Discovery and development of an efficient process to atovaquone

Article abstract of DOI:10.1021/op300165q

The discovery and development of an efficient and more sustainable manufacturing route to the antipneumocystic agent atovaquone (2-((1R,4R)-4-(4-chlorophenyl)cyclohexyl)-3-hydroxynaphthalene-1,4-dione) 1 is described. The existing commercial route to atovaquone delivers a poor yield of product and uses expensive reagents. The new synthesis commences with readily available phthalic anhydride, which is converted to 1,4-isochromandione 5 and then to atovaquone 1 by reaction with 4-(4-chlorophenyl)cyclohexanecarboxylic acid 3 using key bromination, Rosenmund reduction, and rearrangement chemistries. Downstream processing to atovaquone is both high yielding and robust, and the resulting process has been demonstrated on 200-kg scale. The process is simple, uses cheap raw materials, and is more sustainable in that it avoids lowyielding silver-promoted chemistry and isomerisation procedures. It includes a robust, facile, and highly efficient procedure to 1,4-isochromandione 5, and routes to 4-(4-chlorophenyl)cyclohexanecarboxaldehyde 9 have also been developed, including a Rosenmund method that was demonstrated on pilot-plant scale. Also discussed are the route-derived impurities and processing amendments to control their formation.

Full text of DOI:10.1021/op300165q

Article
pubs.acs.org/OPRD
Discovery and Development of an Efficient Process to Atovaquone
Hugh Britton, David Catterick, Andrew N. Dwyer, Andrew H. Gordon, Stuart G. Leach,*
Chris McCormick, Clive E. Mountain, Alec Simpson, David R. Stevens, Michael W. J. Urquhart,*
Charles E. Wade, John Warren, Nick F. Wooster, and Audrey Zilliox
Medicines Research Centre, Glaxo Group Limited, Gunnels Wood Road, Stevenage, Hertfordshire SG1 2NY, U.K.
S
* Supporting Information
ABSTRACT: The discovery and development of an efficient and more sustainable manufacturing route to the anti-
pneumocystic agent atovaquone (2-((1R,4R)-4-(4-chlorophenyl)cyclohexyl)-3-hydroxynaphthalene-1,4-dione) 1 is described.
The existing commercial route to atovaquone delivers a poor yield of product and uses expensive reagents. The new synthesis
commences with readily available phthalic anhydride, which is converted to 1,4-isochromandione 5 and then to atovaquone 1 by
reaction with 4-(4-chlorophenyl)cyclohexanecarboxylic acid 3 using key bromination, Rosenmund reduction, and rearrangement
chemistries. Downstream processing to atovaquone is both high yielding and robust, and the resulting process has been
demonstrated on 200-kg scale. The process is simple, uses cheap raw materials, and is more sustainable in that it avoids low-
yielding silver-promoted chemistry and isomerisation procedures. It includes a robust, facile, and highly efficient procedure to
1,4-isochromandione 5, and routes to 4-(4-chlorophenyl)cyclohexanecarboxaldehyde 9 have also been developed, including a
Rosenmund method that was demonstrated on pilot-plant scale. Also discussed are the route-derived impurities and processing
amendments to control their formation.
from the aldehyde 4-(4-chlorophenyl)cyclohexanecarb-
oxaldehyde 9. The aldehyde 9 has been mentioned within the
literature, but a synthetic method has never been reported.5
To demonstrate the feasibility of the concept, a simpler model
aldehyde 19 was chosen, and initial work was focused on
developing a more efficient synthesis of the 1,4-isochromandione
5. Although there are several methods described which included
oxidation of 1,3-indanedione6 and acidification of 2-diazoacetyl-
benzoic esters,7 methods involving bromination of 2-acetylbenzoic
acid 10 appeared the most promising for scale-up.4,8,9 The earliest
reported synthesis of 5 using this approach was by Gabriel
in 188410 although the structure was incorrectly assigned as
3-formylphthalide 11.9 Evaluation of the bromination of
2-acetylbenzoic acid 10 following the conditions of Estevez4
and Yang8 led to incomplete reactions and the generation of
benzylbromide from bromination of the toluene cosolvent.
Removal of toluene gave an immediate improvement, whereby
the substrate 10 was brominated in glacial acetic acid using a slight
excess of bromine. Interestingly, 2-bromoacetylbenzoic acid was
not observed but rather the cyclised phthalides 12, 13, and 14 in
a ratio of 4:4:1. The reaction appears to be acid catalysed in that
the rate of conversion to products increases as the hydrobromic
acid byproduct forms. The importance of acid has also been
confirmed with remarkably slow bromination reactions resulting
when sodium acetate was used as a buffer, or when bromine was
replaced with N-bromosuccinimide. A small quantity of water was
added in order to cut down on the formation of the dibromide 13,
and it was beneficial to run the reaction with sub-stoichiometric
quantities of bromine to reduce the formation of the poly-
brominated byproducts (14 and 15). Under these new conditions,
INTRODUCTION
■
Atovaquone is approved for marketing in the United States under
the trade name Mepron for the treatment and prophylaxis of
Pneumocystis carinii infection.1 It is also available in combination
with proguanil hydrochloride under the trade name Malarone for
the treatment and prevention of Plasmodium falciparum malaria.
Herein is described the limitations of the current route, the
new route investigation and subsequent scale-up as well as
identification and control of some of the key route impurities.
RESULTS AND DISCUSSION
■
Existing Route to Atovaquone. The existing route to
atovaquone 12 proceeds by the reaction of 2-chloro-1,4-
naphthoquinone 2 and 4-(4-chlorophenyl)cyclohexane-1-car-
boxylic acid 3 in the presence of silver nitrate and ammonium
persulphate, followed by extraction with chloroform (Scheme 1).
Hydrolysis of the intermediate substituted 2-chloro-1,4-
napthoquinone 4 gives a low overall yield of product due to
the inefficiency of the key radical coupling stage. There are other
routes to atovaquone described in the literature3 which follow a
similar approach but are limited by similarly low yields or the use
of non-ideal reagents, reactions, or purification conditions.
Identification of a New Route to Atovaquone 1. To
address the aforementioned issues, we explored alternative
approaches to the synthesis of atovaquone 1, specifically those
which avoided the use of the low-yielding radical couplings
described above. The work of Estevez et al.4 described
condensation of 1,4-isochromandione 5 with benzaldehyde 6,
and the resulting lactone 7 was rearranged to the corresponding
2-hydroxynaphthoquinone 8 using sodium methoxide (Figure 1).
It was unclear whether this transformation could be extended
to aliphatic carboxaldehydes, given their potential for self-
condensation, but if it could, then atovaquone could be prepared
Received: June 19, 2012
Published: September 11, 2012
© 2012 American Chemical Society
1607
dx.doi.org/10.1021/op300165q | Org. Process Res. Dev. 2012, 16, 1607−1617

Organic Process Research & Development
Article
Scheme 1. Existing route to atovaquone
Figure 1. Preparation of 2-hydroxy-3-phenylnaphthoquinone, retrosynthetic analysis of atovaquone to 4-(4-chlorophenyl)cyclohexanecarboxaldehyde
9, and identification of a suitable model system 19.
reasonable levels of acetylbenzoic acid (10) would always remain
(7−10%) with low levels of polybrominated materials persisting,
highlighting the poor selectivity for this reaction. Attempts to
convert the crude brominated intermediate 12 to 1,4-
isochromandione 5 using sodium acetate4,9 in methanol led to
an immediate red colouration and the formation of polymeric
byproduct, most notably 17. Other buffered cyclisation conditions
were evaluated, and use of dipotassium hydrogenphosphate,
delivered low levels of product but this was accompanied by
the formation of the condensation byproduct 16, 17, and 18
(Scheme 2). It was clear that 1,4-isochromandione 5 is highly
reactive under these cyclisation conditions, so that another method
was needed.
overall result, only in this instance a clean sample of the
decomposition product 17 was also isolated. The formation of
compounds 16 and 17 were attributed to the presence of
hydrobromic acid leading to enolisation of the intended product
5 which then condensed with another molecule of 5 to deliver
aldol product 16 and subsequent acid-mediated dehydration
secured the diene 17.12
Reaction of the key isochromandione 5 with the model
carboxaldehyde 19 using the conditions highlighted by Estevez4
delivered an excellent conversion to the aldol product 20.
Subsequent treatment with sodium methoxide in methanol
proceeded to the desired naphthoquinone 21 in very good yield
(Scheme 3).
The success of the model system meant that an efficient route
to the isochromandione 5 was now essential, and the prohibitive
cost and limited supply base for 2-acetylbenzoic acid (10)
highlighted that a useful route to this compound would also
be beneficial. Approaching these in turn, a solvent screen was
conducted for the bromination of acetylbenzoic acid (10) which
Returning to the original report by Gabriel,11 the crude
bromomethyl phthalides 12 and 13 were suspended in water, and
the mixture was refluxed to deliver the target isochromandione 5
in poor overall yield (12%) with the dimeric compound 16
being identified as the major byproduct. Telescoping the initial
bromination with an aqueous hydrolysis delivered a similar
1608
dx.doi.org/10.1021/op300165q | Org. Process Res. Dev. 2012, 16, 1607−1617

Organic Process Research & Development
Article
Scheme 2. Bromination of acetylbenzoic acid 10 and conversion to 1,4-isochromandione 5
Scheme 3. Reaction of 1,4-isochromandione, 5, with cyclohexanecarboxaldehyde and subsequent rearrangement to 2-cyclohexyl-
3-hydroxynaphthalene-1,4-dione, 21
Scheme 4. Robust routes to 1,4-isochromandione, 5, and 2-acetylbenzoic acid, 10
highlighted that chlorobenzene was a useful solvent for this
transformation. Indeed, this solvent proved remarkably beneficial
in that over-bromination appeared to be inhibited and complete
consumption of 10 was now possible.13 It was also clear that the
major product under these conditions was the dibromide 13
(Scheme 4). This observation, together with the remarkable sel-
ectivity and the requirement for acid catalysis, strongly advocated
a mechanism whereby conversion of 10 to the isomeric
3-hydroxy-3-methylphthalide 22 is required followed by
dehydration to the olefin 23 which is then brominated to give
13 as the major product. The water produced within the reaction
is effectively removed as a biphase which inhibits formation of the
alcohol 12 which would be needed for the polybrominated
species 14 and 15 to form. It was rationalized that conducting a
thermal cyclisation of this mixture under biphasic conditions
might inhibit the formation of dimeric byproduct because the
hydrobromic acid catalyst would be effectively removed from the
system. With this in mind, once bromination was completed,
water was added, and the resulting biphasic mixture was heated
to reflux whereupon a rapid conversion to alcohol 12 was
observed followed by a smooth conversion to the target 1,4-
isochromandione, 5. The product could then be isolated in very
good yield following a simple extraction, evaporation, and
crystallization procedure. This process was ultimately exempli-
fied on pilot-plant scale where a total of 250 kg of 5 was prepared.
There are several routes described in the literature for the
preparation of 2-acetylbenzoic acid 10, but for this work we
focused on those involving phthalic anhydride and malonic acid
as they were the most cost-effective. These materials have been
reacted together using pyridine,14 and triethylamine,15,16 as
solvent/bases, and there is a report where various solvent/base
combinations were evaluated for this transformation.17 Evalua-
tion of these conditions highlighted that using triethylamine as
solvent and base gave the cleanest conversion to 10. The final
process was based on the conditions reported by Rangnekar and
Telange,15 where malonic acid was added in portions to a
1609
dx.doi.org/10.1021/op300165q | Org. Process Res. Dev. 2012, 16, 1607−1617

Organic Process Research & Development
Article
Scheme 5. Initial routes to 4-(4-chlorophenyl)cyclohexanecarboxaldehyde 9 and conversion to atovaquone 1
suspension of phthalic anhydride in triethylamine at 80 °C;
subsequent acid-mediated hydrolysis and crystallization gave
good yields of pure product without the need for further
purification. This process was exemplified in the pilot plant
where a total of 350 kg of 10 was prepared.
impurity 27 from 1 using crystallization proved unsuccessful;
therefore, alternative amines were screened to avoid its future
formation. For early development work, morpholine proved to
be a useful alternative, but this was ultimately replaced by
isobutylamine which gave superior yields (Schemes 5 and 7).
The methoxide-induced chemistry where the lactone 26 is
converted through to 1 proved to be more complex than was
initially thought. HPLC analysis of the reaction mixture showed
the rapid conversion into the intermediate methyl ester 29 as
expected, as well as the hydrolysis product 30, but the lactone 31
was also formed. On continued reaction it became apparent that
both the ester 29 and lactone 31 were converted slowly through
to the target product 1, whereas the hydrolysis product 30 took
no further part within the reaction. Further evidence for the rapid
formation of the intermediates followed by slower conversion
into product 1 was obtained from ReactIR data (Figure 2).
Authentic samples of intermediates 29 and 31 were readily
accessed from methanolysis of the lactone 26 using dimethyla-
minopyridine in methanol with and without toluene as a
cosolvent, respectively. Both of these materials were individually
reacted with sodium methoxide to give 1 in excellent yields.
Formation of the hydrolysis product 30 was attributed to residual
water; however, unlike intermediates 29 and 31, 30 does not
convert to 1 under the reaction conditions, so the removal of
water became key to obtaining high yields for this chemistry.
A standard of 30 was easily prepared by reacting the lactone 26
with potassium hydroxide in methanol whereby an 86% yield of
the hydrolysis product was obtained (Scheme 6).
Replacing sodium methoxide with sodium ethoxide for the
conversion of the lactone 26 to 1 was also successfully demon-
strated, although the rate of this reaction was significantly
reduced, presumably due to steric factors. Therefore, this
alternative reagent was not pursued further.
Having demonstrated the new route to 1, alternative
approaches to the key aldehyde 9 were considered as the routes
described above required expensive reagents and complicated
processing. The classical Rosenmund reduction21 was of
particular interest, given the availability of 4-(4-chlorophenyl)-
cyclohexanecarboxylic acid 3. Also the conversion of carboxylic
acids to their acid chlorides is typically clean and high yielding,
and catalytic hydrogenations are generally simple and atom-
efficient processes. Therefore, this kind of approach should
simplify the process to deliver 9. Initial work to prepare the acid
chloride 32 from acid 3 demonstrated that oxalyl chloride with
Having secured a route to the key 1,4-isochromandione 5 and
exemplified the feasibility of the new route to atovaquone from a
successful model study, it was now important to prepare the
relevant carboxaldehyde 9. Given that the existing manufacturing
route to atovaquone used 4-(4-chlorophenyl)cyclohexane-
carboxylic acid 3 and that both this acid and it is methyl ester
24 are known,18,19 it seemed appropriate to use these molecules
as starting points for the synthesis of the aldehyde 9. Treatment
of the ester 24 with lithium borohydride, or lithium aluminium
hydride, in tetrahydrofuran gave the corresponding alcohol 25 in
excellent yield which was in turn converted through to the target
aldehyde 9 using the modified Swern conditions20 of pyridine
sulfur trioxide in dimethylsulfoxide. A more direct route to
this aldehyde could be achieved by reacting ester 24 with
diisobutylaluminium hydride, although this approach did require
low-temperature conditions and was always accompanied by
some over-reduction to the alcohol 25. Subsequent condensa-
tion of the carboxaldehyde 9 with isochromandione 5 gave an
excellent yield of the corresponding conjugated ketone 26 which
rearranged to atovaquone on treatment with sodium methoxide.
The aldehyde 9 proved difficult to handle, given that it was a low-
melting solid and susceptible to further oxidation to the acid 3.
With this in mind, the processes to the aldehyde 9 were further
modified such that 9 was not isolated but rather used as a solution
within the next stage. The resulting atovaquone derived from this
process proved of reasonable purity, but a new impurity was
observed at low level (approximately 0.4%) which was ultimately
characterised as the pyridine 27. The dibenzopyranopyridine
motif in 27 appears to be a novel ring system and looks to
have formed from the condensation of a molecule of the iso-
chromandione 5 with the unsaturated ketone 26 and ammonia.
Indeed, heating the isochromandione 5 with the unsaturated
ketone 26 in the presence of ammonium acetate in an acetic
acid/toluene mixture led to the target pyridine 27 being prepared
albeit in low yield. Also formed during this process was the
pyridine 28 which appeared to have been formed from the
condensation of two molecules of 5 with ammonia and
“formaldehyde” although the mechanism for this transformation
is not clear. Attempts to remove the dibenzopyranopyridine
1610
dx.doi.org/10.1021/op300165q | Org. Process Res. Dev. 2012, 16, 1607−1617

Organic Process Research & Development
Article
Figure 2. React IR profile showing the formation of the methyl ester 29 and its conversion through to atovaquone 1.
Scheme 6. Rearrangement of lactone 26 to atovaquone 1 and
respective intermediates
the robustness was not sufficient to allow routine operation.
Therefore, alternative solutions were sought. A range of organic
and inorganic bases were screened in this chemistry. Inter-
estingly, we found that the choice of base had a significant effect
on the rate of dechlorination compared to the desired reaction,
with good correlation between pKa of the base and the extent
of dechlorination. Stronger bases (such as triethylamine) gave
increased dechlorination, whereas weaker bases suppressed
this side reaction. If the base was too weak, the rate of desired
reaction was unacceptably slow, as was observed with 2,6-di-tert-
butylpyridine and tetramethylpyrazine. However, it was found
that the slightly less basic 2-methylquinoline (quinaldine) gave
a good compromise, leading to a very clean and robust reac-
tion. Gratifyingly, it was also found that the catalyst system
with quinaldine as base was not susceptible to poisoning from
the crude acid chloride input, contrary to what was seen when
2,6-lutidine was used. It was therefore unnecessary to carry
out a carbon treatment prior to hydrogenation, and this
step was removed from the process. The reaction also proved
to be hampered by the presence of water which leads to the
formation of the symmetrical anhydride 33 and can have a
profound impact on yield. With this in mind, it was necessary
to use anhydrous catalyst, and excellent conversion to pro-
duct could be achieved under these conditions. Once again,
it proved beneficial to telescope the chemistry such that
isolation of the highly air-sensitive aldehyde 9 was avoided
(Scheme 7).
catalytic DMF gave a smooth conversion to product.22 Standard
palladium on charcoal proved a superior choice to the classic
poisoned palladium on barium sulfate catalyst due to the relative
stability of the aliphatic acid chloride. 2,6-Lutidine was used as
the base such that conversion to product 9 could be delivered.23
It was, however, necessary to carry out a carbon treatment of the
input acid chloride solution prior to hydrogenation, in order to
avoid catalyst poisoning and reaction stalling. Careful control of
the reaction time was required in order to control the competing
reduction of the aromatic chloride which resulted in the
persisting des-chloro impurity. Initial attempts to minimize this
over-reduction focused on choice of metal catalyst as well as
loading, but while it was possible to improve the selectivity and
achieve a reaction which gave product of the required purity,
The use of thionyl chloride rather than oxalyl chloride for the
Rosenmund chemistry looked economically attractive. Unfortu-
nately, in our hands the resulting hydrogenation of acid chloride
32 derived from reacting 3 with thionyl chloride was retarded
significantly, possibly due to catalyst poisoning. After optimiza-
tion of the acid chloride formation, Rosenmund reduction, and
condensation sequence key lactone 26 was prepared in 81%
overall yield on pilot-plant scale.
1611
dx.doi.org/10.1021/op300165q | Org. Process Res. Dev. 2012, 16, 1607−1617

Organic Process Research & Development
Article
Scheme 7. Rosenmund route to 4-(4-chlorophenyl)cyclohexanecarboxaldehyde, 9
1,4-Isochromandione (5). A stirred mixture of 2-acetylben-
zoic acid 10 (82 kg, 500 mol) and chlorobenzene (905 kg, 820 L)
was treated with 5.5 M hydrobromic acid in acetic acid (5 L) and
bromine (25.2 L, 78.6 kg, 492 mol) and then warmed to
approximately 30 °C. After 3 h, water (820 L) was added and the
reaction heated to reflux. After 3 h the reaction was cooled to
60 °C and the organic layer removed. The aqueous layer was
extracted with chlorobenzene (364 kg, 328 L), and the combined
organic layers were concentrated under reduced pressure to
approximately 260 L. Propan-2-ol (322 kg, 410 L) was charged,
and the slurry was cooled to 0 °C and filtered. The damp cake was
washed with propan-2-ol (128 kg, 163 L then a further 64 kg,
82 L) and then dried under vacuum at 50 °C to give the title
compound as pale-yellow/brown crystals (59 kg, 73% th yield,
100% a/a HPLC purity); 1H NMR (400 MHz, CDCl3): δH 5.14
(2H, s), 7.83−7.91 (2H, m), 8.08−8.10 (1H, m), 8.28−8.30
(1H, m); 13C NMR (100 MHz, CDCl3): δC 73.4, 125.6, 128.0,
130.9, 131.8, 134.6, 135.9, 161.3, and 189.5; HRMS (APCI,
negative ion, [M − H]−) calculated for C9H5O3, 161.0244, found
161.0241.
4′-Hydroxy-3′,4′-dihydro-1H,1′H-3,4′-bi-2-benzopyran-
1,1′,4(3H)-trione (16). Isochromandione 5 (10 g, 61.7 mmol)
was suspended in THF (50 mL), stirred under nitrogen, and
10% aqueous potassium hydrogen carbonate (1 m L) added. The
mixture was stirred overnight at room temperature and then
clarified by filtration. The solution was concentrated at reduced
pressure to give a brown oil. Trituration with THF (10 mL) and
TBME (10 mL), filtration, and drying under vacuum gave the
title compound as a cream-colored solid (6.6 g, 67% yield): 1H
NMR (400 MHz, DMSO-d6): δH 4.31 (1H, d), 4.50 (1H, d), 5.33
(1H, s), 6.57 (1H, s, exchangeable), 7.60−7.66 (2H, m), 7.81−
7.84 (1H, m), 7.94−8.01 (3H, m), 8.05 (1H, d), 8.18 (1H, m).
13C NMR (100 MHz, DMSO-d6): δC 68.9, 71.4, 84.6, 122.7,
125.1, 127.1, 128.1, 129.0, 129.4, 129.5, 132.5, 133.5, 134.6,
135.7, 142.7, 161.5, 163.4, 190.0; HRMS (positive ion electro-
spray, MH+) calculated for C18H13O6 325.0712, found 325.0719.
4-Hydroxy-1H,1′H-3,4′-bi-2-benzopyran-1,1′-dione (17).
Isochromandione 5 (10 g, 61.7 mmol) was suspended in acetic
acid (30 mL) with stirring under nitrogen, and isobutylamine
(1 mL, 10 mmol) was added after which there was some fuming.
The reaction mixture was heated to 100 °C for 18 h. The mixture
was cooled to room temperature and the product collected by
filtration, washed with acetic acid (10 mL), and then dried under
vacuum at 50 °C to give the title compound as a pale-yellow solid
SUMMARY
■
To conclude, a sustainable manufacturing route to atova-
quone has been discovered and developed which is robust and
higher yielding than the original supply route in that atovaquone
1 is delivered in a total yield of 71% from carboxylic acid 3
compared to 20% from the existing route using a comparable
number of steps. The new route also has significant sustainability
advantages, with a demonstrated mass efficiency of 1.50%, which
is expected to rise to 2.03% with solvent recovery, compared to
0.93% for the current manufacturing route24 (see Scheme 8).
The new synthesis exemplifies classical bromination and
Rosenmund chemistries whilst also taking advantage of a key
rearrangement to provide atovaquone. In addition, the utility of
the under-utilized but highly effective Rosenmund reduction as a
tool in the preparation of aldehydes on a large scale is described.
EXPERIMENTAL SECTION
■
General. Commercially available reagents were used without
further purification. Reactions requiring anhydrous conditions
were performed under an atmosphere of dry nitrogen. 4-(4-
Chlorophenyl)cyclohexanecarboxylic acid and it is methyl ester
were obtained commercially from custom synthesis suppliers.
The experimental conditions below refer to the procedures used
for the multikilogram pilot-plant campaign. Key NMR spectra
are included within the Supporting Information.
2-Acetyl Benzoic Acid/3-Hydroxy-3-methylisobenzofuran-
1(3H)-one (10/22). A stirred mixture of phthalic anhydride (170
kg, 1.15 kmol) and triethylamine (171 kg, 1.23 kmol) were
heated to 80 °C. Ten equal portions of malonic acid (10 × 14.4
kg; 144 kg total, 1.38 kmol) were charged over a period of 4 h,
and the reaction mixture was maintained at 80 °C for a further 10
h. Hydrochloric acid (4 M, made up from 365 kg of concentrated
aqueous hydrochloric acid and 530 kg of water) was charged and
the reaction stirred for a further 30 min at 80 °C before being
cooled to 25 °C, and the resulting slurry was filtered. The damp
cake was washed with water (2 × 340 kg) and then dried under
vacuum at 50 °C to give the title compound as light-brown
1
crystals (126 kg, 67% th yield, 98.7% a/a HPLC purity); H
NMR (700 MHz, DMSO): δH 2.51 (3H, s, CH3), 2.47 (1H, br s,
OH) 7.52−7.70 (2H, m), 7.79−7.80 (2H, d); 13C NMR (100
MHz, DMSO): δC 26.0, 106.4, 122.4, 124.4, 130.2, 134.6, 150.4
and 167.8; HRMS (positive ion electrospray, MH+) calculated
for C9H9O3 165.0546, found 165.0644.
1612
dx.doi.org/10.1021/op300165q | Org. Process Res. Dev. 2012, 16, 1607−1617

Organic Process Research & Development
Article
Scheme 8. Comparison of existing manufacturing route to atovaquone 1 with new process
(6.7 g, 71%): 1H NMR (400 MHz, DMSO-d6): δH 7.61 (1H, d),
7.68−7.75 (2H, m), 7.88 (1H, m), 7.93 (1H, s), 7.94−8.00 (2H,
m), 8.23 (1H, d), 8.27 (1H, d), 9.19 (1H, s); 13C NMR (100
MHz, DMSO-d6): δC 110.1, 120.6, 120.9, 122.2, 124.6, 129.1,
129.1, 129.2, 129.3, 131.1, 134.5, 135.0, 135.4, 135.5, 135.6,
148.0, 160.7, 160.9; HRMS (positive ion electrospray, MH+)
calculated for C18H11O5 307.0601, found 307.0605.
room temperature for 23 h. The red solution was treated with
acetic acid (2 mL, 35.1 mmol), but no solids precipitated, and a
red solution persisted. The dark solution was diluted with water
(18 mL), whereby a yellow solid precipitated, and the suspension
was stirred for 10 min. The solid was collected by filtration under
vacuum, washed with water (3 × 20 mL), and pulled dry to give
1
the title compound as a yellow solid (2.55 g, 85%); H NMR
(Z)-3-(Cyclohexylmethylene)isochroman-1,4-dione (20).
1,4-Isochromandione 5 (4 g, 24.67 mmol) was suspended in
acetic acid (30 mL), cyclohexane carboxaldehyde 19 (3.1 mL,
25.9 mmol) was added, and the mixture was treated with a slurry
of ammonium acetate (2 g, 25.9 mmol) in acetic acid (10 mL) at
35 °C. The suspension was heated to 55 °C and stirred for 15 h
during which period a solid was precipitated. The suspension was
diluted with water (40 mL) and stirred for 10 min. The solid was
collected by filtration under vacuum, washed with water (3 ×
20 mL), and pulled dry to give the title compound (5.73 g, 91%)
(400 MHz, CDCl3): δH 1.23−1.46 (m, 3), 1.56−1.67 (m, 2H),
1.68−1.77 (m, 1H), 1.76−1.87 (m, 2H), 1.98 (qd, J = 12.4, 2.9,
2H), 3.08 (tt, J = 12.3, 3.5, 3.4, 1H), 7.45 (s, 1H), 7.66 (td, J = 7.6,
1.2, 1H), 7.74 (td, J = 7.6, 1.2, 1H), 8.06 (dd, J = 7.6, 1.2, 1H),
8.11 (dd, J = 7.8, 1.0, 1H); 13C NMR (100 MHz, CDCl3): δC
26.0, 26.7, 29.2, 35.2, 125.9, 126.9, 127.9, 129.2, 132.7, 133.2,
134.9, 152.8, 181.9, 184.6.
[trans-4-(4-Chlorophenyl)cyclohexyl]methanol (25). Lith-
ium borohydride (2.0 g, 91.8 mmol) was added to a stirred
solution of methyl ester 24 (20.0 g, 79.1 mmol) in tetrahy-
drofuran (50 mL) at ambient temperature under argon; the
reaction mixture was heated to reflux and stirred for 2 h. The
reaction mixture was carefully quenched with 1 M aqueous
hydrochloric acid (50 mL, 50 mmol), and then ethyl acetate
(50 mL) was added. The aqueous was separated and extracted
with ethyl acetate (50 + 25 mL), and then the combined organics
were washed with water (50 mL) and brine (20 mL). Theresulting
organic phase was dried (MgSO4), filtered, and concentrated
undervacuumto givethe title compound asa pale-yellow oil which
crystallised on standing to give a cream-colored solid (17.1 g,
1
as a pale-yellow solid; H NMR (400 MHz, CDCl3): δH1.16−
1.32 (m, 3H), 1.33−1.48 (m, 2H), 1.65−1.83 (m, 5H), 2.89 (q,
J = 10.7, 1H), 6.39 (d, J = 10.0, 1H), 7.79−7.91 (m, 2H), 8.19−
8.27 (m, 1H), 8.28−8.36 (m, 1H); 13C NMR (100 MHz,
CDCl3): δC 25.3, 25.8, 31.8, 34.6, 126.7, 127.4, 130.6, 131.4,
133.5, 134.8, 135.0, 145.1, 158.7, 176.7.
2-Cyclohexyl-3-hydroxynaphthalene-1,4-dione (21). (Z)-3-
(Cyclohexylmethylene)isochroman-1,4-dione 20 (3 g, 11.71
mmol) was suspended in methanol (18 mL), sodium methoxide
(3.2 mL, 14.05 mmol) was added, and the mixture was stirred at
1613
dx.doi.org/10.1021/op300165q | Org. Process Res. Dev. 2012, 16, 1607−1617

Organic Process Research & Development
Article
1
96%); H NMR (400 MHz, CDCl3): δH 1.07−1.17 (2H, m),
stirred solution of methyl trans-4-(4-chlorophenyl)cyclo-
hexanecarboxylate 24 (50.5 g, 200 mmol) in tetrahydrofuran
at 0 °C under argon and the mixture stirred for 30 min. The
reaction mixture was cooled to 0 °C, 1 M aqueous hydrochloric
acid (500 mL, 500 mmol) was carefully added, and the mixture
stirred for 5 min. The aqueous was separated and extracted with
ethyl acetate (250 mL), and then the combined organic phases
were washed with water (2 × 125 mL). The resulting organic
phase was concentrated under vacuum to a volume of approxi-
mately 150 mL, ethyl acetate (250 mL) was added and the
mixture concentrated under vacuum to a volume of 150 mL.
This “put and take” procedure was repeated a two more times.
Dimethyl sulfoxide (100 mL) and triethylamine (112 mL,
804 mmol) were added to the stirred mixture, at ambient
temperature under argon, and the mixture was cooled to 0 °C.
Pyridine sulphur trioxide complex (64 g, 402 mmol) was added
in 4 equal portions, and the reaction mixture stirred between
0 and 20 °C for 2 h. Ethyl acetate (400 mL) was added, the
mixture was cooled to 0 °C with stirring, and 1 M aqueous
hydrochloric acid (400 mL, 400 mmol) was added. The resulting
mixture was stirred at ambient temperature for 5 min, and then
the organic layer was separated, diluted by the addition of ethyl
acetate (500 mL), and washed with water (250 mL). The
resulting organics were concentrated under vacuum to a volume
of 200 mL, the residue was diluted by the addition of ethyl acetate
(500 mL), and then concentrated to a volume of 200 mL. Acetic
acid (250 mL), isochromandione 5 (32.4 g, 200 mmol), and
morpholine (17.5 mL, 200 mmol) were added to the resulting
stirred mixture, at room temperature under argon. The reaction
mixture was heated to 40 °C and stirred for 4 h and allowed to
cool to ambient and stir overnight. Water (250 mL) was added
and the slurry stirred for 15 min and then filtered. The filter cake
was washed with TBME (2 × 125 mL) and dried to give the title
compound as a yellow solid (62.4 g, 85%). This material was
spectroscopically identical to that reported above.
1.33−1.35 (1H, m), 1.39−1.49 (2H, m), 1.52−1.61(1H, m),
1.91−1.94 (4H, m), 2.42−2.50 (1H, d), 3.50−3.53 (2H, m),
7.12−7.14 (2H, m), 7.24−7.26 (2H, m); 13C NMR (100 MHz,
CDCl3): δC 29.7, 33.6, 40.1, 44.0, 68.5, 128.2, 128.4, 131.5, 145.9;
HRMS (APCI, negative ion, [M − H]−) calculated for C13H16ClO
223.0895, found 223.0894.
trans-4-(4-Chlorophenyl)cyclohexanecarbaldehyde (9).
Triethylamine (2.2 mL, 16 mmol) was added to a stirred
solution of alcohol 25 (0.9 g, 4 mmol) in dimethyl sulfoxide
(4 mL) at ambient under argon. A solution of pyridine sulphur
trioxide complex (1.27 g, 8 mmol) in dimethyl sulfoxide (8 mL)
was added and the reaction mixture stirred at ambient for 1 h.
The stirred reaction mixture was cooled to 15 °C, and then water
(30 mL) and ethyl acetate (10 mL) were added. The organic
layer was separated, diluted by the addition of ethyl acetate
(20 mL), and washed with water (20 mL), 1 M aqueous hydro-
chloric acid (20 mL, 20 mmol), and 50% saturated aqueous brine
(20 mL). The resulting organic phase was dried (MgSO4),
filtered, and concentrated under vacuum to give the title com-
pound as an oily solid (852 mg, 95%);25,26 1H NMR (400 MHz,
CDCl3): δH 1.37−1.53 (4H, m), 2.00−2.02 (2H, m), 2.11−2.14
(2H, m), 2.25−2.33 (1H, m), 2.45−2.51 (1H, m), 7.11−7.15
(2H, m) and 7.25−7.28 (2H, m), 9.68 (1H, s); 13C NMR (100
MHz, CDCl3): δC 26.2, 32.9, 43.2, 49.8, 128.1, 128.5, 131.8,
145.1, and 204.2; HRMS (APCI, negative ion, [M − H]−)
calculated for C13H14ClO 221.0739, found 221.0739.
If necessary, further purification could be achieved in the
following manner. Crude aldehyde 9 (600 mg, 2.7 mmol) was
suspended in n-heptane (2 mL) with stirring at ambient temp-
erature. The resulting solution was cooled to −5 °C and the
resulting slurry stirred for 30 min at −5 °C. The slurry was
filtered, and then the filter cake was washed with cold, −5 °C,
n-heptane (2 mL) and dried to give the title compound as a white
solid (360 mg, 60%).
(3Z)-3-{[trans-4-(4-Chlorophenyl)cyclohexyl]methylidene}-
1H-2-benzopyran-1,4(3H)-dione (26). Isochromandione 5
(0.51 g, 3.13 mmol) and ammonium acetate (0.24 g, 3.13 mmol)
were added to a stirred solution of the carboxaldehyde 9 (0.87 g,
3.13 mmol) in acetic acid (5 mL). The mixture was heated to 60 °C,
stirred for 1.5 h, and then cooled to ambient. Water (5 mL) was
added dropwise to the stirred mixture. The yellow solid was
collected by filtration under vacuum, washed with water (2 × 5 mL)
and TBME (10 mL), and then dried to give the title compound as
a pale-yellow solid (0.83 g, 72%). 1H NMR (400 MHz, CDCl3):
δH 1.38−1.48 (2H, m), 1.53−1.64 (2H, m), 1.93−1.98 (4H, m),
2.48−2.56 (1H, m), 2.88−3.00 (1H, m), 6.40 (1H, d, J = 10.0),
7.15 (2H, m), 7.27 (2H, m), 7.85−7.89 (2H, m), 8.24 (1H, m),
8.33 (1H, m); 13C NMR (100 MHz, CDCl3): δC 31.2, 33.2, 34.3,
43.1, 113.6, 126.8, 128.1, 128.2, 128.5, 128.5, 130.6, 130.7, 133.4,
135.0, 135.1, 145.5, 158.6 and 176.7; HRMS (APCI, negative ion,
[M + e]−) calculated for C22H19ClO3 366.1028, found 366.1028.
A second crop of material was available from the filtration
liquors by dilution with 2-methyltetrahydrofuran (5 mL), separa-
tion of the aqueous phase, and extraction with 2-methyltetrahy-
drofuran (5 mL). The combined organic phases were dried
(MgSO4), the solvent was removed under vacuum and the
residual oil triturated with TBME (5 mL), giving the title
compound as a pale-yellow solid (90 mg, 8%).
DIBAH Route to (3Z)-3-{[trans-4-(4-Chlorophenyl)-
cyclohexyl]methylidene}-1H-2-benzopyran-1,4(3H)-dione
(26). A 1 M solution of diisobutylaluminium hydride in dichloro-
methane (110 mL, 110 mmol) was added to a stirred solution
of methyl trans-4-(4-chlorophenyl)cyclohexanecarboxylate 24
(25.3 g, 100 mmol) in dichloromethane at −78 °C under argon,
and the mixture was stirred for 90 min. Methanol (125 mL)
was added to the stirred mixture which was allowed to warm
to −10 °C after which 1 M aqueous hydrochloric acid (250 mL,
250 mmol) was added. The aqueous layer was extracted with
dichloromethane (250 mL), and the combined organics were
washed twice with water (2 × 125 mL). The organics were
concentrated under vacuum to a volume of 75 mL, the residue
was diluted by the addition ethyl acetate (125 mL), and the
mixture was then concentrated under vacuum to a volume of
75 mL. Acetic acid (125 mL), isochromandione 5 (16.2 g,
100 mmol), and ammonium acetate (7.7 g, 100 mmol) were
added to the resulting stirred mixture, at ambient temperature
under argon. The reaction mixture was then heated to 60 °C and
distilled for 2 h. The resulting slurry was cooled to ambient
temperature; water (75 mL) was added, stirred for 30 min, and
then filtered. The filter cake was washed with TBME (62.5 mL)
and dried to give the title compound as a yellow solid (24.8 g,
68%). This material was spectroscopically identical to that
reported above.
Reduction/Swern Oxidation Route to (3Z)-3-{[trans-4-(4-
Chlorophenyl)cyclohexyl]methylidene}-1H-2-benzopyran-
1,4(3H)-dione (26). A 1 M solution of lithium aluminium
hydride in tetrahydrofuran (100 mL, 100 mmol) was added to a
Rosenmund Route to (3Z)-3-{[trans-4-(4-Chlorophenyl)-
cyclohexyl]methylidene}-1H-2-benzopyran-1,4(3H)-dione
(26). A suspension of 4-(4-chlorophenyl)cyclohexane-1-carboxylic
1614
dx.doi.org/10.1021/op300165q | Org. Process Res. Dev. 2012, 16, 1607−1617

Organic Process Research & Development
Article
acid (3) (50.0 kg) in ethyl acetate (300 L) with catalytic DMF
(82.5 mL) was heated to 55 °C, and oxalyl chloride (19.5 L) was
added, followed by a line wash of ethyl acetate (30 L). This mixture
was stirred at 55 °C until all solids had dissolved and reaction
was complete. The mixture was distilled down to 150 L and cooled
to 20 °C, and quinaldine (39.8 L) was added, followed by a line
wash of ethyl acetate (30 L). The mixture was transferred to a
hydrogenation vessel containing 5% Pd/C (3 kg), followed by
a line wash of ethyl acetate (250 L). The reaction mixture was
stirred under hydrogen gas at 20 °C until reaction was complete.
The mixture was filtered to remove the catalyst, washing with
ethyl acetate (150 L) and charged to a separate vessel. 1,4-
Isochromandione (5) (32.6 kg) was added, washing with ethyl
acetate (30 L) followed by addition of acetic acid (150 L) and
isobutylamine (56.5 L) and a line wash of ethyl acetate (30 L).
The mixture was heated to 38 °C and stirred at this temperature
until reaction was complete and cooled to 20 °C. The product
was collected by filtration, washing with isopropanol (2 × 125 L)
before drying under vacuum at 70 °C to give the title com-
pound as a white solid (62.2 kg, 81% th yield, 99.6% a/a HPLC
purity).27 This material was spectroscopically identical to that
reported above.
by filtration and washed with acetonitrile (2 × 2 mL) to give the
title compound and pyridine 28 as a grey solid (1.06 g). Further
purification was achieved by suspending the crude 27 and 28
(0.50 g) in DMSO (10 mL), stirring, and heating to 110−120 °C.
The resulting suspension was cooled to ambient and the solid
collected by filtration under vacuum, washed with DMSO (2 mL)
followed by methanol (2 × 3 mL), and dried to give a grey solid
consisting of 27 and 28 in a ratio of 2.3:1. 1H NMR (400 MHz,
CDCl3): δH 1.75 (2H, m), 1.93 (2H, m), 2.06 (2H, m), 2.58 (2H,
m), 2.87 (1H, m), 3.80 (1H, m), 7.22 (2H, m), 7.31 (2H, m), 7.74
(1H, m), 7.98 (1H, m), 8.42 (2H, m), 8.86 (2H, m); 13C NMR
(100 MHz, CDCl3): δC 29.6, 34.2, 34.8, 43.0, 121.9, 123.8, 128.3,
128.5, 130.1, 130.6, 130.9, 131.6, 133.1, 135.3, 135.6, 145.8, and
159.7; HRMS (positive ion electrospray, MH+) calculated for
C31H23ClNO4, calculated mass 508.1316, found 508.1313.
Methyl 2-{3-[4-(4-Chlorophenyl)cyclohexyl]-2-
oxopropanoyl}benzoate (29). A suspension of 26 (250 g,
681 mmol) and dimethylaminopyridine (8.33 g, 68.1 mmol) in
toluene (2.5 L) and methanol (250 mL) was heated to 67 °C and
stirred at this temperature for 4 h and cooled to ambient,
and water (1.25 L) was added followed by hydrochloric acid
(11.4 mL, 136 mmol). The mixture was stirred for 5 min, the
aqueous phase was removed, and the organic phase was washed
with water (750 mL). The aqueous phase was removed, and the
solvent was removed by distillation until an internal volume
of 750 mL was reached. The solution was cooled to 60 °C,
methanol (2 L) was added, and the solution was concentrated by
distillation until an internal volume of 750 mL was reached. The
solution was again cooled to 60 °C and diluted with methanol
(1.25 L), and this too was concentrated by distillation until an
internal volume of 750 mL was reached. The solution was cooled
to 64 °C at which point methanol (2 L) was added, the solution
was cooled to 48 °C, a seed of 29 (100 mg) was added and the
suspension stirred for 15 min. The stirred suspension was cooled
to 20 °C and aged for 30 min. Filtration and washing with
methanol (2 × 325 mL) gave the title compound as a light-
2-[4-(4-Chlorophenyl)cyclohexyl]-3-hydroxy-1,4-naphtha-
lenedione (1). A 30 wt % solution of sodium methoxide in
methanol (37.0 kg, 205 mol) was added to a stirred suspension of
26 (60.45 kg, 165 mol) in methanol (233 kg) at 20 °C. The solids
rapidly dissolved, and the resulting dark-red solution was stirred
at 20 °C for 18 h or until conversion to 1 was complete.28
The solution was added via an in-line filter to a stirred solution of
acetic acid (94.0 kg, 1.57 kmol) and methanol (143 kg) in water
(12.0 kg) at ambient using a line rinse of methanol (50 kg) with
the precipitation of a bright-yellow solid. The solid was collected
in a filter dryer, washed with a mixture of methanol and water
(2 × 108 kg, 1:1 v/v) and methanol (96 kg), and then dried
at 40 °C to give the title compound as a bright-yellow solid
1
(51.85 kg, 85.8% th yield, 98.9% a/a HPLC purity): H NMR
1
(400 MHz, CDCl3): δH 1.48−1.64 (2H, m), 1.71−1.78 (2H, m),
1.92−2.02 (2H, m), 2.13−2.25 (2H, m), 2.64 (1H, m), 3.17 (1H,
m), 7.18 (2H, d), 7.27 (2H, d), 7.48 (1H, s, OH), 7.68 (1H, m),
7.76 (1H, m), 8.08 (1H, d) and 8.14 (1H, d); 13C NMR (100
MHz, CDCl3): δC 29.2, 34.4, 34.5, 43.2, 126.0, 127.0, 127.3,
128.2, 128.4, 129.2, 131.5, 132.8, 133.2, 135.0, 146.0, 153.0,
181.8, 184.5; HRMS (APCI, negative ion, [M − H]−) calculated
for C22H18ClO3 365.0950, found 365.0951.
7-[4-(4-Chlorophenyl)cyclohexyl]-5H,9H-bis[2]-
benzopyrano[3,4-e:4′,3′-b]pyridine-5,9-dione (27). A suspen-
sion of 5 (2.60 g, 16.03 mmol), 26 (5 g, 13.63 mmol), and
ammonium acetate (1.20 g, 15.57 mmol) in toluene (100 mL)
and acetic acid (10 mL) was heated to 100 °C (hot bath at
110 °C) and stirred for 24 h. All solids dissolved during the
heating-up period, and some precipitation was seen during the
reaction. The reaction mixture was allowed to cool to ambient
with the precipitation of a grey solid. The mixture was aged for
1 h and filtered, and the residue was washed with toluene (2 ×
10 mL) and then pulled dry to give the title compound as a grey
solid containing a quantity of 5H,9H-bis[2]benzopyrano[3,4-
e:4′,3′-b]pyridine-5,9-dione (28) (approximately 25%) (wt = 0.92 g).
The filtrate was evaporated under vacuum and the residue
triturated with acetone to afford a second crop of product as an
orange solid (wt = 0.23 g).
yellow solid (209.9 g, 77%): H NMR (400 MHz, CDCl3): δH
1.16−1.28 (2H, m), 1.42−1.55 (2H, m), 1.70−2.05 (5H, m),
2.44−2.54 (1H, m), 2.95 (2H, d), 3.87 (3H, s), 7.12 (2H, d),
7.24 (2H, d), 7.48 (1H, d), 7.58 (1H, dd), 7.67 (1H, dd), 7.99
(1H, d); 13C NMR (100 MHz, CDCl3): δC 32.7, 33.2, 34.0,
43.5, 43.6, 52.8, 128.2, 128.4, 129.0, 129.3, 129.4, 131.0, 131.4,
133.2, 138.9, 145.9, 167.1, 194.2 and 198.4; HRMS (APCI,
negative ion, [M + e]−) calculated for C23H23ClO4 398.1290,
found 398.1281.
2-[4-(4-Chlorophenyl)cyclohexyl]-3-hydroxy-1,4-naphtha-
lenedione (1). The ester 29 (3.0 g, 7.54 mmol) was suspended in
methanol (18 mL), stirred, and treated with a 25% solution of
sodium methoxide in methanol (2.02 g, 9.35 mmol), whereby
solids began to dissolve, and a red solution was formed.29 The
solution was stirred at room temperature for 23 h and quenched
by the dropwise addition of 5 M phosphoric acid (1.8 mL,
9.0 mmol), and the resulting yellow slurry was stirred at room
temperature for 24 h. The yellow suspension was filtered, and the
residue was washed with methanol (4 mL + 5 mL) and hot water
(2 × 9 mL) and then dried to give the title compound as a bright-
yellow solid (2.30 g, 83%). This material was spectroscopically
identical to that reported above.
3-(2-((1r,4r)-4-(4-Chlorophenyl)cyclohexyl)acetyl)-3-hy-
droxyisobenzofuran-1(3H)-one (30). Sodium hydroxide sol-
ution (2 M, 10 mL, 20.00 mmol) was added to a suspension of
26 (5 g, 13.63 mmol) in methanol (30 mL), and the mixture was
stirred at ambient for ∼18 h. The mixture was quenched by the
Both solid residues were combined, suspended in acetonitrile
(20 mL), and heated to reflux. The suspension was stirred at
reflux for 30 min and cooled to 40 °C, and the solid was collected
1615
dx.doi.org/10.1021/op300165q | Org. Process Res. Dev. 2012, 16, 1607−1617

Organic Process Research & Development
Article
addition of aqueous hydrochloric acid (25%, 3.2 mL), leading to
the precipitation of a solid and a pH of <2. The slurry was stirred
at ambient for 30 min and filtered, and the residue was washed
with water−methanol (2:1, 2 × 10 mL) and then dried to give the
Generation, Control and Engineering Team, and the Cork team
for scale-up support.
REFERENCES
1
■
title compound as an off-white solid (4.52 g, 86%): H NMR
(1) Latter, V. S.; Gutteridge, W. E. U.S. Patent 4,981,874, 1991.
(2) Hudson, A. T.; Randall, A. W. U.S. Patent 5,053,432, 1991.
(3) (a) Williams, D. R.; Clark, M. P. Tetrahedron Lett. 1998, 39, 7629−
7632. (b) Zhu, F.; Qiao, H.; Bekhazi, M. PCT WO/2010/001379, 2010.
(4) Barcia, J. C.; Cruces, J.; Estevez, J. C.; Estevez, R. J.; Castedo, L.
Tetrahedron Lett. 2002, 43, 5141−5141.
(400 MHz, CDCl3): δH 0.86 (1H, m), 1.02 (1H, m), 1.42 (2H,
m), 1.5−2.0 (8H, m), 2.36 (2H, m), 5.65 (1H, s), 7.08 (2H, m),
7.23 (2H, m), 7.45 (1H, d), 7.71 (1H, m), 7.77 (1H, m), 8.0
(1H, d); 13C NMR (100 MHz, CDCl3): δC 32.8, 33.2, 33.7, 42.1,
43.3, 123.0, 126.2, 127.4, 128.1, 128.4, 131.6. 131.8, 135.2, 145.4,
178.0; HRMS (APCI, negative ion, [M − H]−) calculated for
C22H20ClO4 383.1056, found 383.1053.
(5) (a) Buchecker, R.; Schadt, M.; Takatsu, H. Eur. Patent EP 0 475,
254A1, 1991. (b) Tanaka, Y.; Takatsu, H.; Takeuchi, K.; Schadt, M.;
Buchecker, R. Eur. Patent EP 0 452, 274 A2, 1991. (c) Zhao, Y.;
Snieckus, V. Canadian Patent 2686915 A1, 2009.
3-{[4-(4-Chlorophenyl)cyclohexyl]acetyl}-3-(methyloxy)-2-
benzofuran-1(3H)-one (31). The lactone 26 (3.0 g, 8.18 mmol)
and dimethylaminopyridine (0.12 g, 0.98 mmol) were suspended
in methanol (60 mL), and the stirred mixture was heated at reflux
for 25.5 h. The solution was cooled to ambient, the solvent was
removed under vacuum, and the residue was dissolved in tert-
butylmethylether (20 mL) whereby white crystals started to
crystallise. The mixture was allowed to stand at ambient overnight,
and the solid was collected by filtration under vacuum, washed
with tert-butylmethylether (2 × 10 mL), and dried, giving the title
compound as white crystals (0.48 g, 14.7%):30 1H NMR (400
MHz, CDCl3): δH 0.98−1.15 (2H, m), 1.36−1.51 (2H, m), 1.70−
1.98 (5H, m), 2.34−2.46 (1H, m), 2.55 and 2.73 (2H, 2 × dd),
3.24 (3H, s), 7.08 (2H, d), 7.24 (2H, d), 7.58 (1H, d), 7.64 (1H,
dd), 7.75 (1H, dd), 7.94 (1H, d); 13C NMR (100 MHz, CDCl3):
δC 32.7, 32.9, 33.1, 33.8, 33.8, 43.5, 44.6, 52.1, 107.5, 124.2, 125.9,
127.6, 128.1, 128.4, 131.5, 131.6, 134.8, 142.7, 145.7, 167.8, and
201.1; HRMS (APCI, negative ion, [M + e]−) calculated for
C23H23ClO4 398.1290, found 398.1278.
2-[4-(4-Chlorophenyl)cyclohexyl]-3-hydroxy-1,4-naphtha-
lenedione (1). A 25% solution of sodium methoxide in methanol
(0.21 mL, 0.905 mmol) was added to a suspension of the lactone
31 (0.30 g, 0.754 mmol) in methanol (2 mL) at room
temperature. The solids gradually dissolved, giving a red solution
which was stirred at room temperature for 18 h31 and quenched
by the dropwise addition of 5 M phosphoric acid (0.2 mL,
1.0 mmol), and the resulting yellow slurry was stirred at room
temperature for 24 h. The yellow suspension was filtered, and the
residue was washed with methanol (2 × 1 mL) and hot water
(2 × 1 mL) and then dried to give the title compound as a bright-
yellow solid (0.21 g, 76%). This material was spectroscopically
identical to that reported above.
(6) Kus, N. S.; Kazaz, C. Asian J. Chem. 2008, 20, 1226−1230.
(7) Duggleby, P. M.; Holt, G. J. Chem. Soc. 1962, 3579−3582.
(8) Yang, R.-Y.; Kizer, D.; Wu, H.; Volckova, E.; Miao, X.-S.; Ali, S. M.;
Tandon, M.; Savage, R. E.; Chan, T. C. K.; Ashwell, M. A. Bioorg. Med.
Chem. 2008, 16, 5635−5643.
(9) Knott, E. B. J. Chem. Soc. 1963, 402−410.
(10) Gabriel, S. Chem. Ber. 1884, 17, 2521−2527.
(11) Gabriel, S. Chem. Ber. 1907, 40, 71−83.
(12) Clean samples of the impurities 16 and 17 were prepared
unambiguously by treatment of the isochromandione 5 with potassium
hydrogen carbonate in THF or isobutylamine in hot acetic acid,
respectively.
(13) Investigational work showed that polybrominated products 14
and 15 do not convert to the desired isochromandione 5 but rather
hydrolyse to form phthalide-3-carboxylic acid.
(14) (a) Yale, H. L. J. Am. Chem. Soc. 1947, 69, 1547−1548. (b) Zhai,
X.; Li, J.; He, L.; Zheng, S.; Zhang, Y. B.; Gong, P. Chin. Chem. Lett.
2008, 19, 29−32. (c) Li, J.; Zhao, Y.-F.; Yuan, X.-Y.; Xu, J.-X.; Gong, P.
Molecules 2006, 11, 574−582.
(15) Rangnekar, D. W.; Telange, V. R. Ultra Sci. 2000, 12, 211−215.
(16) Newman, M. S. J. Chem. Educ. 1977, 54, 191.
(17) Iwasaki, H. Takahashi, H. U.S. Patent 4,298,530 1981.
(18) (a) Cai, D.; Wu, X. Faming Zhuanli Shenqing 2011, 101973872.
(b) Sathe, D. G.; Mantripagada, N. R.; Patel, G. R.; Sawant, D. K.; Naik,
A. T.; Thoovara, S. M. Indian Pat. Appl. 2007MU01397, 2009.
(19) Kumai, Y. Jpn. Kokai Tokkyo Koho JP 02019344, 1990.
(20) Parikh, J. R.; Doering, W. E. J. Am. Chem. Soc. 1967, 89, 5505−
5507.
(21) Rosenmund, K. W. Chem. Ber. 1918, 51, 585. Rylander, P. N.
Catalytic Hydrogenation over Platinum Metals; Academic Press: New
York, 1967; pp 478−485.
(22) Use of thionyl chloride gave a much slower and dirtier reaction
profile, as well as having a detrimental effect on the subsequent
hydrogenation reaction.
ASSOCIATED CONTENT
(23) Use of tertiary alkylamine bases for this transformation generally
led to significant over-reduction whereby the aromatic chloride function
was also reduced.
■
S
* Supporting Information
1H and 13C NMR spectra. This material is available free of charge
via the Internet at http://pubs.acs.org.
(24) Green metrics were calculated for both routes including solvent
recovery. The new route metrics included the synthesis of 5 from
phthalic anhydride, whereas the old route was calculated with
naphthoquinone 34 as starting material. It is acknowledged that 2 is
derived from naphthalene via oxidation which is in turn derived from
coal tar, so the comparison is deemed appropriate.
AUTHOR INFORMATION
■
Corresponding Author
*Michael.W.Urquhart@gsk.com (M.W.J.U.); Stuart.G.Leach@
gsk.com (S.G.L.).
(25) Levels of the cis product were also evident at up to 25% from 1H
NMR inspection.
Notes
(26) Product typically contains unreacted alcohol 26 at 10%.
(27) Percentage yield range observed: 78−89% (from pilot-plant
batches), average yield 83%. The purity quoted includes 1.1% a/a of the
cis-1,4-cyclohexane isomer and 1.0% E-alkene, both of which
interconvert to atovaquone 1 in the following chemical stage.
(28) Note: LC check after 1 h of reaction shows a ratio of 29 to 31 of
approx 1:1. These products diminish on continued reaction, confirming
their intermediacy within the rearrangement reaction.
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
We thank the whole atovaquone team, in particular, Clare
McKinlay for analytical support, Dave Grimshaw for Process
Safety, Felicity Millns for Technical Sourcing, the Particle
1616
dx.doi.org/10.1021/op300165q | Org. Process Res. Dev. 2012, 16, 1607−1617

Organic Process Research & Development
Article
(29) LC check after 30 min of reaction shows a ratio of 29 to 31 of
approx 1:1 which confirms the equilibrating nature of this trans-
formation.
(30) Please note there is appreciable product 31 (approximately 32%)
remaining within the filter liquors with the remaining material being
methyl 2-{3-[4-(4-chlorophenyl)cyclohexyl]-2-oxopropanoyl}benzoate
(30).
(31) LC check after 1 h of reaction shows a ratio of 31 to 29 of approx
3:2 which confirms the equilibrating nature of this transformation.
1617
dx.doi.org/10.1021/op300165q | Org. Process Res. Dev. 2012, 16, 1607−1617