Diastereospecific enolate addition and atom-efficient benzimidazole synthesis for the production of l/t calcium channel blocker act-280778

Article abstract of DOI:10.1021/op400269b

A scalable access to 1 (ACT-280778), a potent L/T calcium channel blocker, has been developed. The synthesis, amenable to kilogram manufacturing, comprises 10 chemical steps from enantiomerically pure 5-phenylbicyclo[2.2.2]oct-5-en-2-one (3) and 1,4-dimethoxybenzene with a longest linear sequence of 7 steps. Key to the success of this fit-for-purpose approach are a robust and atom-efficient access to benzimidazole 4, the substrate-controlled diastereoselective enolate addition toward carboxylic acid 2 that was isolated by simple crystallization with high dr (>99:1), the convenient selective N-deacylation of intermediate 10, and the identification of a suitable solid form of 1 as the bis-maleate salt (1·2 C₄H₄O₄). As an illustration of the robustness of this process, 14 kg of drug substance, suitable for human use, was produced with an overall yield of 38% over the longest linear sequence (7 steps).

Full text of DOI:10.1021/op400269b

Article
pubs.acs.org/OPRD
Diastereospecific Enolate Addition and Atom-Efficient Benzimidazole
Synthesis for the Production of L/T Calcium Channel Blocker ACT-
280778
Jacques-Alexis Funel,† Stephan Brodbeck,‡ Yves Guggisberg,∥,⊥ Remy Litjens,§ Thomas Seidel,∥
Martin Struijk,§ and Stefan Abele*,†
†Process Research Chemistry, Actelion Pharmaceuticals Ltd., Gewerbestrasse 16, CH-4123 Allschwil, Switzerland
‡Technical Operations, Drug Substance, Actelion Pharmaceuticals Ltd., Gewerbestrasse 16, CH−4123 Allschwil, Switzerland
§CML Europe BV, Vliesvenweg 1, P.O. Box 8, Weert, The Netherlands
∥Valsynthese SA, Fabrikstrasse 48, CH-3900 Brig, Switzerland
S
* Supporting Information
ABSTRACT: A scalable access to 1 (ACT-280778), a potent L/T calcium channel blocker, has been developed. The synthesis,
amenable to kilogram manufacturing, comprises 10 chemical steps from enantiomerically pure 5-phenylbicyclo[2.2.2]oct-5-en-2-
one (3) and 1,4-dimethoxybenzene with a longest linear sequence of 7 steps. Key to the success of this fit-for-purpose approach
are a robust and atom-efficient access to benzimidazole 4, the substrate-controlled diastereoselective enolate addition toward
carboxylic acid 2 that was isolated by simple crystallization with high dr (>99:1), the convenient selective N-deacylation of
intermediate 10, and the identification of a suitable solid form of 1 as the bis-maleate salt (1·2 C4H4O4). As an illustration of the
robustness of this process, 14 kg of drug substance, suitable for human use, was produced with an overall yield of 38% over the
longest linear sequence (7 steps).
solution of HCl in AcOEt followed by concentration to dryness
(90%).
INTRODUCTION
■
Chiral bicyclic benzimidazole 1 (ACT-280778) is a L/T
calcium channel blocker potentially indicated for the treatment
of hypertension and angina pectoris (Scheme 1). There was a
need at Actelion for a robust, safe and scalable production of
the active pharmaceutical ingredient (API) 1 to supply material
for preclinical and clinical studies.
At the outset of this project, a number of aspects had to be
considered to enable the large scale production of 1: (1)
development of a shorter and scalable way to produce
benzimidazole 4, (2) identification of a chromatography-free
access to carboxylic acid 2, (3) improvement of the efficiency of
the introduction of the isobutyroyl ester moiety in the last stage
toward free base API 1, (4) identification of a suitable solid
form of 1. A maximum of 5 g of API as HCl salt was prepared
in multiple batches with 14% overall yield from 3. This was
sufficient to profile 1 until its selection for further development;
however, the throughput was too low to consider the synthesis
of multikilogram amounts of this substance.
In this account,1 we report a fit-for-purpose approach
towards 1 that went through two optimization cycles in two
production campaigns, starting with the route implemented in
Discovery Chemistry (Schemes 2 and 3).2 Two practical
syntheses of 3 on multikilogram scale have already been
reported by our laboratories for both racemic3,4 and
enantiomerically pure5,6 forms. Discovery Chemistry designed
an 8-step sequence to free base API 1 from bicyclic ketone 3
starting with the addition of the lithium enolate of tert-
butylacetate onto rac-3 (Scheme 2). It afforded mixtures of
diastereoisomers (6 and 7) with an average ratio of 3.5−4:1
which were separated by chromatography. The enantiopure
adduct 6 was subsequently separated from its undesired
enantiomer via preparative HPLC in 31% overall yield. After
saponification of 6 (98% yield), the optically pure acid 2 was
coupled with the benzimidazole building block 4, prepared in
six steps from 1,4-dimethoxybenzene (Scheme 4), to yield
amide 8 (79%). Product 8 was reduced (76%) with sodium
bis(2-methoxyethoxy)aluminium (RedAl) followed by acylation
with an excess of isobutyroyl chloride to yield 10. Selective N-
deacylation of 10 was accomplished by treatment with silica gel
for a period of one week (81%). The amorphous API was
isolated as the bis-HCl salt (1·2HCl) after treatment with a
RESULTS AND DISCUSSION
■
Synthesis of Benzimidazole 4. The benzimidazole 4 was
first synthesized following the Discovery Chemistry route
(Scheme 4).2b This sequence did not meet our requirements
for scale-up due to the high costs of some raw materials and
reagents7 and the need for chromatography (phenylenediamine
12a) to obtain pure 4.
Still, the first two steps, i.e. the dinitration of readily available
1,4-dimethoxybenzene (5) and the subsequent reduction to the
phenylenediamine 12a as substrate for the benzimidazole
formation were attractive. Three major challenges remained to
be addressed for scale up: (1) the dinitrobenzene derivatives
Received: September 26, 2013
© XXXX American Chemical Society
A
dx.doi.org/10.1021/op400269b | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
Scheme 1. Retrosynthesis of 1
Scheme 2. Discovery Chemistry synthesis of 2 (part 1 of 2)
Scheme 3. Discovery Chemistry synthesis of 1 (part 2 of 2)
(11a, 11b) have a high decomposition energy (see Exper-
imental Section),8 (2) both phenylenediamine 12a and
benzimidazole 4 are highly water-soluble rendering an isolation
very difficult, and (3) the phenylenediamine 12a was obtained
as deeply colored (black) substance that decomposed to a tarry
black mass when exposed to air.
Several methods are known for the synthesis of benzimida-
zoles carrying a polar side chain at C2. The Phillips reaction,9
i.e. the condensation of o-phenylenediamines with carboxylic
acids in the presence of diluted mineral acids had been used
with various amino acid derivatives.10 We quickly focused on
the unprecedented Phillips reaction with N-methyl pyrrolidi-
none (NMP) as condensation reagent. This strategy has the
advantage of using a readily available solvent as reagent
containing all required atoms for the targeted benzimidazole.
Herein, we report the optimization of the process from the first
kilograms in campaign 1 (15 kg) to 141 kg of 4 produced in
campaign 2, describing how the issues encountered during the
development have been solved. The chosen synthetic scheme
for the production of 4 is short, comprising three synthetic
steps (Scheme 5) and uses readily available raw materials. The
benzimidazole 4 was built up in one step from the
phenylenediamine intermediate 12a by condensation with
NMP in aqueous HCl.
Nitration. The dinitration of 1,4-dimethoxybenzene 5 with
conc. HNO3 (70%) is known to give a mixture of
regioisomers.11 Efficient stirring was crucial. The reaction rate
correlated with the dissolution rate that increased with the
stirrer speed: after addition of a portion of 5, a brown spot12
was visible that disappeared within seconds, and the product 11
was formed and precipitated within minutes even at 0 °C.13 As
suggested by the literature precedence, the reaction was aged at
50 °C for 1 h. However, this had no beneficial effect on yield,
purity, or selectivity. The regioisomeric ratio (HPLC) was 4.8:1
and >7:1 at 20−25 and 0 °C, respectively. At −10 °C,
approximately 35% accumulation was observed, mainly due to
solubility issues of the starting material. Crystallization of the
crude product from alcohols, toluene, and acetic acid were
B
dx.doi.org/10.1021/op400269b | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
within several hours. Amongst two appropriate Pd and Pt
catalysts, we chose Pd as it led to a faster decay of the
hydroxylamine I as compared to Pt/V.
Scheme 4. Discovery Chemistry synthesis of 4 (Z =
benzyloxycarbonyl)
The sticky phenylenediamine 12 obtained after filtration over
Celite immediately turned black after contact with air.18 To our
surprise and delight, the undesired 2,5-regioisomer 12b was not
1
detected in the crude product (see H NMR spectra in the
Supporting Information [SI]). We speculate that this
regioisomer was forming−probably via para-chinoid intermedi-
ates−oligomeric side products or was giving oxidation products
that partially remained on the filter cake.19 The first option was
to process 12a as a solution in EtOH without isolation.20 To
this end, on 11-kg scale, a 7:1 regioisomeric mixture of 11a/
11b (11.5 kg, 20% loss on drying (LOD)) was hydrogenated
on 10% Pd/C (10% w/w) in EtOH at 20−30 °C and 1−2 bar
hydrogen within 10 h. Crude 12a was obtained as a solution in
ethanol in yields of 88−92% (95% a/a HPLC). Typical NMR
assays of aliquots of this solution after evaporation to dryness
were 88−90% w/w with no individual byproducts present,
indicating the presence of (polymeric) impurities not detected
by UV or by NMR. This solution was carried through to the
cyclization with NMP. However, the above-mentioned black
impurities severely hampered workup and necessitated several
charcoal treatments (vide infra). Therefore, the removal of the
tar in 12a was mandatory for the next scales. To this end,
different types of charcoal were tested but no decoloration was
obtained and the assay (weight of free base relative to the total
weight as determined by HPLC with a reference standard) only
slightly increased from 90% to maximum 94% w/w. Gratify-
ingly, the HCl salt synthesized by adding either aqueous (47%)
or gaseous HCl (66% yield) to the EtOH solution of crude 12
gave an off-white (sometimes slightly pink) HCl salt 12a·HCl
with an assay of 82% w/w which corresponds to 100%, based
on the mono-HCl salt.21
found to enhance the regioisomeric ratio.14 Regioisomerically
pure 11a thus obtained (11a:11b = 29:1) led to approximately
10% higher yield in the next step. However, the overall yield of
steps 1 and 2 was not higher due to the yield loss in the
crystallization. Therefore, 11 was processed as regioisomeric
mixtures of 6−8:1. The process was scaled up uneventfully
from 5- to 2 × 400-L scale: 1,4-dimethoxybenzene (5, 64 kg)
was added in portions to 70% HNO3 at 0−6 °C, followed by an
aging period at 20 °C for 1−2 h. The suspension was quenched
onto water,15 followed by filtration and washing of the filter
cake with water. The yield for both isomers 11 (99 kg) was 91−
94% with a regioisomeric ratio of approximately 7:1. In total,
235 kg of 11 were produced.
Hydrogenation. The 7:1 mixture of dinitro compound 11
was carried forward as such into the highly exothermic
hydrogenation.16 As the intermediate hydroxylamines are
notoriously highly energetic compounds that could be
explosive, they should not accumulate during the hydro-
genation.17 The main intermediates in the hydrogenation of 11
have been identified by LC−MS (Figure 1). With 1% catalyst
loading, the reaction stalled at the hydroxylamine stage whereas
a 10% load led to full conversion to the phenylenediamine
In campaign 2, a 7:1 regioisomeric mixture of dry 11a/11b
(194 kg) was hydrogenated on 10% Pd/C (5 wt %) in EtOH at
20−30 °C and 1−1.5 bar hydrogen within 12 h. After filtration
of the mixture, HCl (1.5 equiv) was introduced as gas to
precipitate 12a·HCl in a good yield of 70% with an excellent
assay (82% w/w, 100% a/a).22
Cyclization. Initially, the EtOH solution of free base 12a
was telescoped into the next step. Gratifyingly, the trans-
a
Scheme 5. Synthetic scheme used for (a) campaign 1 (15 kg 4·2HCl) and (b) campaign 2 (141 kg 4·2HCl)
a
See detailed process description in the text.
C
dx.doi.org/10.1021/op400269b | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
Figure 1. Monitoring of intermediates I, II/III, and of 11a and 12a during the hydrogenation of 11a by LC−MS. (a) 3% Pt/2.5% V (Degussa CF
1082 RV/W) (b) 10% Pd/C (Engelhard ESCAT 163). Reaction in EtOH, H2 (1 bar), 10% w/w catalyst loading. [LC−MS conditions: Eclipse Plus
C18, 1.8 μm, 2.1 mm × 50 mm, 1 mL/min, 50 °C, eluent A: water, 0.04% TFA, eluent B: acetonitrile, 0.006% TFA, gradient: 2 min 95% B, 2.8 min
95% B, 3.0 min 5% B.]
formation of the free base 12a to the benzimidazole proceeded
well with 2 equiv NMP and 20% HCl23 at 102−104 °C
(reflux); the reaction started as soon as EtOH was removed by
distillation, and the reaction took 1−2 days for completion. The
major issues were the workup and the isolation: benzimidazole
4 is water-soluble, and it is insoluble in 2-methyl-THF, EtOAc,
iPrOAc, MIBK, TBME, or toluene at pH 9. It is sparingly
soluble in dichloromethane and acetone. However, an accept-
able solubility was measured in n-butanol (4 vol at rt) that
allowed for an aqueous workup at both pH 13−14 and at pH 1.
n-Butanol was chosen as solvent for both extractions and for
the salt formation: the HCl salt 4·2 HCl was formed by
addition of the n-butanol solution to HCl in AcOEt at 10 °C.
Another challenge consisted of poor phase separation of the
two black layers that were facilitated on scale by a conductivity
meter. The crude free base obtained in first batches contained
consisting of a third charcoal treatment of 4·2 HCl in refluxing
EtOH, affording 4·2 HCl as off-white solid in 67% yield (15
kg).
The major change in campaign 2 was the use of the stable
HCl salt of 12a as starting material instead of the
corresponding free base. Two out of three charcoal treatments
(being associated with a lot of waste) could be skipped. Careful
monitoring of the assay in weight % (in contrast to the area %
purity by HPLC) in all unit operations was crucial. In order to
suppress oxidative decomposition, all solvents and reagents
were degassed prior to use, and sodium sulfite30 was added to
the cyclization mixture. The batches of campaign 1 contained
1−2% of salts according to sulphated ash. An extractive wash of
the n-butanol layer after basification with 30% NaOH was
investigated. Whereas a brine wash did not decrease the
sulphated ash, a water wash reduced it to 0.5% with an
associated slight loss of yield due to the water solubility of 4·2
HCl. The water content turned out to be crucial for the
crystallization. A water content of <0.5% led to very small
crystals which were difficult to filter, whereas a water content of
3−8% gave larger crystals and hence faster filtration but with a
yield reduced by 8%. A solubility study of 4·2 HCl showed a
minimal solubility at approximately 2% water in n-butanol. As a
compromise between yield and quality, a water content of 3%
was chosen as specification. In practice, 32% HCl and charcoal
were added to the n-butanol phase after the basic extraction,
and this suspension was filtered over Celite. The filtrate was
concentrated to both azeotropically remove water and to trigger
crystallization. At approximately 3.3 vol, the water content was
adjusted to 3% based on the analysis of an aliquot. In campaign
2, seven batches of 4·2 HCl (141 kg) were produced from the
HCl salt 12a·2 HCl with high assay (72% w/w) and purity
(100% a/a HPLC). The water content was 5.3% (correspond-
1
10−30% of undetectable (by HPLC, H NMR) impurities
which were neither NaCl nor solvents, as indicated by
sulphated ash, GC-headspace, chloride-titration and ICP-OES
(Na). We assume that these impurities are tarry byproducts
derived from oxidation of the phenylenediamines or
demethylation products.25 Stress tests with benzimidazole 4
showed that it was not stable in 32 and 20% HCl in a closed
vial at 95 °C for 16 h, leading to 44 and 15% demethylated
product according to LC−MS.26 On the basis of these results,
nonacidic benzimidazole syntheses should be preferred.
However, no viable alternative was found.27 To remove these
tarry byproducts, two charcoal treatments28 were included in
the workup of campaign 1, and the bis HCl monohydrate salt of
4 (four 6-kg batches) was formed in n-butanol/AcOEt as pink
crystals in 77% yield with 97−99% a/a purity. Although these
batches met the current specifications, a user test with 4·2 HCl
in the amide bond formation failed.29 This required a rework
D
dx.doi.org/10.1021/op400269b | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
Scheme 6. Alternative synthesis of 8 by addition of the enolate of acetamide 17
Figure 2. Impurities derived from CDMT-mediated coupling of 2 and 4.
ing to the monohydrate), the chloride content was 21−22%
(corresponding to the dihydrochloride), and the average yield
was 59% (39% overall from 5). In order to assess the impurities
that have not been covered by the standard analytical methods,
the color of a solution of 4·2 HCl (2% w/w solution in water/
acetonitrile 1:1) was compared with color reference solutions
according to the European Pharmacopoeia 5.0.31 This allowed
for a qualitative determination of the colored impurities.
Diastereospecific Synthesis of Carboxylic Acid 2. With
benzimidazole 4 now at hand, another line of attack consisted
of reinvestigating the addition of the lithium enolate of tert-
butylacetate onto optically pure phenyl ketone 3 obtained by
resolution of rac-3 with HPLC (Scheme 2). This reaction was
originally performed in THF at −60 °C to yield a mixture of
diastereoisomer (6 and 7) with a dr of 2.2 to 3.3:1. By lowering
the temperature to −80 °C, together with the use of toluene as
solvent instead of a mixture of THF and toluene, an improved
dr of 4.5 to 5:1 was obtained. This endo selectivity is only driven
by the difference in steric bulk of the CH2−CH2 bridge and the
CHCPh bridge.32 The presence of the tert-butyl ester moiety
as chelating group might stabilize in a solvent-dependent
manner the 6-membered Li complex of the endo intermediate
leading to endo adduct 6. Secondary orbital overlaps may
account for this selectivity as well.
be the optimal solvent to extract the desired mixture of acids
with a minimal loss of product. After a solvent switch to
AcOEt35 followed by slow cooling, the desired acid 2
crystallized and was isolated with a yield of 55−60% over
two steps (dr > 99:1, er > 98:2) starting from bicyclic ketone 3
with 98:2 er. A similar high level of enantiomeric purity (er >
98:2) was reached when the same sequence was conducted
with 3 but with only 95:5 er. The absolute configuration of 3
was unambiguously proven by X-ray crystal structure analysis of
two distinct derivatives [SI].
In order to satisfy the need for development, 90 kg of
optically pure ketone 3 were manufactured. Diels−Alder
cycloaddition on large scale36 and highly productive resolution
with preparative HPLC37 are two key features of the bulk
synthesis of 3. The latter was obtained in several batches of
high optical purity (er > 98:2) but with a maximal chemical
purity of 80% w/w (NMR assay). Fortunately, this variability
had no impact on the quality of the isolated acid 2. It only
affected the throughput at this stage. This protocol was
successfully scaled up. A freshly prepared solution of LDA was
added to a solution of ketone 3 in toluene at −80 °C. tert-Butyl
acetate was added dropwise to afford a mixture of
diastereoisomers that was subsequently saponified. After
extraction with 2-methyl-THF and solvent exchange, the
desired diastereoisomer was isolated by crystallization. On 8-
kg scale, the dr of the acid 2 (isolated in 60% yield with 98%
purity HPLC) slightly dropped to 98.3:1.7. The ratio was
improved to >99.9:0.1 by recrystallization from AcOEt (90%
recovery, 54% overall from 3). Later, it turned out that 2 was
the only solid intermediate isolated by crystallization until 1.
Indeed, all downstream intermediates (notably 8, 9, and 1 as
free base) were foamy materials or viscous oils. The successful
For safety and environmental reasons, BuLi was substituted
by HexLi.33 Other bases such as LiHMDS or NaHMDS were
tested, but none of them led to an improvement in terms of
conversion or selectivity.34 The crude mixture of diaster-
eoisomers was not purified but rather directly used in the next
saponification step after aqueous quench with citric acid. EtOH
was added followed by aqueous KOH to trigger the hydrolysis
of the mixture of tert-butyl esters. 2-Methyl-THF was found to
E
dx.doi.org/10.1021/op400269b | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
Figure 3. Impurities 23, 24a,b.
Scheme 7. Synthesis of 9: (a) campaign 1 (4.9 kg 9) and (b) campaign 2 (10.8 kg 9·2C2H2O4)
isolation of pure bicyclic carboxylic acid 2 was, hence, crucial
for the development of a practical route to 1.
next step to yield amine 9. A side reaction of CDMT with the
benzimidazole side chain 4 giving 18 (Figure 2) required a
slight overcharge of reagent. Under the reaction conditions,39
acylation of the nitrogen atom of the benzimidazole ring was
observed to yield impurity 19. Three side products (endo-
methylene 20, exo-methylene, cis- and trans-21a,b, Figure 2)
derived from dehydration of 8 were identified by LC−MS.
Several acidic and basic washings did not fully remove (below
0.2% a/a) the series of dehydrated products (20, 21a,b) that
were carried through to the final product with minimal purging
options in the absence of solid intermediates.40
The requirement to tightly control impurities in order to
reproducibly deliver high-purity material triggered additional
investigations in the preparation of the second kg-campaign.
The use of EDCI/HOBT led to less dehydrated amide
impurities (0.1% a/a vs 1% a/a with CDMT). The fact that
incomplete conversion was observed during the first familiar-
ization experiments using HOBT/EDCI was retrospectively
assigned to the low assay of early batches of benzimidazole side
chain 4. With upgraded quality of this material, full conversion
was observed so that CDMT could eventually be replaced. On
the occasion of a 5.5-kg batch using EDCI, only traces of
dehydrated impurities (less than 0.05% a/a) were formed. The
use of EDCI, however, led to the formation of up to 3% a/a of
the urea obtained by addition of the tertiary hydroxyl group
onto the carbodiimide. This side product was partially removed
by acidic washings (down to 1% a/a) and subsequently
removed as alcohol 22 in the next reduction step. The desired
In an attempt to further improve the diastereoselectivity of
this transformation, based on related precedents at Hoffmann-
La Roche,38 we envisioned that the use of functionalized
acetamide 17 may increase the convergency of the sequence.
Acetamide 17 was prepared from benzimidazole 4 using
isopropenyl acetate [SI]. Initial experiments to add 17 to 3
showed that the corresponding adducts were obtained in a 7:1
ratio in favor of the desired amide endo-8 (Scheme 6). By using
LiCl as additive, this ratio increased to 9:1. Nevertheless, this
approach was set aside since the undesired diastereoisomer
could not be rejected below an acceptable limit without a
significant loss of yield after formation of the corresponding
oxalate salt of 9 in the next step.
Coupling of 2 and 4. With such well-defined solid building
blocks at hand, the next step toward amide 8 was investigated
(Scheme 3). The amide coupling of 2 and 4 was originally
performed using HOBT/1-ethyl-3-(3-dimethylaminopropyl)-
carbodiimide·HCl (EDCI·HCl). In some reactions, incomplete
conversion (90−95%) was observed. A rapid and complete
conversion to the expected amide 8 was observed with 2-
chloro-4,6-dimethoxy-1,3,5-triazine (CDMT) and propane
phosphonic acid anhydride (T3P), whereas degradation was
observed with isobutylchloroformate, pivaloyl chloride, and
oxalyl chloride. When T3P was used, the bad smell arising after
LAH reduction in the next stage was a concern. CDMT was
hence selected for the first kg-campaign that afforded the amide
8 as a foam that was not isolated and engaged as such in the
F
dx.doi.org/10.1021/op400269b | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
Scheme 8. Overall synthesis of 1·bis maleate salt (17% overall yield) from cyclohexenone
amide 8 (10.2 kg) was obtained with a purity of 97.4% a/a
(HPLC) and was engaged as such in the next step.
step with the isolation of the oxalate salt of 9 helped to further
increase the robustness of this process.
Reduction of Amide 8. Amide 8 was initially reduced by
addition of a 65% solution of RedAl in toluene (Scheme 3).
Even though the conversion was satisfactory, the reaction
mixture in toluene was heterogeneous and difficult to stir, and
RedAl was replaced by LAH in THF. During the workup of the
preceding amide coupling, a final wash with diluted acid was
implemented after treatment with aqueous base. This led to
residual HCl in 8 which was isolated in 76% yield on 7-kg scale
(92.6% a/a HPLC). In this case, 2.3 equiv of LAH was
necessary to observe complete conversion to amine 9. Under
these conditions, the reaction mixture typically contained 3−
6% a/a (HPLC) of the mixture of dehydrated products (23,
24a, 24b, Figure 3). By adding the amide 8 to the LAH in THF,
complete reduction of the amide (used as free base) was
observed with only 2 equiv of LAH, as demonstrated during the
second production campaign. After hydrolysis of the aluminum
salts using aqueous KOH,41 the crude mixture was filtered over
Celite and stored as a solution in THF or 2-methyl-THF.
Despite extensive investigations, the unwanted dehydration
reaction could not be suppressed below 0.5% a/a. To bypass
this problem, the formation of the oxalate salt was envisioned.
Upon treatment with oxalic acid of the crude solution of 9 in
iPrOH, the oxalate salt crystallized out with an improved purity
(from 98.3% a/a with 0.70% a/a of 23, 24a,b to 99.5% a/a with
0.48% a/a of 23, 24a,b). The filtration and drying of the very
fine suspension in iPrOH, however, proved troublesome and
took 3 days to be completed on a 9-kg scale.42
Scalable Preparation of 1 (As Free Base). In order to
complete the synthesis of 1, the tertiary alcohol had to be
esterified to the isobutyroyl ester. To this end, the acylation of
9 was conducted in the presence of excess of isobutyroyl
chloride (2.5 equiv) to observe complete consumption of
starting material. However, the formation of 1 was accom-
panied by the simultaneous generation of compound 10
derived from overacylation on the nitrogen of the benzimida-
zole 4 (Scheme 3). Trials to tune the regioselectivity of this
transformation using other solvents, bases or reagents
(isobutyroyl anhydride) proved to be unsuccessful. In an
attempt to purify 10 over silica gel, partial cleavage to 1
occurred. This observation led to a preparative isolation of the
API by Discovery Chemistry. Nevertheless, this process could
not be reliably reproduced.43
We hypothesized that selective hydrolysis of the N-
isobutyroyl moiety over the tertiary ester function may be
possible in the presence of a mild nucleophilic base. In a first
trial, we were pleased to find that treatment of the crude
reaction mixture with NaOMe in MeOH led to a rapid, mild,
and selective cleavage of the N-acyl group, leaving the tertiary
ester unaffected. Only under stress conditions with excess
NaOMe under reflux, transesterification (up to 30%) was
observed. This pivotal finding secured a practical access to
larger quantities of 1. Nevertheless, the identification of a
suitable salt form of 1 remained a major challenge.
Preparation of the bis-maleate salt of 1. Before the
isolation of the oxalate salt of amino alcohol 9, all intermediates
derived from 2 and the API itself as free base were isolated as
oils or foams. Under these circumstances, the need for a well-
defined solid form became pressing. When first treated with
Using HOBT/EDCI instead of CDMT on the one hand and
inverting the order of addition of LAH on the other hand
allowed significant improvement of the overall purity of amine
9 (Scheme 7). The introduction of an additional purification
G
dx.doi.org/10.1021/op400269b | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
anhydrous HCl, the API was isolated as an amorphous material.
Attempts to crystallize the free base were all unsuccessful; thus,
a salt screening using a concise set of pharmaceutically
acceptable acids in various solvent systems was performed.
The key discriminating factors were good filterability and ease
of drying. The bis-maleate salt (Scheme 8) turned out to be a
breakthrough.44 Indeed, a subsequent in-depth screening
confirmed that this salt was the only form suitable for a clinical
formulation. It showed the largest stability range among other
candidates and exhibited good physicochemical properties.
Most significantly, this salt was readily isolated by filtration after
slow crystallization in contrast to other salts that formed thick
and fine suspensions causing lengthy filtrations and drying. The
same polymorphic form was consistently obtained. These
favorable properties supported the selection of the bis-maleate
salt for long-term development.
The crude API as a solution in AcOEt was therefore treated
with a solution of maleic acid in MeOH under reflux. Upon
cooling, the corresponding bis-maleate salt (1·2 C4H4O4) was
isolated with good yield (74% over two steps) and excellent
chemical purity (99.7% a/a), suitable for human use. The
identification of this salt turned out to be the cornerstone of
this through process since it allowed for an efficient and robust
purification that delivered material within specifications after
three telescoped steps. Overall, this synthetic sequence
delivered 14 kg of 1 with a yield of 38% (seven linear steps)
from chiral bicyclic ketone 3, and 17% yield (14 linear steps)
from cyclohexenone (Scheme 8).
the HPLC/MS and DSC methods are listed in the Supporting
Information. All temperatures are internal temperatures and
yields are presented as is, unless otherwise stated.
1,4-Dimethoxy-2,3-dinitrobenzene (5). (Caution: highly
energetic substance!8 5 was not shock sensitive (falling hammer test
according to Lutolf). However, a detonation was observed in the
̈
falling hammer test once the powder was wrapped in aluminium
foil). In a 600-L inox reactor, 1,4-dimethoxybenzene (5, 32 kg,
white crystalline chips) was charged in 5 portions to 70%
HNO3 (482 kg, 10 vol, 23 equiv) at 0−6 °C within 50 min with
good stirring (475 rpm, Caution: exothermic addition!). After
stirring at 0−5 °C for 30 min, the mixture was warmed to 20
°C and stirred at 20 °C for 1 h. IPC (TLC) indicated full
conversion. A second nitration batch (32 kg 5) was produced in
a second 600-L inox reactor. The suspensions of both 600-L
reactors were added to water (1500 L) in a 4300-L SS reactor
at 9−18 °C within 35 min. The suspension was filtered, washed
with water (3 × 500 L, pH of filtrates: 1.4, 3.5, and 4.0), and
dried in a tray dryer at 70 °C, 100 mbar for approximately 40 h
to afford 11 as a yellow crystalline solid. Yield: 99.2 kg (94%).
Purity (HPLC method 2): 98.9% a/a sum of 11a and 11b, ratio
1
11a:11b = 7.6:1, 11a tR = 11.2 min, 11b tR = 10.8 min; H
NMR and 13C NMR data in accordance with literature data.11a
3,6-Dimethoxybenzene-1,2-diamine hydrochloride
(12a·HCl). Pd/C (9.8 kg, 10% Pd/C, type E 101 R/W, 5%
w/w) and 11 (195.5 kg, regioisomeric mixture 7.6:1, 11a:11b)
were charged into an inerted 4300-L glass-lined reactor,
followed by EtOH (1080 kg). The vessel was inerted. After
purging three times with hydrogen, the stirrer was started, and
the mixture was hydrogenated at 20−30 °C for 12 h at 1.5 bar.
(Caution: exothermic dinitro hydrogenation, cooling with brine!)
IPC (HPLC) showed approximately 3% intermediates left.
After inertization, the mixture was circulated through a
cartridge filter. The filter was washed with EtOH (386 kg).
The combined filtrates were charged into the vessel and HCl
gas (45.7 kg, 1.5 equiv) was added at 15−23 °C within 1 h 45
min to reach pH < 0.5. (Caution: exothermic addition!). The
suspension was cooled to 10 °C, stirred at 10 °C for 30 min
and filtered. The filter cake was washed with EtOH (540 kg)
and dried in a tray dryer at 50−60 °C, 100 mbar for 16 h to
afford 12a·HCl as an off-white solid. Yield: 123.4 kg (70%). Mp
217 °C; assay (HPLC, free base): 82.2% (theoretical maximum
82% w/w for mono-HCl salt); purity (HPLC method 2):
CONCLUSION
■
A practical synthesis of ACT-280778 (1, as the bis-maleate salt)
was developed and demonstrated on kilogram scale. Using the
present process, 14 kg of drug substance was prepared, enough
to complete phase II clinical trials. This became only possible
after the manufacturing of large amounts of chiral bicyclic
ketone 3 had been secured. A first achievement was the scalable
synthesis of the benzimidazole side chain 4 based on a solvent
(NMP) as atom-efficient reagent, and on the stable phenyl-
enediamine hydrochloride 12a. The development of a
convenient way to isolate carboxylic acid 2 in diastereoisomeri-
cally pure form was crucial for purging impurities from the
preceding steps. The use of sodium methoxide to selectively
afford 1 from bis-acylated byproduct 10 and the identification
of a convenient access to bis-maleate salt of 1 were the two
other salient features. This route consists of a total of 7 + 3
steps from optically pure ketone 3 with 38% yield for the
longest linear sequence. Notably, no chromatographic
purification was required despite most of the intermediates
being oils. Although further investigations will be required
before larger scale campaigns may be undertaken, the work
presented herein resulted in a rapid progression of this
candidate into early clinical trials.
1
99.0% a/a, tR = 3.9 min; H NMR (D6-DMSO): δ 6.99−8.96
(br s, 4 H), 6.56 (s, 2 H), 3.77 (s, 6 H); Anal. Calcd. For
C8H12N2O2·HCl: C: 46.95; H: 6.40; N: 13.69; Cl: 17.32.
Found: C: 46.74; H: 6.41; N: 13.54; Cl: 17.05.
3-(4,7-Dimethoxy-1H-benzo[d]imidazol-2-yl)-N-meth-
ylpropan-1-amine dihydrochloride (4·2 HCl). A 64-L
glass-lined reactor was charged with 21.5% HCl (41.4 kg, 4.1
equiv) and NMP (14.9 kg, 2.54 equiv). After degassing with N2
at 20 °C for 15 min, sodium sulfite (350 g, 0.05 equiv) and 12a·
HCl (12.1 kg, assay 82.2% w/w) were added, and the mixture
was stirred at 107−109 °C for 19 h. IPC (HPLC) indicated less
than 0.1% 12a·HCl. In a 400-L stainless-steel reactor, n-butanol
(109 kg), 30% NaOH (52.4 kg) and water (20 kg) were added,
followed by degassing with N2 at 20 °C for 15 min. The black
mixture from the 64-L reactor was added to the 400-L reactor
at 20−30 °C. (Caution: exothermic quench!) The pH was >12.
After phase separation the organic phase was washed with water
(2 × 33 kg) and transferred into a 200-L glass-lined reactor.
37% HCl (10.4 kg, 1.8 equiv) was added at 20−30 °C to adjust
the pH to <3. After addition of activated charcoal (10.0 kg, 1
EXPERIMENTAL SECTION
■
General Remarks. Respectively, 1 vol or 1 wt means 1 L of
solvent or 1 kg of reagent, with respect to the reference starting
material; er (enantiomeric ratio), dr (diastereomeric ratio); er
and dr reported in this paper have not been validated by
1
calibration. Compounds are characterized by H NMR (400
MHz, Bruker) or 13C NMR (100 MHz, Bruker); internal
standard for quantitative NMR was 1,4-dimethoxybenzene
affording the assay data (in % w/w) in this paper. Details for
H
dx.doi.org/10.1021/op400269b | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
wt), the mixture was stirred at 60−80 °C for 2 h. After cooling
to 20 °C, Celite (7.0 kg) was added, and the mixture was stirred
at 20−25 °C for 30 min. The following filtration (200-L
reactor) and distillation (in the cleaned 64-L reactor) were run
in parallel. The mixture was pumped over a filter nutsche and
circulated back into the 200-L reactor. After the solution was
clear, it was transferred into the 64-L reactor via a 5-μm PAL
filter. As soon as the 64-L reactor contained 30−40 L, solvent
was removed at 70−80 °C (head temperature 60−70 °C) and
300 mbar. At the end of the filtration, the filter cake was washed
with warm (50 °C) n-butanol (57 kg), and the filtrate was
added to the 64-L reactor. The distillation was resumed in the
64-L reactor until approximately 40 L of residual volume (a
total of 140−160 L was removed). An aliquot was sampled at
approximately 60 °C to determine the water content by Karl
Fischer. Water was added to adjust to 2.5−3.5% water. The
mixture was cooled to 0 °C within 3 h and filtered. The filter
cake was washed with AcOEt (2 × 38 kg) and dried in a tray
dryer at 70 °C, <100 mbar for 15 h to afford 4·2HCl
(monohydrate) as an off-white powder. Yield: 12.1 kg (59%).
Mp 174 °C (DSC peak temperature); assay (HPLC method 2,
free base): 72.4% (theoretical maximum 73% w/w for
monohydrate dihydrogen chloride salt); purity (HPLC method
2): 100.0% a/a, tR = 3.4 min; water: 5.2% (Karl Fischer);
sulphated ash: 0.51%; ICP: Pd < 0.1 ppm, Na: 0.17%, Cl:
20.9%; color: R7; 1H NMR (D6-DMSO): δ 14.48−15.81 (br s,
2 H), 9.34 (br s, 2 H), 7.00 (s, 2 H), 4.48−5.64 (br 1, 2 H),
3.95 (s, 6 H), 3.23 (t, J = 7.4 Hz, 2 H), 2.86−2.97 (m, 2 H), 2.5
(s, under DMSO peak, 3 H), 2.23 (quint, J = 7.5 Hz, 2 H); 13C
NMR (D6-DMSO): δ 152.9, 141.2, 122.9, 106.8, 56.7, 47.2,
32.5, 23.8, 23.6. Data of analytical sample: Anal. Calcd. For
C13H19N3O2·2 HCl·H2O: C: 45.89; H: 6.81; N: 12.35; O:
14.10; Cl: 20.84. Found: C: 45.11; H: 6.51; N: 12.22; O: 14.44;
Cl: 21.05.
tert-Butyl 2-((1R,2R,4R)-2-Hydroxy-5-phenylbicyclo-
[2.2.2]oct-5-en-2-yl)acetate (Mixture of Diaster-
eoisomers, 6 and 7). To a 2.5 M solution of hexyl lithium
in hexane (12.6 L, 31.5 mol, 1.25 equiv) in toluene (60 L, 12
vol) at −15 °C was added dropwise diisopropylamine (4.6 L,
32.8 mol, 1.3 equiv) between −15 °C and +5 °C. The resulting
solution was stirred at −15 to −10 °C for 10 min. The
suspension was cooled to −25 °C and tert-butyl acetate (4.25 L,
31.5 mol, 1.25 equiv) was added dropwise between −25 and
−15 °C. The mixture was aged at −25 °C for 10 min. It was
cooled to −80 °C, and a solution of 3 (5.0 kg, 25.2 mol, 1
equiv) in toluene (10 L, 2 vol) was added dropwise between
−85 and −75 °C in 30−45 min. The resulting turbid solution
was stirred at −78 °C for 30 min, warmed to −25 °C, and
quenched with 20% w/w citric acid (30 L, 6 vol) at <10 °C.
The organic layer was separated and washed with water (20 L,
4 vol) at 20 °C to pH ≥ 5−7. The organic phase was
concentrated to 7−8 vol at atmospheric pressure. EtOH
(absolute, 90 L) was added, and the solution was again
concentrated to 7−8 vol. The solution was used as such in the
next stage as a mixture of diastereoisomers (dr: 4.8:1). Yield:
100%, used as is. Purity (HPLC method 1): 76.6% a/a (6), tR
12.5 min; 16.0% a/a (7), tR 13.0 min; [M − tBu − H2O + 1]+ =
241. An analytical sample was purified by chromatography over
silica gel. 1H NMR (CDCl3): δ 7.45−7.43 (m, 2H), 7.38−7.35
(m, 2H), 7.29−7.25 (m, 1H), 6.51 (br d, J = 6.9 Hz, 1H), 4.1
(br s, 1H), 3.16 (br s, 1H), 2.68−2.67 (m, 1H), 2.40−2.34 (m,
3H), 1.85−1.80 (m, 1H), 1.76−1.73 (m, 1H), 1.49 (br s, 10
H), 1.42−1.33 (m, 1H), 1.24−1.15 (m, 1H); 13C NMR
(CDCl3): δ 172.9, 144.8, 138.6, 128.5 (2 C), 127.2, 127.1,
124.8 (2 C), 81.6, 74.1, 47.5, 42.1, 41.7, 33.8, 28.1 (3 C), 24.3,
19.7.
2-((1R,2R,4R)-2-Hydroxy-5-phenylbicyclo[2.2.2]oct-5-
en-2-yl)acetic Acid (2). KOH (3.3 kg, 50.4 mol, 2 equiv) in
water (10 L) at 30−35 °C was added to the previously
described solution of 6 and 7 (dr 4.8:1) in EtOH. The resulting
turbid mixture was heated at reflux (∼80 °C) for 1 h. The
reaction mixture was cooled to 35 °C and concentrated to ∼5
vol. 2-Methyl-THF (64 L) and water (12 L) were added, and
the mixture was concentrated to ∼5 vol. 2-Methyl-THF (17 L),
and 2 M HCl (5.6 L) were added to the mixture until pH ≈ 2.0
at 10−20 °C. Water (12 L) was added, and the mixture was
heated and stirred at 35 °C to obtain a good separation. The
following solvent exchange was repeated 3 times: the organic
phase was concentrated at atmospheric pressure to 38 L, AcOEt
(40 L) was added, and the solution was concentrated at
reduced pressure to 38−40 L at ≤45 °C. At 1 atm, the solution
was allowed to cool to 65 °C. At this temperature,
crystallization started. The mixture was stirred at 55−65 °C
for 1 h. The suspension was cooled to 5 °C. It was filtered over
filter cloth, and the filter was washed with cold (∼5 °C) AcOEt
(8 L). The filter cake was dried under a flow of nitrogen for 10
h. Yield: 3.83 kg (59%, two steps). Purity (HPLC method 1):
1
99.0% a/a, tR 1.52 min, [M − H2O + 1]+ = 241; H NMR
(MeOD): δ 7.42−7.39 (m, 2H), 7.33−7.29 (m, 2H), 7.24−
7.20 (m, 1H), 6.51(dd, J = 7.0, 1.8 Hz, 1H), 4.87 (br s, 2H),
3.15−3.12 (m, 1H), 2.75−2.72 (m, 1H), 2.42 (br s, 2H), 2.33−
2.25 (m, 1H), 1.83−1.76 (m, 1H), 1.69−1.68 (m, 2H), 1.43−
1.36 (m, 1H), 1.23−1.15 (m, 1H); 13C NMR (MeOD): δ
174.2, 145.1, 138.5, 128.1 (2 C), 126.6, 126.5, 124.5 (2 C),
74.0, 46.8, 41.8, 41.1, 33.8, 23.9, 19.4.
N-(3-(4,7-Dimethoxy-1H-benzo[d]imidazol-2-yl)-
propyl)-2-((1R,2R,4R)-2-hydroxy-5-phenylbicyclo[2.2.2]-
oct-5-en-2-yl)-N-methylacetamide (8). A reactor was
charged with EDCI·HCl (5.2 kg, 27.1 mol, 1.27 equiv),
HOBT hydrate (1.47 kg, 10.9 mol, 0.5 equiv), and 4·2 HCl
(7.73 kg, 24.0 mol, 1.1 equiv). THF (105 L) was added. In
another reactor, Et3N (11.05 kg, 109.2 mol, 5 equiv) was added
at 15−25 °C to a solution of 2 (5.54 kg, 21.4 mol, 1.0 equiv) in
THF (22 L). The solution of 2 was added dropwise at 10 to 25
°C to the first vessel. The reaction mixture was stirrred at 20−
25 °C for 23−30 h until conversion (HPLC method 1) was
>98%. THF was exchanged with DCM under reduced pressure
at 45 °C. At 10−20 °C, the organic phase was washed twice (33
and 24 kg respectively) with 2 M HCl followed by 2 M NaOH
to reach pH = 12 and water (27 L). DCM was exchanged to
THF under reduced pressure at 45 °C, and the resulting
solution of 8 was used as such. Yield: 10.2 kg (97%, used as is);
weight of the solution: 59.72 kg, concentration by evaporation
of an aliquot: 21.0% w/w, HPLC assay of the aliquot: 81.3% w/
w. Purity (HPLC method 1): 97.4% a/a, tR 3.11 min, [M + 1]+
1
= 490; water content: 0.05% w/w; H NMR (major rotamer,
MeOD): δ 7.42−7.39 (m, 2H), 7.32−7.28 (m, 3H), 7.23−7.19
(m, 1H), 7.16−7.12 (m, 1H), 6.96 (br s, 2H), 6.50 (dd, J = 6.9,
1.7 Hz, 1H), 4.0 (br s, 6H), 3.55−3.51 (m, 2H), 3.14−3.10 (m,
3H), 3.05 (br s, 3H), 2.72−2.69 (m, 1H), 2.54−2.52 (m, 2H),
2.33−2.23 (m, 1H), 2.17−2.08 (m, 2H), 1.82−1.74 (m, 1H),
1.71−1.50 (m, 2H), 1.42−1.29 (m, 1H), 1.21−1.13 (m, 1H);
13C NMR (MeOD): δ 173.6, 152.8, 145.2, 141.1, 138.4, 128.1
(2 C), 127.9, 126.7, 126.6, 124.4 (2 C), 123.0, 106.0 (2 C),
74.8, 55.5 (2 C), 48.6, 46.2, 44.0, 41.7, 41.6, 35.0, 33.8, 24.4,
23.9, 23.5, 19.4.
I
dx.doi.org/10.1021/op400269b | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
(1R,2R,4R)-2-(2-((3-(4,7-Dimethoxy-1H-benzo[d]-
imidazol-2-yl)propyl) (methyl)amino) ethyl)-5-
phenylbicyclo[2.2.2]oct-5-en-2-ol (9). A reactor was
charged with 10% w/w solution of lithium aluminium hydride
in THF (15.9 L, 41.9 mol, 2.0 equiv). THF was added (41 L),
and the resulting solution was cooled to 0−10 °C. The above
solution of crude 8 in THF (17.4% w/w, 58.45 kg, 20.8 mol)
was added dropwise to keep the temperature <15 °C. The
reaction mixture was allowed to warm up to 20 °C for 1.5 h. A
solution of KOH (1.61 kg, 1 wt relative to LAH) in water (3.2
L, 2 wt relative to LAH) was added cautiously at 0−15 °C
followed by water (1.6 L, 2 wt relative to LAH). The
suspension was stirred at 0−15 °C for 10 min. After filtration
over Celite, the filtrate was washed with THF (2 × 20 L) and
concentrated under reduced pressure to 7.5 vol at T ≤ 45 °C.
THF (20 L) was added, and the solution was concentrated
under reduced pressure to 7.5 vol at T ≤ 45 °C. The solution of
9 in THF (H2O: 0.87% w/w) was used as such in the next
stage. Yield: 9.0 kg (91%, used as is); weight of the solution:
63.15 kg, concentration by evaporation of an aliquot: 14.3% w/
w. Purity (HPLC method 1): 98.3% a/a, tR 1.85 min, [M + 1]+
= 476. An analytical sample was purified by chromatography
over silica gel. 1H NMR (MeOD): δ 7.38−7.36 (m, 2H), 7.30−
7.27 (m, 2H), 7.21−7.18 (m, 2H), 6.60 (br s, 2H), 6.48 (br d, J
= 6.9 Hz, 1H), 3.91 (s, 6H), 3.11−3.09 (m, 1H), 2.88 (t, J = 7.4
Hz, 2H), 2.63−2.59 (m, 3H), 2.46 (t, J = 7.4 Hz, 2H), 2.33−
2.22 (m, 2H), 2.25 (s, 3H), 2.05−1.97 (m, 2H), 1.80−1.75 (m,
1H), 1.64−1.50 (m, 4H), 1.40−1.34 (m, 1H), 1.19−1.13 (m,
1H); 13C NMR (MeOD): δ 153.4, 144.5, 138.6, 128.1 (2 C),
127.0, 126.5, 124.4 (3 C), 102.1 (2 C), 76.0, 67.4, 56.5, 54.9 (2
C), 52.5, 42.1, 41.4, 40.7, 38.6, 33.9, 26.1, 25.8, 25.3, 25.1, 24.3,
19.3.
(1R,2R,4R)-2-(2-((3-(4,7-Dimethoxy-1H-benzo[d]-
imidazol-2yl)propyl)(methyl)amino)ethyl)-5-
phenylbicyclo[2.2.2]oct-5-en-2-ol Oxalate (9·2 C2H2O4).
A THF solution of 9 (63.15 kg, 14.3% w/w) was charged into a
reactor. THF was exchanged with iPrOH at 50 °C under
reduced pressure, and the resulting solution was transferred
into a drum. A solution of oxalic acid dihydrate (5 kg, 39.9 mol,
2.1 equiv) in water (9 L), iPrOH (18 L) and AcOEt (9 L) was
stirred at 70−75 °C. The above-prepared solution of 8 was
added dropwise. The mixture was heated to 70−75 °C for 15
min. AcOEt (100 L) was added, followed by aging at 70−75 °C
for 2 h. The reaction mixture was allowed to cool to 5 °C over
4 h and filtered (filtration time: 38 h), rinsed with AcOEt and
dried under a flow of nitrogen for 3 days. Yield: 10.80 kg (87%,
3 steps). KF: 3.0% w/w; purity (HPLC method 1): 96.5% a/a,
tR 1.85 min, [M + 1]+ = 476; 1H NMR (D6-DMSO): δ 8.9 (br
s, 4H), 7.39−7.37 (m, 2H), 7.32−7.27 (m, 2H), 724−7.19 (m,
1H), 6.59 (br s, 2H), 6.53 (dd, J = 6.9, 1.4 Hz, 1H), 3.85 (br s,
6H), 3.23−3.08 (m, 5H), 2.93−2.89 (m, 2H), 2.73 (br s, 3H),
2.54−2.50 (m, 1H), 2.20−2.10 (m, 3H), 1.77−1.65 (m, 3H),
1.54−1.40 (m, 2H), 1.30−1.24 (m, 1H), 1.09−1.02 (m, 1H);
13C NMR (D6-DMSO): δ 163.7 (4 C), 152.8, 144.2, 142.9,
138.5, 129.8, 128.9, 127.9 (2 C), 127.3, 125.0 (2 C), 103.1,
73.7, 67.5, 56.1 (2 C), 55.3, 51.6, 42.0, 41.6, 37.5, 33.4, 26.1,
25.6, 24.6, 21.9, 20.2.
afford a clear biphasic mixture. The layers were separated, and
the organic layer was dried azeotropically. Et3N (4.20 kg, 41.0
mol, 2.5 equiv) was added to the solution of 9 in 2-methyl-THF
at 0−5 °C. Isobutyroyl chloride (2.65 kg, 24.7 mol, 1.5 equiv)
was added at 0−10 °C over 15 min. (Caution: delay in
exotherm!) The turbid mixture was allowed to warm to rt and
then stirred for 45 min. At this point, 92.7% conversion (HPLC
method 3) was observed. Et3N (1.50 kg, 14.8 mol, 0.9 equiv)
was added at 0−10 °C over 5 min, followed by isobutyroyl
chloride (1.55 kg, 14.8 mol, 0.9 equiv) at 0−10 °C over 5 min.
After 45 min, full consumption of the starting material was
observed (HPLC method 3). 30% w/w NaOMe in MeOH
(10.55 kg, 49.4 mol, 3 equiv) was added dropwise at 0−10 °C.
Stirring was continued for 45−60 min at 10−15 °C. Water (53
L) was added at 10−20 °C to the organic phase. The layers
were separated. 10% w/w citric acid solution (50 L) was added
at 10−20 °C until 5.5 ≤ pH ≤ 6.5. The phases were separated,
and 10 wt % NaOH (∼1 equiv) was added until pH ≥ 13. After
phase separation, the organic phase was washed with water (32
L) and concentrated at reduced pressure to 5−6 vol AcOEt (85
L) was added, and the solution was concentrated at ≤45 °C
under reduced pressure to 5−6 vol. This exchange with AcOEt
was repeated twice. The solution of 1 in AcOEt was used as
such in the next step. Yield: 8.79 kg (98%, used as is); weight of
the solution: 45.3 kg, concentration by evaporation of an
aliquot: 19.4% w/w. Purity (HPLC method 3): 98.8% a/a, tR
3.26 min, [M + 1]+ = 546, 1H NMR (MeOD): δ 7.39−7.37 (m,
2H), 7.31−7.27 (m, 2H), 7.23−7.20 (m, 1H), 6.59 (br s, 2H),
6.42 (dd, J = 7.1, 1.6 Hz, 1H), 3.90 (s, 6H), 3.25−3.22 (m,
1H), 3.15−3.14 (m, 1H), 2.86 (t, J = 7.4 Hz, 2H), 2.55−2.48
(m, 1H), 2.42−2.37 (m, 4H), 2.18 (s, 3H), 2.16−2.11 (m, 1H),
2.08−1.87 (m, 6H), 1.72−1.64 (m, 2H), 1.45−1.38 (m, 1H),
1.25−1.18 (m, 1H), 1.16 (s, 3H), 1.14 (s, 3H); 13C NMR
(MeOD) δ 176.8, 153.0, 146.0, 142.8, 137.9, 129.3, 128.1 (2
C), 126.9, 124.9, 124.5 (3 C), 102.2, 85.4, 56.6, 54.8 (2 C),
51.2, 40.7, 40.0, 38.6, 36.1, 34.7, 34.3, 33.3, 26.4, 23.7, 23.6,
19.7, 19.0, 18.1, 18.0.
(1R,2R,4R)-2-(2-((3-(4,7-Dimethoxy-1H-benzo[d]-
imidazol-2-yl)propyl)(methyl)-amino)ethyl)-5-
phenylbicyclo[2.2.2]oct-5-en-2-yl Isobutyrate Maleate
(1·2 C4H4O4). A filtered solution of crude 1 in AcOEt (45.0
kg, 19.4% w/w) was heated at 35−40 °C for 10 min. EtOH (7
L) was added. A filtered solution of maleic acid (3.7 kg, 32.0
mol, 2.0 equiv) in MeOH (7 L) was slowly added at 35−40 °C
over 10−15 min. After a rinse with EtOH (2.5 L), the solution
was stirred at 35−40 °C for 30 min. Seed crystals (10 g) were
added at 35 1 °C. The mixture was stirred at 35−40 °C for 2
h. The suspension was cooled to 20 °C over 2 h. AcOEt (181
L) was added at 20−25 °C over 1 h. The mixture was cooled to
5 °C over 1 h. The suspension was filtered, and the filter cake
was washed with AcOEt (3 × 17.5 L). The white solid was
dried under a flow of nitrogen for 2 days. Yield (9.40 kg, 74%,
two steps). Purity (HPLC method 3): 99.7% a/a, tR 3.26 min,
1
[M + 1]+ = 546, H NMR (MeOD): δ 7.33−7.20 (m, 5H),
6.73 (br s, 2H), 6.41 (dd, J = 7.0, 1.3 Hz, 1H), 6.25 (s, 4H),
3.91 (s, 6H), 3.30−3.11 (m, 9H), 2.86 (s, 3H), 2.63−2.56 (m,
1H), 2.53−2.38 (m, 2H), 2.31−2.24 (m, 2H), 2.05−1.93 (m,
2H), 1.76−1.69 (m, 2H), 1.47−1.39 (m, 1H), 1.29−1.22 (m,
1H), 1.20 (d, J = 2.4 Hz, 3H), 1.18 (d, J = 2.4 Hz, 3H); 13C
NMR (MeOD) δ 176.9, 168.8, 151.8, 146.3, 141.9, 137.7, 133.9
(4 C), 128.2 (2 C), 127.0, 126.1, 124.5 (3 C), 124.4, 104.3 (2
C), 84.6, 55.7, 55.1 (2 C), 51.3, 40.4, 38.8, 38.5, 34.6, 33.2,
33.0, 25.0, 23.6, 21.5, 19.6, 18.0, 17.9; HRMS (ESI) for [M +
(1R,2R,4R)-2-(2-((3-(4,7-Dimethoxy-1H-benzo[d]-
imidazol-2-yl)propyl)(methyl)amino)ethyl)-5-
phenylbicyclo[2.2.2]oct-5-en-2-yl Isobutyrate (1). 9·2
C2H2O4 (10.75 kg, 16.4 mol) was suspended in 2-methyl-
THF (85 L). A solution of 85% KOH (5.40 kg, 82.0 mol, 5
equiv) in water (41 L) was added dropwise until pH > 13 to
J
dx.doi.org/10.1021/op400269b | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
H+] C33H44N3O4: Calcd. 546.3332; Found: 546.3334. Anal.
Calcd. For C41H51N3O12: C: 63.31; N: 5.40; O: 24.68. Found:
C: 63.23; N: 5.31; O: 24.85.
2794. (b) After completion of this work, a similar dinitration−
hydrogenation sequence was reported: Besset, T.; Morin, C. Synthesis
2009, 1753−1756.
(12) Maybe indicative of NOx formed during reaction.
(13) Bulk amounts of 1,4-dimethoxybenzene (5) are typically
delivered as white chips. Delumping with a mortar gave faster
dissolution; however, on large scale, 5 was used as received.
(14) Two consecutive crystallizations from HOAc (8−9 vol)
improved the regioisomeric ratio from 8.8:1 to 29:1 with a yield of
60%. One crystallization from toluene (4−6 vol) raised the ratio to
13:1 with 90% yield.
(15) On kilo-scale, the quench on a water/ice mixture was
endothermic: the temperature was −20 °C after quench. A mixture
of conc. HNO3 and ice (2:1) is commonly known as freezing mixture
to reach −50 °C.
(16) (a) Hydrogenation of 12a on PtO2: Weinberger, L.; Day, A. R.
J. Org. Chem. 1959, 24, 1451−1455. (b) With sodium dithionite:
Shaikh, I. A.; Johnson, F.; Grollman, A. P. J. Med. Chem. 1986, 28,
1329−1340. (c) With H2 on Pd/C in AcOEt, see ref 11.
(17) (a) Kosak, J. R. Chem. Ind. (Dekker) 1988, 33 (Catal. Org.
React.), 135−147. (b) Tong, W. R.; Seagraves, R. L.; Wiederhorn, R.
Loss Prev. 1977, 11, 71−75.
(18) When the filtration was performed carefully under inert
conditions, the clear filtrate was yellow that turned pink and finally
dark-violet after short contact with air.
ASSOCIATED CONTENT
* Supporting Information
Analytical methods and characterization data of 1, 4·2 HCl, 12a
and 12b, DSC traces of 11 and 12, X-ray analyses of rac-6, and
experimental details for 17. This material is available free of
charge via the Internet at http://pubs.acs.org.
■
S
AUTHOR INFORMATION
Corresponding Author
*stefan.abele@actelion.com. Telephone: +41 61 565 67 59.
■
Present Address
⊥Lonza AG, Rottenstrasse 6, CH-3930 Visp, Switzerland.
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
J.-A.F. and S.A. thank Dr. Dorte Renneberg, Dr. Francis Hubler,
and Dr. Kurt Hilpert from Drug Discovery Chemistry for the
■
(19) When running the hydrogenation with a pure sample of the
regioisomer 11b, the yield was <20%, although the Celite filter bed
́
fruitful collaboration; Stephanie Combes, Luke Harris, Roman
Inauen, and Ivan Schindelholz for their skillful experimental
work; Dr. Daniel Bur for the X-ray structures; and Dr. Markus
von Raumer for sharing his expertise during the salt screening.
We are indebted to Dr. Thomas Weller for continuous support.
1
was copiously washed with EtOH and acetone (280 vol). The H
NMR of this product was differerent from that of 12a and
corresponded to 12b.
(20) The black solution was stable for 1−2 weeks at rt in the dark.
Longer storage led to precipitation of a sticky tar.
REFERENCES
(21) (a) The lower yield with aqueous HCl was attributed to the
water solubility of 12a. (b) The stoichiometry of the salt was proven
by elemental analysis and chloride titration.
(22) Considering the 88% content of 11a in the starting material and
the assay of the crude product prior to HCl salt formation, the
maximum theoretical yield is approximately 80%. Hence, approx-
imately 10% of the product was lost in the mother liquor. This was
corroborated by HPLC assay of the mother liquor; when the
temperatures for filtration were 10 and 5 °C, 10.4% and 8.9% w/w
of 12 HCl were detected, respectively.
(23) HCl and water have a negative azeotrope with a bp of 110 °C
and 20.2% w/w HCl. The HCl concentration varied between 18.5 and
21.5% for seven consecutive batches on 10-kg scale.
(24) Although dichloromethane gave a similar recovery during the
extraction at pH 12, it was not appropriate for the salt formation and
crystallization.
■
(1) Part of this work was preliminarily disclosed: Funel, J.-A. In
Practical Synthesis of L/T Calcium Channel Blocker ACT-280778, 30th
SCI Process Development Symposium, Cambridge, UK, December
5−7, 2012; Funel, J.-A.; In Practical Synthesis of 5-Phenylbicyclo[2.2.2]
oct-5-en-2-one toward L/T Calcium Channel Blocker ACT-280778.
Application of the Diels−Alder Reaction on kg-Scale, 1st Smart Synthesis
and Advanced Purification Conference, April 21−23, 2013; Lyon, FR.
(2) (a) Hilpert, K.; Hubler, F.; Renneberg, D. WO/2008/132679A1,
2008. (b) Hubler, F.; Hilpert, K.; Renneberg, D. WO/2009/
130679A1, 2009.
(3) Funel, J.-A.; Schmidt, G.; Abele, S. Org. Process Res. Dev. 2011, 15,
1420−1427.
(4) Abele, S.; Schwaninger, M.; Fierz, H.; Schmidt, G.; Funel, J.-A.;
Stoessel, F. Org. Process Res. Dev. 2012, 16, 2015−2020.
(5) (a) Abele, S.; Inauen, R.; Funel, J.-A.; Weller, T. Org. Process Res.
Dev. 2012, 16, 129−140. (b) Abele, S.; Funel, J.-A. WO/2012/
052943A1, 2012. (c) Abele, S.; Funel, J.-A. WO/2012/052939A2,
2012.
(25) The redox activity of demethylated benzoquinone-2,5-diamines
is used in coordination chemistry, molecular switches, and catalysis,
see: Schweinfurth, D.; Weisser, F.; Sarkar, B. Nachr. Chem. 2009, 57,
462−466.
(26) Demethylation of related dimethoxybenzimidazoles, see:
́ ́
(a) Alvarez, F.; Gherardi, A.; Nebois, P.; Sarciron, M.-E.; Petavy, A.-
(6) Abele, S.; Inauen, R.; Spielvogel, D.; Moessner, C. J. Org. Chem.
2012, 77, 4765−4773.
(7) Z-GABA (14): CHF 3000/kg.
F.; Walchshofer, N. Bioorg. Med. Chem. Lett. 2002, 12, 977−979.
(b) Weinberger, L.; Day, A. R. J. Org. Chem. 1959, 24, 1451−1455.
(27) (a) A neat melt process followed by distillation as described for
the condensation of caprolactame or pyrrolidinone with o-phenyl-
enediamine led to decomposition: Botta, A. DE 2321054, 1973; Botta,
A. Liebigs Ann. Chem. 1976, 336−337. (b) A basic method that relied
on the condensation with a nitrile in the presence of sodium
methoxide only gave traces of product with decomposition:
Zarrinmayeh, H.; Nunes, A. M.; Ornstein, P. L.; Zimmerman, D.
M.; Arnold, M. B.; Schober, D. A.; Gackenheimer, S. L.; Bruns, R. F.;
Hipskind, P. A.; Britton, T. C.; Cantrell, B. E.; Gehlert, D. R. J. Med.
Chem. 1998, 41, 2709−2719. (c) 2-Imino-1-methylpyrrolidinone as
electrophile for the nucleophilic attack by 2,5-dimethoxyaniline gave
no product: Rasmussen, C. R.; Gardocki, J. F.; Plampin, J. N.;
Twardzik, B. L.; Reynolds, B. E.; Molinari, A. J.; Schwartz, N.;
(8) Differential scanning calorimetry (DSC) of a mixture 11a:11b =
7.7:1 shows an exotherm with left limit 290 °C, peak 338 °C, and the
decomposition energy of −3154 kJ/kg.
(9) Phillips, M. A. J. Chem. Soc. 1928, 2393−2399.
(10) Glutamic acid: Cescon, L. A.; Day, A. R. J. Org. Chem. 1962, 27,
581−586. (b) β-Alanine: Revankar, G. R.; Siddappa, S.; Umarani, D.
C. Indian J. Chem. 1964, 2, 489−490. (c) Sarcosine: Cardwell, T. J.;
Edwards, A. J.; Harthorn, R. M.; Holmes, R. J.; McFasyen, W. D. Aust.
J. Chem. 1997, 50, 1009−1015. (d) 4-Dimethylaminobutanoic acid:
Menichincheri, M.; Ballinari, D.; Bargiotti, A.; Bonomini, L.;
Ceccarelli, W.; D’Alessio, R.; Fretta, A.; Moll, J.; Polucci, P.; Soncini,
C.; Tibolla, M.; Trosset, J. Y.; Vanotti, E. J. Med. Chem. 2004, 47,
6466−6475.
(11) (a) Hammershøj, P.; Reenberg, T. K.; Pittelkow, M.; Nielsen, C.
B.; Hammerich, O.; Christensen, J. B. Eur. J. Org. Chem. 2006, 2786−
K
dx.doi.org/10.1021/op400269b | Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
Bennetts, W. W.; Price, B. E.; Marakowski, J. J. Med. Chem. 1978, 21,
1044−1054. (d) No product was obtained via N-chloroamidines of
2,5-dimethoxyaniline: Ichikawa, M.; Nabeya, A.; Muraoka, K.; Hisano,
T. Org. Prep. Proced. Int. 1978, 10, 205−209. (e) A recent protocol
relying on the use of NaH as base for the reaction of N-
methylbenzimidazoles with carbonitriles was not tried: Sluiter, J.;
Christoffers, J. Synlett 2009, 63−66.
(28) Charcoal treatments (Acticarbone 3S, CECA, 1 wt) of (a) the
reaction mixture, 80 °C, 2 h, and of (b) the n-butanol phase after
basification with 32% NaOH, 50 °C, 2 h.
(29) More benzimidazole 4·2 HCl had to be added to reach full
conversion, and the phase separations were difficult due to intensive
coloration.
(30) A degassed solution of the diamine hydrochloride 12a·HCl in
water−acetonitrile in the dark got colored within 6−12 h, whereas the
same solution containing 0.5% w/w sodium sulfite did not show
coloration within several days. However, as long as the solvents and
reagents were degassed with N2, no effect on the color of the product
4·2HCl was observed.
(31) Chapter 2.2.2, red colors R1−R7, higher numbers indicating less
colorized solutions.
(32) There is no direct example with the bicyclo[2.2.2]octene
framework. However, the following references allow the comparison
with additions to related bicyclic systems: Devaux, J. F.; Hanna, I.;
Lallemand, J. Y.; Prange, T. J. Org. Chem. 1993, 58, 2349−2350.
(b) Almqvist, F.; Torstensson, L.; Gudmundsson, A.; Frejd, T. Angew.
Chem., Int. Ed. 1997, 36, 376−377. (c) Olsson, C.; Helgesson, S.;
Frejd, T. Tetrahedron: Asymmetry 2008, 19, 1484−1493.
(33) Freshly prepared LDA gave full conversion in contrast to the
commercially available solution.
(34) No conversion was observed when a Reformatsky-type reaction
(using Zn and tert-butyl bromoacetate) was attempted.
(35) iPrOH was also suitable as the solvent for diastereoselective
crystallization, but gave 5−10% reduction in yield.
(36) For a recent review on industrial applications of the Diels−Alder
reaction: Funel, J.-A.; Abele, S. Angew. Chem., Int. Ed. 2013, 52, 3822−
3863; Angew. Chem. 2013, 125, 3912−3955.
(37) Details to be disclosed elsewhere.
(38) (a) Fleming, M. P.; Harrington, P. J. WO/98/49149A1, 1998;
(b) Harrington, P. J.; Wong, J. WO/98/49148A1, 1998.
(39) On 3.5-kg scale, the solid reagents (1 equiv carboxylic acid 2, 1
equiv benzimidazole 4 and 1.1 equiv CDMT) were suspended in
DCM. DIPEA was added followed by the addition of N-
methylmorpholine. After 1−2 h at 20 °C, the reaction was completed.
(40) Impurity 19 posed less of an issue since it was converted as well
to amine 9 and diol 22 after hydride reduction. The latter was
ultimately purged away in the final crystallization step.
(41) It was observed that the filtration of aluminum salts derived
from the hydrolysis with KOH as base performed better than with
NaOH.
(42) Even though this issue was anticipated in the laboratory, time
constraints did not allow for removing this issue prior to scale up.
(43) Spontaneous deacylation of 10 to 1 was also detected in a crude
batch of 10 exposed to daylight over a week.
(44) Abele, S.; Combes, S.; Funel, J.-A.; Hilpert, K.; Hubler, F.;
Reichenbaecher, K.; Renneberg, D.; Von Raumer, M. WO/2010/
046857A1, 2010.
L
dx.doi.org/10.1021/op400269b | Org. Process Res. Dev. XXXX, XXX, XXX−XXX