pilot Plant Real Book; FXM Engineering

and Design: Worcester, MA, 2002; Chapter 6.

REFERENCES

■

(1) (a) National Heart, Lung and Blood Institute, What is COPD?

National Institutes of Health: Bethesda, MD, http://www.nhlbi.nih.

gov/health/public/lung/copd/what-is-copd/index.htm. (b) National

Heart, Lung, and Blood Institute, Morbidity and Mortality: 2007 Chart

Book on Cardiovascular, Lung, and Blood Diseases, National Institutes of

Health: Bethesda, MD, 2007. Available at http://www.nhlbi.nih.gov/

resources/docs/07a-chtbk.pdf.

(2) (a) Mathers, C. D.; Loncar, D. PLoS, Med. 2006, 3 (11), 442. (b)

Global Initiative for Chronic Obstructive Lung Disease (GOLD).

Global Strategy for Diagnosis, Management, and Prevention of COPD,

2006, http://www.goldcopd.com.

(3) (a) Barnes, P. J. Br. Med. J. 2006, 333, 246. (b) Rabe, K. F.; Hurd,

S.; Anzueto, A.; Barnes, P. J.; Buist, S. A.; Calverley, P.; Fukuchi, Y.;

Jenkins, C.; Rodriguez-Roisin, R.; van Weel, C.; Zielinski, J. Am. J.

Respir. Crit. Care Med. 2007, 176 (6), 532. (c) Hogg, J. C.; Chu, F.;

Utokaparch, S.; Woods, R.; Elliott, M. W.; Buzatu, L.; Cherniack, R.

́

M.; Rogers, R. M.; Sciurba, F. C.; Coxson, H. O.; Pare, P. D. N. Engl. J.

Med. 2004, 350 (26), 2645. (d) National Heart, Lung, and Blood

Institute. What is COPD? National Institutes of Health: Bethesda, MD,

http://www.nhlbi.nih.gov/health/health-topics/topics/copd/.

(4) Barnes, P. J.; Stockley, R. J. Eur. Respir. J. 2005, 25, 1084.

(5) (a) Mantell, S. J. Monaghan, S. M. WO 0077018, 2000;

(b) Smith, J. D. Silk, T. V. WO 048180, 2003; (c) Mantell, S. J.

Stephenson, P. T. WO 094368, 2001.

(6) (a) Sullivan, G. W. Curr. Opin. Invest. Drugs 2003, 4 (11), 1313.

(b) Polosa, R.; Holgate, S. T. Curr. Drug Targets 2006, 7, 699.

483

dx.doi.org/10.1021/op200365n | Org. Process Res. Dev. 2012, 16, 470−483

Article Doi

Route selection and process development of a multikilogram route to the inhaled A2a agonist UK-432,097

DOI: 10.1021/op200365n

Source and publish data:

Authors:

Article abstract of DOI:10.1021/op200365n

Full text of DOI:10.1021/op200365n

Products guided by the article

R&D Labs maybe for 1360054-66-5

Relevant to this article

Hot Product