Structural similarity and its surprises: Endothelin receptor antagonists -process research and development report

Article abstract of DOI:10.1021/op000040n

Process research and pilot plant processes are described for three endothelin (ET) receptor antagonists. The efficient synthesis of the parent compound Darusentan proceeds via a Darzens reaction from chloroacetate with benzophenone, addition of methanol to the resulting epoxide, saponification of the alkyl propionate and optical resolution of the racemic acid by crystallisation with a chiral amine. The final stage of the synthetic sequence involves the introduction of a pyrimidine moiety. Intermediates formed during this process can be used as starting materials for the synthesis of the two other ET receptor antagonists BSF 420627 and BSF 302146. An ether exchange reaction, which replaces the methoxy with a phenethyloxy substituent, enabled BSF 420627 to be prepared. The synthetic route to BSF 302146 employs trimethylaluminum to methylate the epoxide produced by the Darzens reaction.

Full text of DOI:10.1021/op000040n

Organic Process Research & Development 2001, 5, 16−22
Structural Similarity and Its Surprises: Endothelin Receptor Antagonists -
Process Research and Development Report
R. Jansen,* M. Knopp,* W. Amberg, H. Bernard, S. Koser, S. Mu¨ller, I. Mu¨nster, T. Pfeiffer, and H. Riechers
Hauptlaboratorium, BASF AG, 67056 Ludwigshafen, Germany
Scheme 1: Does structural similarity imply the same
synthetic route?
Abstract:
Process research and pilot plant processes are described for
three endothelin (ET) receptor antagonists. The efficient
synthesis of the parent compound Darusentan proceeds via a
Darzens reaction from chloroacetate with benzophenone, ad-
dition of methanol to the resulting epoxide, saponification of
the alkyl propionate and optical resolution of the racemic acid
by crystallisation with a chiral amine. The final stage of the
synthetic sequence involves the introduction of a pyrimidine
moiety. Intermediates formed during this process can be used
as starting materials for the synthesis of the two other ET
receptor antagonists BSF 420627 and BSF 302146. An ether
exchange reaction, which replaces the methoxy with a phen-
ethyloxy substituent, enabled BSF 420627 to be prepared. The
synthetic route to BSF 302146 employs trimethylaluminum to
methylate the epoxide produced by the Darzens reaction.
Results and Discussions
Process Research and Pilot Plant Synthesis of Darusen-
tan. The retrosynthetic strategy developed in our medicinal
chemistry group for alkoxy-type ET receptor antagonists such
as Darusentan or BSF 420627 featured the glycidate ester 3
as the central intermediate. The ester is then transformed to
the desired drugs via the 2-hydroxy methyl 3-methoxy-3,3-
diphenylpropionate 4. Initially, the methyl methoxy-3,3-
diphenylpropionate 4 was directly treated with 4,6-dimeth-
yloxy-2-methylsulfonylpyrimidine 8. The subsequent hy-
drolysis of the LU 135252-methyl ester led to the racemic
drug which was then optically resolved by coupling with
quinine. This strategy had several disadvantages. The hy-
drolysis of the LU 135252-methyl ester required very harsh
conditions using KOH in dioxane under reflux. Under these
conditions two side products were obtained: one from the
partial elimination of methanol, the other from the partial
cleavage of the pyrimidyl ether. Furthermore, half of the
racemic drug rac-LU 135252 produced initially would be
wasted if optical resolution was carried out at the final step.
For these reasons, the synthetic strategy had to be revised
before the process could be used in pilot plant scale
production. An important objective of the revised strategy
was to introduce racemic separation at an earlier stage of
the sequence. The hydroxy acid 5 was considered to be a
suitable intermediate. By performing optical resolution at the
hydroxy acid 5 stage, the amount of heterocyclic derivative
8 needed would also be reduced. Furthermore, the acid (S)-5
or the acid salt (S,S)-7 would undergo the final nucleophilic
substitution rather than the corresponding ester 4. The revised
pathway used to synthesize Darusentan is outlined in Scheme
2.
Introduction
Endothelins (ET) are a family of 21 amino acid peptides
which act as modulators of vascular tone, cell proliferation
and hormone production. Their actions are mediated by two
G-protein coupled receptors whose absolute and relative
amounts vary depending on the tissue. The controlled
interference of the endothelin receptor interaction appears
to be a promising approach in a number of different
therapeutic areas, and our medicinal chemistry group had
identified three compounds with different binding profiles.1
The chemical development task involved establishing
efficient and robust routes with which the target compounds
could be prepared in amounts suitable for clinical develop-
ment. At first sight, the close structural similarity (Scheme
1) between Darusentan and BSF 420627 suggested that
essentially the same synthetic route could be used, in which
addition of an alcohol to an epoxide would result in the core
structure of the 3-alkoxy-2-pyrimidinyloxy-propionic acid
derivatives. It was unclear whether the structural similarity
between Darusentan and BSF 302146 could be synthetically
exploited. Early attempts to add a methyl group to the
epoxide had failed and the medicinal chemistry group
developed an entirely different approach to synthesizing the
target structure.
* To whom correspondence should be addressed. Telephone: +49 621 60
52953, +49 621 60 41433. Fax: +49 621 60 46923. E.mail: Rolf.Jansen@
basf-ag.de, Monika.Knopp@basf-ag.de.
(1) Riechers. H.; Albrecht, H. P.; Amberg, W.; Baumann. E.; Bernard, H,;
Bo¨hm, H. J.; Klinge, D.; Kling, A.; Mu¨ller, S.; Raschack, M.; Unger, L.;
Walker, N.; Wernet, W. J. Med. Chem. 1996, 39, 2123-2128; Amberg, W.;
Hergenro¨der, S.; Hillen, H.; Kettschau, G.; Kling, A.; Klinge, D.; Raschack,
M.; Riechers, H.; Unger, L. J. Med. Chem. 1999, 42, 3026-3032.
16
•
Vol. 5, No. 1, 2001 / Organic Process Research & Development
10.1021/op000040n CCC: $20.00 © 2001 American Chemical Society and The Royal Society of Chemistry
Published on Web 11/05/2000

Scheme 2: Pilot plant synthesis of Darusentan
The methyl glycidate intermediate 3 was obtained in a
Darzens reaction by treating benzophenone 1 with methyl
chloroacetate 2 in the presence of NaOMe. The epoxide 3
was then opened via an acid-catalyzed addition of methanol
to yield the R-hydroxy ester 4. Hydrolysis of the ester 4 led
to the corresponding R-hydroxy acid 5. The R-hydroxy acid
5 could be produced in a one pot procedure without isolation
of any intermediate. This sequence was performed at the 500
mol scale to produce 95 kg in one batch with a chemical
yield of 80% for the three steps and a purity of >99.5%
(HPLC). Unfortunately, the first pilot plant scale synthesis
was accompanied by an unwelcome surprise.
In contrast to the laboratory experiments and the calori-
metric experiments, the reaction was initiated very violently
after half of the methyl chloroacetate had been added, with
the result that most of the MTBE evaporated within seconds.
As the reaction produced methanol which helped to dissolve
the methylate, we observed a self-accelerated reaction.
Addition of more MTBE allowed the reaction to go to
completion. This sudden violent reaction was probably
caused by an accumulation of methyl chloroacetate because
the rate at which the reactants were consumed was too low.
The delayed start of the reaction was not due to the low
temperature of -10 °C because laboratory experiments had
already demonstrated that the reaction started directly after
the addition of methyl chloroacetate at -20 °C. Stirring
appeared to be a more likely cause of the problem, especially
since a heterogeneous reaction was involved (suspension of
NaOMe). The effect of low stirring energy had then been
tested in laboratory experiments. It was observed that the
suspension became very viscous after half the methyl
chloroacetate had been added. The anchor stirrer in the lab
reactor was able to stir this suspension, but the impeller
agitator in the pilot plant vessel was not. Tests of a number
of other solvents demonstrated that the problem could be
solved by replacing MTBE by tetrahydrofuran. The increased
solubility of the reactants in tetrahydrofuran meant that the
reaction mixture did not become viscous and the reaction
could therefore start immediately after addition of the methyl
chloroacetate without prior build-up of the reactants.
Once the conditions for the production of the racemic
hydroxy acid had been established, attention turned to the
question of optical resolution. Several chiral amines were
tested. The best results were obtained using L-methyl
prolinate (c.y. 43%, ee > 99%(HPLC)) and (S)-1-(4-
nitrophenyl)ethylamine (c.y. 38%, ee > 99%(HPLC)). The
use of methyl prolinate required an equimolar amount of the
auxiliary, whereas in the case of the (S)-1-(4-nitrophenyl)-
ethylamine only half an equivalent was required. The
disadvantage of both these auxiliaries is that they are
extremely expensive. Additionally, methyl prolinate could
not be completely recycled since free methyl prolinate tends
to form diketopiperazines. Nevertheless, both processes were
subjected to pilot plant scale testing. In both cases compa-
rable yields were obtained. Optical resolution to give the
(S,S)-hydroxy acid salt and isolation of the enantiopure acid
(S)-5 was achieved in 70% with respect to the racemate 5 in
both cases. Although the yields obtained with both chiral
amines were comparable, the more robust process was the
(S)-1-(4-nitrophenyl)ethylamine process. Considering the
high costs of these two auxiliaries (S)-1-(4-chlorophenyl)-
ethylamine 6 proved to be an attractive alternative as it is
the product of another BASF process. Furthermore, by using
(S)-1-(4-chlorophenyl)ethylamine 6, the racemic hydroxyacid
5 did not need to be isolated, and optical resolution could
be performed directly in the MTBE solution. Although it
was stated above that the final stage involved nucleophilic
substitution on the free hydroxy acid (S)-5, it turned out that
the hydroxy acid salt (S,S)-7 could also be subjected to
nucleophilic substitution without loss of yield. Treatment of
the hydroxy acid salt (S,S)-7 with 4,6-dimethyloxy-2-
methylsulfonylprimidine 8 in the presence of LiNH2 in DMF
produces Darusentan with an 85% yield. This very easy five-
step process has been used to provide several hundred
kilograms of Darusentan to date with an overall yield of 31%.
Only two intermediates have to be isolated during the whole
sequence.
We supposed that the Darusentan process described above
could be easily modified to access other structurally analo-
gous ET receptor antagonists such as BSF 420627.
Vol. 5, No. 1, 2001 / Organic Process Research & Development
•
17

Scheme 3: Approaches unsuitable for the pilot plant
production of BSF 420627
Scheme 4: Transetherification: the key to efficient
production
the ester 4 applying the same conditions as when the
methanol should be added to the epoxide, but attempt to
move the equilibrium toward the epoxide by distilling the
methanol out of the solution. If this process is performed in
the presence of homoveratryl alcohol, the desired compound
should be created. The chiral acid (S)-5 was available from
the production of Darusentan. We knew from our attempts
at crystallising glycidic acid 11 that the free acid 7 would
not survive the acid-catalyzed transetherification reaction.
We therefore had to introduce the additional step of esteri-
fying (S)-5 to generate the starting material (S)-4 for the key
reaction (Scheme 4).
Experiments with toluene added to remove the methanol
showed that the ester 15 was indeed created during the
reaction, but that the conditions were too harsh and decom-
position occurred before the reaction was complete. Since
we had no experience with the reaction but suspected the
high temperature as the reason for decomposition we replaced
toluene by chloroform and later dichloromethane to get quick
information whether the reaction could work at all. Using
dichloromethane we isolated the desired product in satisfac-
tory yield. Later detailed studies revealed that really the
critical reaction parameter was temperature. If the temper-
ature was too low, no reaction took place. If the temperature
was too high, decomposition occurred. Scaling up the
reaction using dichloromethane as a solvent presented new
problems since the reaction slowed and instead of a reaction
time of 8 h we observed reaction times of more than 36 h.
Longer reaction times also meant a higher degree of
decomposition, that is, elimination products, and so the
overall space-time-yield dropped. Since our experience
indicated that temperature was the most critical factor, we
decided to fine-tune the reaction temperature by using toluene
as solvent under low-pressure conditions. This allowed us
to set the temperature to any value between room temperature
and 111 °C. The optimum temperature for the reaction was
found between 50 and 55 °C. In laboratory scale experiments,
we achieved yields of more than 90%. In the pilot plant the
reaction time was reduced to 6 h and the yield was about
75% (Scheme 5).
Process Research and Pilot Plant Synthesis of BSF
420627. Our first attempt at synthesizing BSF 420627
involved repeating the Darusentan sequence but adding
homoveratryl alcohol 14 instead of methanol to the glycidic
ester 3. Although the first steps proceeded smoothly, dif-
ficulties arose when we tried to repeat the racemic resolution
by the same means as used in the Darusentan route. Adding
chiral amines such as (S)-p-chlorophenylethylamine, (S)-p-
nitrophenylethylamine, quinines, quinidines, or amino acid
derivatives and so forth to solutions of acid 13 did not lead
to satisfactory crystallisation and consequently did not yield
any chiral resolution. We presumed that the long lipophilic
side chain was responsible for this surprising reluctance to
crystallise. Attempts to overcome the problem by crystallising
an earlier intermediate like 11, which did not yet have the
lipophilic side chain, did not lead to a synthetic route suitable
for pilot plant production. Whilst the preparation of gram
amounts of BSF 420627 had been achieved, the instability
of the glycidic acid, which decomposed liberating CO2,
prevented effective scale-up. Experiments based on the
literature recommendation of performing the Darzens reac-
tion2 in a diasteroselective manner were not promising.
Although diastereoisomer ratios of 4:1 were achieved using
2-phenylcyclohexenyl alcohol as the chiral auxiliary, the
separation of the optically active forms proved to be very
tedious. An enantioselective approach to epoxidize 3-phenyl
cinnamic esters 10 via Jacobsen’s methodology3 yielded the
glycidic ester with an enantioselectivity of 75% ee, but again
we faced problems in purifying the product. A further well-
established means of generating enantiopure acids is that of
enzymatic resolution. However, in our experiments we did
not observe any conversion; esterases accepted neither the
3,3-diphenyl glycidate derivatives nor the 2-hydroxyesters
as substrates (Scheme 3).
A solution to the problem was found based on the
following considerations. The addition of the methanol to
the epoxide 3 should be a reversible process. If it is
reversible, one could produce a chiral glycidic ester by taking
Using the transetherification reaction, we were thus able
to base our chiral acid (S)-16 on a chiral intermediate from
the synthesis of Darusentan. Nevertheless, to achieve this
objective, a three step detour was necessary. Whether the
mechanism of the transetherification reaction follows the
epoxide path as our initial considerations indicate or whether
it is a carbocation mechanism has still to be investigated.
(2) Ohkata, K.; Kimura, J.; Shinohara, Y.; Takagi, R.; Hiraga, Y. J. Chem.
Soc., Chem. Commun. 1996, 21, 2411-2412.
(3) Deng, L.; Jacobsen, E. N. J. Org. Chem. 1992, 57, 4320-4323.
18
•
Vol. 5, No. 1, 2001 / Organic Process Research & Development

Scheme 5
Scheme 6: Original synthesis of BSF 302146
in toluene in the presence of MsOH led to the cyanohydrine
19, which then was converted to the R-hydroxy acid 20
without further purification using a mixture of sulfuric acid,
acetic acid and water. As in the production of Darusentan,
the racemic acid was optically resolved using (S)-1-(4-
chlorophenyl)ethylamine (S)-6. Nucleophilic substitution of
the 4,6-dimethyl-2-methylsulfonyl-1,3-pyrimidine 16 in the
presence of LiNH2 in DMF completed the sequence to yield
the desired drug. This procedure was used in the first pilot
plant process and rapidly produced 14 kg of BSF 302146
(14% overall yield).
Unfortunately, this strategy had a number of general
disadvantages. Besides the requirement of handling HCN on
a large scale, the hydrolysis of the cyanohydrine 19 was of
poor yield (50-60%). Due to the impurities of the hydroxy
acid 20, the yield of the optical resolution step was also low.
The main disadvantage, however, was the extremely high
cost of the starting material diphenyl propionitrile. In an effort
to avoid the extremely expensive diphenylpropionitrile and
to exploit the synthetic routes to other ET receptor antago-
nists, the possibility of using the glycidic ester 3 as the
starting material was investigated. One idea was the intro-
duction of the methyl group using an organometallic species,
for example, an alkyl aluminum reagent, and trimethyl
aluminum (TMA) turned out to be the reagent of choice.4
TMA can be purchased in bulk as a 2 M solution and can
be handled without severe safety problems.
The formation of the R-hydroxy ester 22 represented the
key step of our new strategy for BSF 302146. Depending
on the reaction parameters, either the methyl 2-hydroxy-
butyrate 22 was obtained on its own or as a mixture with
the methyl 3-hydroxy-2-methylpropionate 23 (Scheme 7).
The yields and the regioselectivity of the reaction were
both strongly dependent on the reaction conditions. To
determine the best conditions, solvent polarity, temperature,
stoichiometry, and concentration were varied. A typical
Saponification of the ester 15 with sodium hydroxide
proceeded as in the Darusentan synthesis as did the final
transformation to introduce the pyrimidine moiety. Problems
arose again during product isolation since the free acid of
BSF 420627, which was the original target molecule, was
amorphous. To supply the product at the required level of
purity meant loss of yield. A further problem was that of
the racemization of the free acid of BSF 420627 in solvents
such as dichloromethane. For these reasons, we decided to
prepare a number of different salts. Fortunately, simply
changing from the free acid to the sodium salt solved the
above problem. The salt BSF 420627 was present as a fine
crystalline substance of reproducible quality and with proper-
ties well suited to pharmaceutical development.
Process Research and Pilot Plant Synthesis of BSF
302146. The original route to the ET receptor antagonist BSF
302146 developed by our medicinal chemistry group was
organized completely different (Scheme 6). Starting from
diphenyl propionitrile 17, the first step was the reduction of
the nitrile to the corresponding aldehyde 18 using DIBAH
in toluene. Treatment of the crude aldehyde 18 with NaCN
(4) Fukumasa, M.; Furuhashi, K.; Umezawa, J.; Takahashi, O.; Hirai, T.
Tetrahedron Lett. 1991, 32, 1059; Kuran, W.; Pasynkiewicz, S.; Serzyko,
J. J. Organomet. Chem. 1974, 73, 187; Pfaltz, A.; Mattenberger, A. Angew.
Chem., Int. Ed. Engl. 1992, 21, 71; Poon, T.; Mundy, B. P.; Favaloro, F.
G.; Goudreau, C. A.; Greenberg, A.; Sullivan, R. Synthesis 1998, 832;
Ishibashi, N.; Miyazawa, M.; Miyashita, M. Tetrahedron Lett. 1998, 39,
3775; Miyashita, M.; Hoshino, M.; Yoshikoshi, A. J. Org. Chem. 1991,
56, 6483.
Vol. 5, No. 1, 2001 / Organic Process Research & Development
•
19

Scheme 7: Addition of trimethyl aluminum to the glycidate
Scheme 8: Pilot plant process of BSF 302146
3.
Table 1. Addition of methyl to the epoxide
conc
(%)
temp
(°C)
time AlMe3 yield
regio
solvent
(h) (equiv) (%)a selectivityb
8.5 toluene
8.5 toluene
8.5 toluene
8.4 toluene
6.25 cyclohexane
0
10
0
0.5
0.5
0.5
2.0
1.5
1.5
1.5
1.5
1.3
1.3
1.3
1.3
75
68
91
75
90
83
84
86
83
86:14
78:22
90:10
91:9
97:3
97:3
94:6
90:10
93:7
-5 to -10 1.5
10-15
-5 to -10
-5 to -10
10-15
0.5
1
1
1
1
10
heptane
20
20
15
heptane
cyclohexane
heptane/toluene 31:2 -5 to -10
The glycidate 3 in toluene was added to a mixture of
heptane and a 2 M solution of TMA in toluene at -10 °C.
The glycidate (63.5 kg) was added as a 75% solution in
toluene. The regioisomeric hydroxyesters 22 and 23 were
obtained with a regioisomeric ratio of 98:2. After crystalli-
sation the R-hydroxy ester 22 was obtained as colorless
crystals in 84% yield with a purity of >99%. Altogether,
210 kg of the R-hydroxy ester was produced using this
reordered procedure (Scheme 8).
Hydrolysis of the R-hydroxy ester 22 by heating the ester
in the presence of KOH in 2-propanol produced the R-hy-
droxy acid 20 in almost quantitative yield. The R-hydroxy
acid 20 is an intermediate in the old reaction route and its
optical resolution was performed in the same way as before.
Due to the higher purity of the R-hydroxy acid, racemic
separation worked slightly better, the yield rising by 4 to
36%. The yield could also be improved by exchanging the
solvent. Carrying out the reaction in a DMF/toluene mixture
instead of pure DMF enabled the reaction time to be reduced
and the yield to be improved by 10 to 84%. The (S,S)-
hydroxy acid salt (S,S)-21 was dissolved in DMF/toluene
and treated with 3 equiv of LiNH2 and 1.6 equiv of 4,6-
dimethyl-2-methylsulfonylpyrimidine 16 at 45 °C for 12 h.
After crystallisation from 2-propanol/water, the drug was
isolated in high chemical (99.8%) and optical yield (99.9%,
HPLC). The 2-propanol/water mixture was chosen for
crystallisation as no racemization occurred in the presence
of protic solvents no. This new strategy was used to prepare
74 kg of BSF 302146 with an overall yield of 24%.
a Yield of the R-hydroxy ester. bMeasured by HPLC.
procedure was the addition of trimethylaluminum to a
solution of the glycidate 3 in toluene, heptane, cyclohexane,
or a mixture of these. The results are summarized in Table
1. It is readily seen that unpolar solvents favor the formation
of the 3-methyl-2-hydroxy ester. This can be explained by
the better stabilization of the tertiary carbocation intermediate
(C3 carbocation) compared to the secondary carbocation
intermediate (C2 carbocation), which is formed after addition
of trimethylaluminum to the epoxide ogygen followed by
the cleavage of the C-O bond. Furthermore, regioselectivity
was also enhanced by low temperatures and low concentra-
tions. An excess of 1.3-1.5 equiv of the organometallic
reagent turned out to give the best regioselectivity.
The separation of the regioisomers was easily achieved
by crystallisation. Mixtures of toluene/heptane as well as
2-propanol/heptane were used. The best results were obtained
using a ratio of crude product: 2-propanol: heptane of 1:1:2.
Under these conditions, we observed a chemical yield of 87%
and a purity of 99.2% (HPLC) of the desired regioisomer in
our laboratory batches. Once again, initial attempts to transfer
these results to pilot plant scale production gave surprisingly
disappointing results. For the first batch, TMA solution was
added to a mixture of heptane and the glycidate in toluene
at -10-0 °C. However, screening of the parameters had
indicated that better regioselectivity could be achieved by
decreasing the polarity of the solvent. Toluene was added
to avoid crystallisation of the starting material. The reaction
was done on a 250 mol scale in a 630 L reactor. Due to
poor vessel cooling, the reaction time was 2 days instead of
the theoretically determined 6 h. Regioselectivity suffered
as a result, decreasing to a level of 2:1. The solution to the
problem lay in the order in which reactants were added. If
we added the solution of the glycidate to the TMA solution
instead of the other way round, we were able to reproduce
the lab results or even improve on them in large scale
production.
Summary
In this report, we have demonstrated the surprises that
can lie in store if one assumes that structural similarity
implies similarity of synthetic routes. Previously BSF 302146
had been synthesized by a route very different to that for
Darusentan. However, the synthetic route finally adopted
turned out to be very similar. In contrast, the synthesis of
BSF 420627, a compound with an obvious resemblance to
Darusentan, had to incorporate a three-step detour due simply
to its crystallisation properties. Efficient syntheses have,
however, now been developed for our three target com-
20
•
Vol. 5, No. 1, 2001 / Organic Process Research & Development

pounds, based on intermediates generated in the synthesis
of our most advanced product Darusentan.
product was separated, washed with 600 L of n-heptane and
dried to give 120 kg (85%) of a colorless solid.
BSF 420627. (S)-2-Hydroxy-3-methoxy-3,3-diphenyl-
propionic Acid Methyl Ester (S)-4. (S)-2-hydroxy-3-
methoxy-3,3-diphenylpropionic acid (47.6 kg) and potassium
carbonate (14.5 kg) were suspended in 250 L of dimethyl-
formamide and were stirred for 3 h at 20° - 25 °C. Then
24.5 kg of dimethyl sulfate was added within 1 h while the
reaction mixture was kept at 20°-25 °C. Stirring was
continued for 2 h. Water (750 L) was added, and the product
was extracted once with 300 L of toluene. The organic layer
was washed once with 100 L of water, and the toluenic
solution of the crude methyl ester was transferred in next
stage without further purification.
(S)-2-Hydroxy-3-[2-(3,4-dimethoxyphenyl)ethoxy]-3,3-
diphenylpropionate (S)-15. A solution of 32.4 kg homo-
veratryl alcohol in 120 L of dichloromethane was added to
the toluenic solution of the previous stage corresponding to
48.6 kg of methyl ester and 300 L of toluene. Dichlo-
romethane was distilled, and afterwards 100 L of toluene
was distilled at a maximum temperature of 40 °C under
vacuum (∼100 mbar). After dilution with 250 L of toluene
and addition of 1.7 kg of concentrated sulfuric acid a mixture
of toluene and methanol was distilled at 50°-55 °C under
vacuum (about 100 mbar) for 6 h. The reaction mixture was
cooled to 20°-25 °C. The solution was submitted to the next
stage without further purification.
(S)-2-Hydroxy-3-[2-(3,4-dimethoxyphenyl)ethoxy]-3,3-
diphenylpropionic acid (S)-16. The solution of the previous
stage was charged with 200 L of aqueous sodium hydroxide
(7.5%). Toluene (115 L) was distilled under vacuum (100-
150 mbar) at 50°-55 °C. After cooling to 20 °C separated
water was added again, and 20 L of tetrahydrofuran was
added when a precipitate was observed. After stirring for
24 h at 50°-55 °C the mixture was cooled to 20°-25 °C,
and 140 L of methyl tert-butyl ether was added. Then the
reaction mixture was acidified with hydrochloric acid (65 L
of 20% aqueous HCl), the organic layer was separated and
washed with 100 L of water. Methyl tert-butyl ether (300
L) was added, and the solvent was distilled under vacuum
at max 50 °C until the content of water was below 0.5%.
The solution was cooled to 20°-25 °C. Stirring was
continued until the product crystallised. n-Heptane (150 L)
was added, and the suspension was cooled to 0°-5 °C. The
crystals were separated, washed with a mixture of methyl
tert-butyl ether and n-heptane. The product (56 kg) was
isolated as a colourless solid.
BSF 420627. (S)-2-Hydroxy-3-[2-(3,4-dimethoxyphenyl)-
ethoxy]-3,3-diphenylpropionic acid (37 kg) was suspended
in 160 L of dimethylformamide. Lithium amide (6 kg) was
added at 20°-25 °C within an hour. Dimethylpyrimidine
methyl sulfone (16.6 kg) was added at 20°-25 °C within
an hour and stirred for 16 h at 45-55 °C. Water (480 L)
and methyl tert-butyl ether (140 L) were added, and the
mixture was acidified with 60 L of 20% aqueous hydro-
chloric acid to pH 1-2. The organic layer was separated,
washed with 100 L of water, and was charged to a solution
of 6.4 kg sodium hydroxide in 560 L of isopropyl alcohol.
Experimental Section
NMR spectra were measured on a Bruker DPX 200 (1H,
200 MHz). HPLC analyses were performed using a Merck
Lichrosorpher RP Select-B 5 µm column and a acetonitrile/
water mobile phase. Melting points are meassured on a Bu¨chi
B 540 and are uncorrected.
Darusentan. 3,3-Diphenyl-2,3-epoxy-propionic Acid
Methyl Ester (3). 178 kg benzophenone and 90 kg of sodium
methoxide were suspended in 300 L of tetrahydrofuran.
Chloroacetic acid methyl ester (181 kg) was added at 0 °C
within 4 h. Then 500 L of water was added, and the product
was extracted with 300 L of methyl tert-butyl ether. The
organic layer was washed twice with 5% sodium chloride
solution and concentrated. The crude product was transferred
to the next stage without further purification.
2-Hydroxy-3-methoxy-3,3-diphenylpropionic Acid (5).
The solution from the previous stage was diluted with 800
L of methanol. p-Toluenesulfonic acid monohydrate (1.8 kg)
was added at room temperature. When the exothermic
reaction was finished (∼5 h) the reaction mixture was heated
to reflux. The solvent was distilled from the mixture until
the temperature reached 66 °C. Potassium hydroxide (1120
kg), 10% solution, was added while the reaction mixture was
still refluxing. The organic solvent was distilled from the
mixture until the temperature reached 94 °C. After cooling
to room temperature the mixture was diluted with 400 L of
water. Methyl tert- butyl ether (560 L) was added, and the
reaction mixture was acidified by addition of 880 L of 10%
sulfuric acid. The layers were separated. The organic layer
was transferred in the next stage without further purification.
(S)-p-Chlorophenylethylammonium-(S)-2-hydroxy-3-
methoxy-3,3-diphenylpropionate (7). The solution from the
previous stage was diluted with 390 L of methyl tert-butyl
ether and 665 L of methanol and was heated to reflux. (S)-
1-(4-chlorophenyl)ethylamine (76 kg) was added, and after
cooling to 0-5 °C with a rate of 10 °C/h the precipitated
salt was separated, washed with 400 L of methyl tert-butyl
ether and dried to give 141 kg (33% from benzophenone)
of a colourless solid.
LU 135252 Darusentan. (S)-p-Chlorophenylethylammo-
nium-(S)-2-hydroxy-3-methoxy-3,3-diphenylpropionate (141
kg) was suspended in 360 L of dimethylformamide. After
addition of 23 kg of lithium amide 375 kg of a 25% solution
of 4,6-dimethoxy-2-methyl-sulfonylpyrimidine in dimethyl-
formamide was added within 12 h at 20-30 °C. Stirring
was continued for 4 h at 20-30 °C. After reaction was
finished, the reaction mixture was diluted with 1770 L of
water. Ethyl acetate (800 L) was added, and the mixture was
acidified with 840 L of 10% sulfuric acid. The layers were
separated, and the organic layer was washed with 660 L of
water. The organic layer was dried by azeotropic distillation
(50-60 °C, 400 mbar). The distillation was continued until
a 1 M solution of the product in ethyl acetate was obtained.
n-Heptane (740 L) was added to the hot solution. After
cooling to 0-5 °C with a rate of 10 °C/h the precipitated
Vol. 5, No. 1, 2001 / Organic Process Research & Development
•
21

The solution was stirred until the crystallisation was com-
pleted. The crystals were separated, washed with 50 L of
isopropyl alcohol, and dried to give 36 kg as a colourless
solid.
(S,S)-[1-(4-chlorophenyl)ethylammonium]-2-hydroxy-
3,3-diphenylbutyrate ((S,S)-21). R-Hydroxy acid 20 (465
kg) in 2-propanol (89.8 kg 20) and (S)-1-(4-chlorophenyl)-
ethylamine 6 (33.5 kg) were heated to 80 °C for 1 h. The
solution was cooled to 0 °C with a rate of 5 °C/h. Stirring
was continued for 6 h. The crystalline product was filtered
off, leading to 54.9 kg of a colourless powder (37.6%).
Purity: 99% (HPLC); ee ) 93% (HPLC); mp ) 120-123
°C;1H NMR (200 MHz, d6-DMSO): δ ) 1.3 (d, 3H), 1.7
(s, 3H), 4.2 (q, 1H), 4.4 (s, 1H), 7.0-7.4 (14H) ppm.
(S)-2-(4,6-Dimethyl-1,3-pyrimidyloxy)-3,3-diphenylbu-
tyric Acid BSF 302146. R-Hydroxy acid salt (S,S)-21 (46.5
kg), LiNH2 (7.9 kg), DMF (87 kg), 4,6-dimethyl-2-methyl-
sulfonylprimidine 8 (26.5 kg), and toluene (8 kg) were heated
to 45 °C for 12 h. Water (90 kg) and toluene (80 kg) were
added. The layers were separated. The aqueous layer was
acidified by addition of 75 kg of 30% HCl and extracted
with 415 kg of ethyl acetate. The organic layer was washed
twice with 90 kg of 0.1 N HCl, and after evaporation of the
ethyl acetate the residue was recrystallised from 150 L of
2-propanol/water (1/1) to give 31.2 kg of a colourless powder
(84%). Purity 99.8% (HPLC); ee ) 99.9% (HPLC); mp )
BSF 302146. Methyl 2-Hydroxy-3,3-diphenylbutyrate
(22). Glycidate 3 (66.5 kg) as a solution in 22 kg toluene
was added to a mixture of 210 kg heptane and 138 kg 2 M
TMA solution in toluene at -5-0 °C. Stirring was continued
for 1-2 h. The reaction mixture was poured in a mixture of
95 kg sulfuric acid and 340 kg water. Toluene (135 kg) was
added. The layers were separated, and the aqueous layer was
extracted with 75 kg of toluene. The organic layers were
combined, and the solvent was evaporated. The residue was
dissolved in 58 kg of 2-propanol and heated to reflux.
Heptane (116 kg) was added, and the solution was cooled
to 0 °C with a rate of 5 °C/h. Stirring was continued for 6
h. The crystalline product was filtered off leading to 55.8
kg of a pale yellowish powder (82%). Purity: 99% (HPLC);
1
mp ) 101-102 °C; H NMR (200 MHz, d6-DMSO): δ )
1.7 (s, 3H), 3.2 (s, 3H), 4.9 (d, 1H), 5.8 (d, 1H), 7.0-7.3
(10H) ppm.
2-Hydroxy-3,3-diphenylbutyric acid (20). R-Hydoxy
ester 22 (105.2 kg), KOH (84.2 kg), and 2-propanol (170
kg) were heated to reflux for 1 h. After cooling the reaction
mixture to room temperature 185 kg of MTBE, 250 kg of
water and 80 kg of 30% HCl were added. The layers were
separated, and the aqueous layer was extracted with 80 kg
of MTBE. The organic layers were combined, the solvent
was evaporated, and the residue was dissolved in 510 kg of
2-propanol.
1
240 °C (decomposition); H NMR (200 MHz, d6-DMSO):
δ ) 1.9 (d, 3H), 2.3 (s, 6H), 5.92 (s, 1H), 6.9 (s, 1H), 7.1-
7.4 (11H) ppm.
Received for review April 10, 2000.
OP000040N
22
•
Vol. 5, No. 1, 2001 / Organic Process Research & Development