Large-scale synthesis of a selective inhibitor of the norepinephrine transporter: Mechanistic aspects of conversion of indolinone diol to indolinone aminoalcohol and process implications

Article abstract of DOI:10.1021/op900141r

Development of a scalable synthesis of WAY-315193 is described. Use of LiHMDS as a base and Ti(O-i-Pr)₄ as a Lewis acid was optimal for efficient and reproducible addition of indolinone anion to epoxyalcohol. Conversion of indolinone diol to in

Full text of DOI:10.1021/op900141r

Organic Process Research & Development 2009, 13, 880–887
Large-Scale Synthesis of a Selective Inhibitor of the Norepinephrine Transporter:
Mechanistic Aspects of Conversion of Indolinone Diol to Indolinone Aminoalcohol
and Process Implications
Asaf Alimardanov,* Alexander Gontcharov, Antonia Nikitenko, Anita W. Chan, Zhixian Ding, Mousumi Ghosh,
Mahmut Levent, Panolil Raveendranath,† Jianxin Ren, Maotang Zhou, Paige E. Mahaney,‡ Casey C. McComas,‡
Joseph Ashcroft, and John R. Potoski
Wyeth Research, 401 North Middletown Road, Pearl RiVer, New York 10965, U.S.A., and Wyeth Research, 500 Arcola Road,
CollegeVille, PennsylVania 19426, U.S.A.
Abstract:
at -78 °C. The epoxide 4 was opened with the sodium salt of
dimethylfluoroindolinone in DMF to afford the diol. The diol
6 was further elaborated into the final aminoalcohol hydrochlo-
ride 1 in 30-34% yield via tosylation with p-toluenesulfonyl
chloride (TsCl) in pyridine, isolation of the intermediate
monotosylate, treatment with MeNH2, and conversion to HCl
salt. Dimethylfluoroindolinone was prepared by reduction and
bis-methylation of 7-fluoroisatin by a process developed earlier
as described in a prior publication.3
In our view, this synthesis appeared simple and straightfor-
ward enough to be considered as a basis for preparation of larger
quantities of the drug substance for preclinical and early clinical
supply. However, several issues had to be addressed: the
abundance of chromatographic purifications of the intermediates
(3, 4, 6, and the free base of 1 were oils), the inefficient opening
of epoxide 4 with the sodium salt of indolinone 5 requiring
two equivalents of the epoxide, the low total yield of the
conversion of diol 6 to aminoalcohol 1, and the variability in
the yield of that conversion on larger scale. The development
of a scalable process based on this synthetic route, the
elucidation of mechanistic aspects of the conversion of 6 to 1,
subsequent studies of properties of uncovered intermediates and
side products and their possible utilization are described in this
contribution.
Development of a scalable synthesis of WAY-315193 is described.
Use of LiHMDS as a base and Ti(O-i-Pr)4 as a Lewis acid was
optimal for efficient and reproducible addition of indolinone
anion to epoxyalcohol. Conversion of indolinone diol to
indolinone aminoalcohol was achieved via monotosylation-
methylamination. The possibility of selective formation of
the amidine side product, as well as its utilization for
alternative selective preparation of the target aminoalcohol,
was demonstrated.
Introduction
Drugs that possess norepinephrine reuptake inhibition, either
selectively or in combination with serotonin reuptake inhibition,
have been used for multiple indications including major
depressive disorder, attention deficit hyperactivity disorder,
stress urinary incontinence, vasomotor symptoms, and pain
disorders such as diabetic neuropathy and fibromyalgia.1 In the
search for new candidates with improvements in both potency
and selectivity, one of the lead compounds in the 1-(3-amino-
2-hydroxy-1-phenylpropyl)indolin-2-one series, WAY-315193
(1), was identified.2 Development of a scalable process and
delivery of kilogram quantities of 1 for further biological
evaluation was required.
Addition to the Epoxide
The synthetic route used initially for preparation of 1 is
shown in Scheme 1. The key step of the synthesis was the
Sharpless epoxidation of fluorocinnamic alcohol 3 which
selectively introduced both relative and absolute configurations
at the C-2 and C-3 positions. At the early stages of the project,
allylic alcohol 3 was prepared in two steps from commercially
available fluorocinnamic acid 2 by treatment with MeI in the
presence of Cs2CO3 in acetone, followed by DIBAL reduction
The preparation of allylic alcohol 3 was modified to avoid
the use of expensive Cs2CO3, volatile and toxic iodomethane,
and cryogenic temperatures. The alcohol 3 was prepared in 95%
yield from 3-fluorocinnamic acid 2 via pTSA-catalyzed esteri-
fication in methanol followed by DIBAL reduction in toluene
at -15 to -8 °C. A Sharpless epoxidation has been successfully
scaled up in an industrial environment, but the literature
accounts of such work are scarce. We relied on one of the
original Sharpless publications describing an example done on
a 100-g scale.4 Treatment of a mixture of allylic alcohol 3,
Ti(Oi-Pr)4, and D-(-)-diisopropyl tartrate (DIPT) with a solution
of tert-butylperoxide (TBHP) in methylene chloride afforded
* Corresponding author. E-mail: alimara@wyeth.com.
† Deceased.
‡ Wyeth Research, Collegeville, PA.
(1) (a) For a review on norepinephrine reuptake inhibitors, see: Babu,
R. P. K.; Maiti, S. N. Heterocycles 2006, 69, 539. (b) Krell, H. V.;
Leuchter, A. F.; Cook, I. A.; Abrams, M. Psychosomatics 2005, 46,
379. (c) Hajos, M.; Fleishaker, J. C.; Filipiak-Reisner, J. K.; Brown,
M. T.; Wong, E. H. W. CNS Drug ReV. 2004, 10, 23. (d) McCormack,
P. L.; Keating, G. M. Drugs 2004, 64, 2567.
(3) Wu, Y.; Wilk, B. K.; Ding, Z.; Shi, X.; Wu, C. C.; Raveendranath,
P.; Durutlic, H. Process for the Synthesis of Progesterone Receptor
Modulators. U.S. Patent Publ. Appl. US 2007/027327, 2007.
(4) (a) Gao, Y.; Hanson, R. M.; Klunder, J. M.; Ko, S. Y.; Masamune,
H.; Sharpless, K. B. J. Am. Chem. Soc. 1987, 109, 5765. (b) For a
recent example of large-scale asymmetric epoxidation, see: Henegar,
K. E.; Cebula, M. Org. Process Res. DeV. 2007, 11, 354.
(2) Kim, C. Y.; Mahaney, P. E.; Trybulski, E. J.; Zhang, P.; Terefenko,
E. A.; McComas, C. C.; Marella, M. A.; Coghlan, R. D.; Heffernan,
G. D.; Cohn, S. T.; Vu, A. T.; Sabatucci, J. P.; Ye, F. Phenylamino-
propanol Derivatives and Methods of Their Use. U.S. Patent 7,517,899,
2009.
880
•
Vol. 13, No. 5, 2009 / Organic Process Research & Development
10.1021/op900141r CCC: $40.75  2009 American Chemical Society
Published on Web 08/17/2009

Scheme 1. Initial synthesis of aminoalcohol 1
the epoxide 4 as a clear viscous oil, typically in 90-94% yield
and 89-91% ee after workup. The epoxide was of sufficient
purity to be used directly in the next step. An alternative protocol
was also developed, allowing the epoxidation to be performed
in toluene at a higher concentration (10 v) compared to the
dichloromethane conditions (30 v) and using TBHP solution
in decane to afford epoxyalcohol 4 in 89-93% ee. However,
this higher throughput came at the expense of the yield
(68-82%), in part due to formation of 5-7% (HPLC) of an
epoxyalcohol self-condensation byproduct.
The epoxide opening with indolinone was addressed next.
Fluorodimethylindolinone 5, being an amide, is a poor nucleo-
phile which needs to be deprotonated in order to make it
sufficiently reactive towards the epoxide.5 It is also a weak acid
which would require a strong base to effect deprotonation: the
pKa of the parent 3,3-dimethylindolinone is reported to be 18.5
in DMSO.6 On the other hand, epoxide opening with a
nucleophile is known to be greatly facilitated in the presence
of a Lewis or Brønsted acid.7 The reaction in the absence of an
acid was found to be sluggish. Thus, presence of both a strong
base and an (Lewis) acid is required in the reaction mixture at
the same time for the reaction to occur. Figure 1 shows a
putative schematic of epoxide opening with Ti(IV) as a Lewis
acid. Enhanced acidity of the free OH coordinated to the Lewis
acid may complicate the picture further, affecting the balance
between the acids and the bases.
combined. Extensive decomposition of the epoxide under these
conditions led to the need to use 2 equiv of 4. As the yields
varied substantially from run to run and significant amounts of
byproducts were present, chromatography was required to
isolate diol 6. The reaction was also hard to control as the
mixing of Ti(O-i-Pr)4 complex and NaH was exothermic and
accompanied by vigorous gas evolution.
A screening of a series of base-Lewis acid combinations
was therefore undertaken (NaH, t-BuOK, K2CO3, KOH, i-
PrMgCl, LiHMDS, Et2Zn, tetramethylguanidine, and BuLi were
explored as bases, and Ti(O-i-Pr)4, ZnCl2, BF3 ·Et2O, MgCl2,
Et2AlCl, BCl3, and TiCl4 as Lewis acids). Incomplete conver-
sion, possibly due to acid-base quench, and epoxide decom-
position, possibly due to Payne rearrangement and/or opening
with a nucleophilic base, were the major issues observed in
these experiments. Gratifyingly, it was found that the use of
excess LiHMDS as a base and 1.3 equiv of titanium isopro-
poxide as a Lewis acid with 1:1.3 molar ratio of the indolinone
to the epoxide resulted in virtually complete conversion, giving
the desired diol 6 reproducibly. The product was isolated as a
solid after crystallization from toluene in 72% yield and >97%
purity, sufficient to take it into the next step without further
purification.
Conversion to Aminoalcohol
Early small-scale batches were prepared using sodium
hydride and titanium isopropoxide in DMF. NaH was mixed
with indolinone 5 to form the anion while Ti(O-i-Pr)4 was aged
with the epoxide 4 to form the complex, then the solutions were
The conversion of diol 6 to aminoalcohol 1 was performed
via tosylation and subsequent amination (Scheme 1). Selective
monotosylation of the terminal hydroxy group was carried out
with TsCl and triethylamine in dichloromethane in the presence
of 1-2 mol % of dibutyltin oxide which has been shown to
(5) (a) For indolinone deprotonation for epoxide opening, see: Proudfoot,
J. R.; Regan, J. R.; Thomson, D. S.; Kuzmich, D.; Lee, T. W.;
Hammach, A.; Ralph, M. S.; Zindell, R.; Bekkali, Y. Preparation of
Propanol and Propylamine Derivatives and Their Use as Glucocorticoid
Ligands. WO 2004/063163, 2004. (b) Gillman, K.; Bocchino, D. M.
Preparation of Monosaccharides Prodrugs of Fluorooxindoles Useful
in Treatment of Disorders Which are Responsive to the Opening of
Potassium Channels. U.S. Patent Publ. Appl. US 2004/0152646, 2004.
(c) For amide deprotonation for epoxide opening, see: Albanese, D.;
Landini, D.; Penso, M. Tetrahedron 1997, 53, 4787. (d) Chan, W. N.;
Evans, J. M.; Hadley, M. S.; Herdon, H. J.; Jerman, J. C.; Morgan,
H. K. A.; Stean, T. O.; Thompson, M.; Upton, N.; Vong, A. K. J. Med.
Chem. 1996, 39, 4537.
(6) Bordwell, F. G.; Fried, H. E. J. Org. Chem. 1991, 56, 4218.
(7) (a) Smith, J. G. Synthesis 1984, 629. (b) Parker, R. E.; Isaacs, N. S.
Chem. ReV. 1959, 59, 737.
Figure 1. Assumed intermediate to Ti(IV)- and base-promoted
addition to epoxyalcohol.
Vol. 13, No. 5, 2009 / Organic Process Research & Development
•
881

Scheme 2. Transformations of hydroxytosylate 7
increase the selectivity of tosylation of the terminal hydroxyl.8
When we attempted to follow the kinetics of tosylation by
reverse phase HPLC, a rather unexpected trend was observed:
the amount of monotosylate 7 in the reaction mixture increased
initially and arrived at a maximum (about 75 area % conversion,
bis-tosylate and the starting diol at that point were present at
ca. 6-8 area % each), but after approximately 5 h the
monotosylate apparently started to convert back to the diol. The
latter conversion was puzzling since there was only trace water
in the reaction mixture, not sufficient to cause the tosylate
hydrolysis, and the monotosylate has been shown to be stable
under reverse phase HPLC conditions to eliminate the possibility
of its hydrolysis during the analysis. When the reaction mixture
containing the “hydrolyzed monotosylate” was treated with an
excess of methylamine in ethanol, we observed that instead of
the anticipated diol 6 (which would not react with methylamine),
the reaction mixture contained a new product with the same
molecular ion mass as aminoalcohol 1 (MH+ 361). These
observations pointed to a possible participation of the amide
carbonyl of the adjacent indolinone fragment in the displacement
of the terminal tosylate leading to the formation of a putative
cationic intermediate 9 (Scheme 2).9 This intermediate would
accumulate when the tosylation reaction mixture was aged at
room temperature for a period of several hours. Nucleophiles
would react with 9 at the imidate carbon, displacing the imidate
oxygen and opening the six-membered ring.10 Thus, reaction
with water would lead back to the starting amido-diol 6, while
methylamine would give the corresponding amidino-diol 10.
We could not detect the putative intermediate 9 by a direct
analytical technique, but we were able to isolate and characterize
the product of its reaction with methylamine. The product
proved too unstable to withstand chromatographic isolation on
silica gel, affording indolinone diol 6 instead, but it could be
crystallized from acetonitrile. NMR analysis of the isolated
material confirmed its structure as 10, thus supporting our
assumption as to the nature of the putative intermediate 9. A
full account of the structural assignment of 10 is given in the
Supporting Information.
The relatively high rate of decomposition of monotosylate
7 showed that it was critical to ensure that both formation of
the monotosylate and its subsequent conversion occur quickly
lest we incur higher losses of it to the degradant 9 and, further,
to impurity 10 in the final product which was problematic to
remove. To accelerate monotosylation of the diol, a catalytic
amount of DMAP (10 mol %) was added to the reaction mixture
without noticeable effect on the selectivity of the terminal
tosylation: the reaction went to completion in less than 1 h,
and the amounts of the unreacted diol and the bis-tosylate stayed
at 6-8% level.11 Further screening of tosylation conditions
revealed that similar results can be achieved by switching the
reaction solvent from methylene chloride to acetonitrile even
without DMAP. Complete conversion was achieved in under
one hour at 0 °C, and the selectivity of monotosylation was
higher than under the previous conditions (only 3% of the bis-
tosylate formed).
(8) (a) Martinelli, M. J.; Vaidyanathan, R.; Khau, V. V. Tetrahedron Lett.
2000, 41, 3773. (b) Martinelli, M. J.; Nayyar, N. K.; Moher, E. D.;
Dhokte, U. P.; Pawlak, J. M.; Vaidyanathan, R. Org. Lett. 1999, 1,
447. (c) Martinelli, M. J.; Vaidyanathan, R.; Pawlak, J. M.; Nayyar,
N. K.; Dhokte, U. P.; Doecke, C. W.; Zollars, L. M. H.; Moher, E. D.;
Khau, V. V.; Kosmrli, B. J. Am. Chem. Soc. 2002, 124, 3578.
(9) (a) An example of a related imidate salt formation from tertiary amide
at elevated temperature has been reported: Osornio, Y. M.; Miranda,
L. D.; Cruz-Almanza, R.; Muchowski, J. M. Tetrahedron Lett. 2004,
45, 2855. (b) Examples of imidate salt formation from N-(γ-haloalkyl)-
phthalimide or -amide derivatives on treatment with Ag salts have
been reported: Huenig, S.; Geldern, L. J. Prakt. Chem. 1964, 24, 246.
(c) Deslongchamps, P.; Chriyan, U. O.; Taillefer, R. J. Can. J. Chem.
1979, 57, 3262. (d) For cyclic imide formation via O-alkylation of
secondary amides under basic conditions, see for example: Wang, X.;
Gross, P. H. J. Org. Chem. 1995, 60, 1201.
The reaction of the monotosylate with methylamine (aqueous
or ethanolic solution) turned out to be relatively slow, resulting
in the formation of significant amounts of amidine 10 even when
(10) For hydrolysis of imidate salts, see ref 9c and Deslongchamps, P.;
Dube, S.; Lebreux, C.; Patterson, D. R.; Taillefer, R. Can. J. Chem.
1975, 53, 2791, and references cited therein.
(11) Both bis-tosylate and the unreacted diol go through the remaining steps
of the synthesis unchanged and can be removed when the HCl salt of
the product 1 is prepared.
882
•
Vol. 13, No. 5, 2009 / Organic Process Research & Development

Scheme 3. Optimized conditions for the conversion of 4 to 1
monotosylate formation was fast. We also observed the presence
of the intermediate terminal epoxide 8 in the reaction mixture
before the reaction reached completion which suggested that
tosylate displacement may go through formation of that epoxide,
at least as a possible competing process. That prompted us to
attempt conversion of the monotosylate to the terminal epoxide
8 with a hope that (1) we may be able to achieve that conversion
faster than monotosylate reacts with methylamine and (2) the
terminal epoxide may be more stable toward intramolecular
attack by the amide oxygen. The latter argument could be made
if one envisioned the trajectories of the nucleophilic attack by
the amide oxygen on the terminal tosylate and the epoxide:
attack on the terminal carbon of the epoxide would lead to a
more strained cyclic transition state. That happened to be the
case: when treated with concentrated aqueous solution of NaOH
in the presence of a phase transfer catalyst, the monotosylate
quickly collapsed into epoxide 8 which indeed exhibited
enhanced stability toward degradation versus the monotosylate.
In fact, the epoxide could be isolated and stored at 0 °C without
noticeable decomposition.12 Weaker bases, e.g., organic tertiary
amines, also gave the terminal epoxide, but the reaction rate
was much lower which led to competitive degradation of the
tosylate. As an added benefit of the treatment with aqueous
NaOH, any amount of intermediate 9 that had accumulated in
the reaction mixture to that point was converted to diol 6 which
was easier to remove from the final aminoalcohol 1 than
impurity 10.
Thus, crucial for the successful transformation of diol 6 to
aminoalcohol 1 is the acceleration of monotosylate formation
coupled with immediate and rapid in situ conversion to epoxide
8 to ensure conversion via Path A (Scheme 2) and to suppress
decomposition via Path B. The optimized protocol involved
monotosylation of 6 with TsCl in acetonitrile in the presence
of triethylamine and catalytic dibutyltin oxide at 0 °C for 1 h,
followed by treatment with aqueous NaOH at 0 °C for 1 h.
After a solvent switch to toluene, the resulting terminal epoxide
8 was treated with a solution of methylamine in EtOH that led
to slow epoxide opening at the terminal carbon and formation
of the desired aminoalcohol 1 devoid of the side product 10.
Aminoalcohol 1 was crystallized from an MTBE-EtOH
mixture as a hydrochloride salt to afford the target compound
in 54-58% yield, >99% purity and 99.1% ee (Scheme 3).
Reactions of Amidine 10 and Alternative Routes to
Aminoalcohol 1
In the previous section, we reported that the development
of conditions for rapid formation of monotosylate 7 and its fast
consumption in the subsequent step allowed us to avoid the
formation of significant amounts of amidine 10 as a side product
resulting from the side reaction of the monotosylate. Here, we
will show that this side reaction (Path B, Scheme 2) can be
used by leading to an alternative method of conversion of diol
6 to aminoalcohol 1. This alternative route to 1 was discovered
too late for an equal consideration with the first route afore-
mentioned and may be considered for large-scale preparations
in the future.
We have shown that conversion of monotosylate to inter-
mediate 9 can be forced to completion by heating the reaction
mixture to 40 °C. Subsequent treatment with methylamine
afforded amidine 10 which was prepared in this way in
multigram quantities.
Amidine-diol 10 was selectively converted to a monomesi-
tylsulfonate 11 (Scheme 4), which spontaneously cyclized to
form the cyclic amidinium salt, intermediate 12. This process
is analogous to the degradation route of monotosylate 7
discussed earlier, except that it proceeded instantaneously,
presumably due to higher nucleophilicity of the amidine nitrogen
compared to that of the amide oxygen. The cyclic amidinium
salt 12 proved to be much more stable than the corresponding
oxygen analog 9 and could be crystallized from the reaction
mixture and isolated by filtration. Cationic intermediate 12 was
stable under mildly acidic HPLC conditions and could be
characterized by LC/MS. Its facile hydrolysis with aqueous base
led to the formation of amino alcohol 1.
These results prompted us to investigate the behavior of a
cyclic sulfate that was derived from amidine diol 10. Cyclic
sulfates can be made from 1,2-diols via initial conversion to
cyclic sulfites and subsequent oxidation, and they can be used
for reactions with nucleophiles.13 Typically, the intermediate
cyclic sulfites are not useful for ring-opening with nucleophiles
and carbon-nucleophile ring formation. But much to our
surprise, treatment of 10 with SOCl2 resulted in cyclic sulfite
formation and immediate ring-opening with amidine nitrogen
to form a 1:1 mixture of cyclic zwitterionic intermediates 14
and 15 (Scheme 5), which were detected by LC/MS as the
corresponding hydroxy derivatives 17 and 18 resulting from
hydrolysis under HPLC acidic aqueous conditions. Subsequent
treatment with NaOH afforded a 1:1 mixture of aminoalcohol
1 and its isomer 16 in quantitative combined yield.
(12) Alternative approaches involving conversion of diol 6 to the terminal
epoxide 8 via Mitsunobu protocol or sequential treatment with
trimethyl orthoacetate, acetyl bromide, and base were explored and
found to be less efficient.
(13) Gao, Y.; Sharpless, K. B. J. Am. Chem. Soc. 1988, 110, 7538.
Vol. 13, No. 5, 2009 / Organic Process Research & Development
•
883

Scheme 4. Conversion of monotosylate 7 to amino alcohol 1 via amidine intermediate 10
Scheme 5. Cyclic sulfite preparation from amidine 10 and formation of isomer 16
The selectivity of the cyclic sulfite 13 opening was greatly
improved when the reaction was conducted at -70 °C, affording
a mixture of 16 and 1 in a 6:1 ratio as judged by 1H NMR (8:1
ratio of isolated compounds). The isomers were separated by
flash chromatography on silica gel, affording the isomeric
aminoalcohol 16 in 70% yield.14
This behavior of amidine diol 10 is in contrast with amide
diol 6 (Scheme 6). Cyclic sulfite 19 formed from amide diol 6
exhibited the expected reactivity and did not undergo intramo-
lecular ring-opening - ring closing even at 80 °C. Subsequent
aqueous NaIO4 oxidation did not afford the corresponding cyclic
sulfate 20, and only the product of its hydrolysis, hydroxysul-
fonic acid 21 or its regioisomer, was formed as judged by LC/
MS, possibly forming with the anchimeric assistance of the
amide carbonyl. It should be noted that direct treatment of cyclic
sulfite 16 with methylamine resulted in its hydrolysis to diol 6.
These results demonstrate that one can use the “unwanted”
side-process (Path B, Scheme 2) of cyclization-ring-opening
to selectively generate amidine 10. By applying a second set
of cyclization-ring-opening transformations one can effectively
achieve a formal nitrogen-oxygen switch to convert an amidine
containing an alkyl alcohol to an amide containing an alkyl
amine. Moreover, in the case of amidine diol 10, a selective
conversion to either regioisomer of amide aminoalcohol 1 or
16 is possible.
Conclusions
A scalable process for the synthesis of 1 has been developed.
Isolation of an amidine side product led to conclusions about
the processes involved in the conversion of diol 6 to aminoal-
cohol 1. That allowed the development of a controllable process
884
•
Vol. 13, No. 5, 2009 / Organic Process Research & Development

Scheme 6. Approach through cyclic sulfate of indolinone diol
for the transformation, and the demonstration of an alternative
selective approach either to the target aminoalcohol 1 or its
isomer.
into a 37% aqueous solution of HCl (16 kg, 160 mol, 4 equiv),
maintaining the internal temperature below 45 °C. Phases were
separated. The aqueous layer was extracted with toluene (11.2
kg). The combined organic phase was washed with 5% aqueous
NaHCO3 (1.2 kg) and 10% brine (2.3 kg) and concentrated via
atmospheric distillation to afford 15.35 kg of a solution that
contained 5.85 kg of the title compound by HPLC strength assay
(95% yield) of >97 area% purity. Spectroscopic data are in
agreement with reported values.15
Experimental Section
General. HPLC analysis of the intermediates and reaction
monitoring was carried out on an Agilent 1100 liquid chro-
matograph equipped with a Phenomenex Prodigy ODS3 4.6
× 50 mm column. Standard method: 90:10 to 10:90 over 8
min gradient of water-acetonitrile containing 0.02% TFA, flow
rate 1 mL/min. LC/MS data were obtained on an Agilent 1100
LC system with an Agilent 1100 LC/MS detector equipped with
a 4.6 mm × 50 mm Chromolith SpeedROD column. Standard
method: 90:10 to 10:90 over 4 min gradient of water-acetonitrile
containing 0.02% TFA, flow rate 4 mL/min.
(E)-3-(3-Fluorophenyl)prop-2-en-1-ol (3). To a slurry of
3-fluorocinnamic acid 2 (6.6 kg, 39.1 mol) in MeOH (39 kg),
was added p-toluenesulfonic acid (0.8 kg, 3.7 mol, 10 mol %)
at 20-25 °C. The suspension was stirred at reflux (65-68 °C)
for 4 h. Methanol was replaced with toluene by distillation at
atmospheric pressure to the volume of 26 L, addition of toluene
(51.5 kg), and distillation at atmospheric pressure to the final
volume of 50 L. The resulting solution was washed successively
with 5% aq. NaHCO3 (2.5 kg in 49.5 kg of water) and water
(49.5 kg) and then concentrated via atmospheric distillation to
a volume of ca. 30 L. Karl Fisher titration indicated the presence
of <0.05% water.
(2R,3R)-3-(3-Fluorophenyl)-2-(hydroxymethyl)oxirane (4).
Method A. A reactor was charged with toluene (9.9 kg),
powdered 4 Å molecular sieves (3.2 kg) and D-(-)-diisopropyl
tartrate (1.00 kg, 4.27 mol). The resulting mixture was cooled
to -35 °C, and titanium isopropoxide (0.862 kg, 3.03 mol) was
added. The mixture was stirred at -30 to -40 °C for 50 min,
then a solution of allylic alcohol 3 (4.61 kg, 30.3 mol) in toluene
(7.80 kg) was added over 30 min, keeping the temperature
below -30 °C. The mixture was stirred at -30 to -40 °C for
50 min. A 5.3 M solution of TBHP in decane (9.46 kg, 62
mol) was added over 3 h maintaining the temperature in the
-30 to -40 °C range. The mixture was stirred at that
temperature range, and the reaction progress was monitored by
HPLC. After stirring for 16 h (83.4% conversion) the temper-
ature was raised to -20 °C, and the stirring was continued for
6 h until the level of the starting allylic alcohol 3 reduced to
below 3%. The mixture was warmed up to 20 °C and filtered
through a pad of Celite. The pad was rinsed with toluene (18
kg), and the rinse was combined with the filtrate. A solution of
NaCl (0.24 kg) and NaOH (2.8 kg) in water (2.8 kg) was added
to the batch at 0 to 10 °C, and the resulting biphasic mixture
was stirred at 20 °C for 2 h. A solution of sodium metabisulfite
(3.2 kg) and citric acid (2.3 kg) in water (30 kg) was added at
0 to 10 °C (exotherm!), and the resulting mixture was stirred
at 20 °C for 1 h until the level of the residual peroxide was
The solution of 3-(3-fluorophenyl)acrylic acid methyl ester
in toluene prepared as described above (30 L, 39.1 mol) was
added to a 25% solution of diisobutylaluminum hydride in
toluene (51.7 kg, 61.1 L, 89.9 mol, 2.3 equiv) maintaining
internal temperature between -15 and -8 °C. The reaction
mixture was stirred at -15 to -8 °C for 1 h, then quenched
(14) The corresponding cyclic carbonate could also be formed from amidine
diol 10 by treatment with triphosgene. However, its intramolecular
ring opening-ring closing proceeded much slower and was incomplete
even after 1.5 h at 40 °C, with a 4:1 ratio of 16 and 1 observed after
basic hydrolysis. The reaction could be accelerated and driven to
completion by the addition of Et3N, albeit at the expense of selectivity
(1.7:1 ratio of 16 and 1 after NaOH treatment).
(15) Boulet, S. L.; Filla, S. A.; Gallagher, P. T.; Hudziak, K. J.; Johansson,
A. M.; Karanjawala, R. E.; Masters, J. J.; Matassa, V.; Mathes, B. M.;
Rathmell, R. E.; Whatton, M. A.; Wolfe, C. N. 3-Aryloxy/thio-2,3-
substituted Propanamines and their Use in Inhibiting Serotonin and
Norepinephrine Reuptake. WO 2004/043903, 2004.
Vol. 13, No. 5, 2009 / Organic Process Research & Development
•
885

below 0.5 mg/L (determined using peroxide test strips). The
aqueous layer was separated and discarded. The organic layer
was washed with NaHCO3 solution (1.5 kg of NaHCO3 in 30
kg of water) and brine (3.0 kg of NaCl in 30 kg of water), and
concentrated in vacuum to a volume of ca. 15 L, keeping the
batch temperature below 50 °C. The resulting solution of the
epoxide 4 (13.66 kg, 25.5% assayed concentration) was used
in the next step. Assayed yield: 3.48 kg (68% based on the
amount of allylic alcohol 3). Spectroscopic data are in agreement
with reported values.15 Chiral purity: 93% ee. Chiral HPLC
conditions: column: Chiralpak ADH, 0.46 cm × 25 cm; mobile
phase: isocratic 15% heptane, 85% EtOH; detection wavelength:
215 nm; 1 mL/min, 25 °C.
concentration, 19.3 mol). Toluene (12.3 kg) was added, and
the resulting solution was cooled to 3 °C. Titanium(IV)
isopropoxide (5.1 kg, 17.9 mol) was added over 20 min, keeping
the temperature below 20 °C. The resulting solution was added
to the contents of reactor A, keeping the temperature below 20
°C. This was chased with 8.5 kg of toluene. The mixture was
stirred at 35-45 °C for 4.5 h. HPLC assay at that point showed
0.4 area % of unreacted indolinone 5. The mixture was cooled
to 10 °C and added to 18% HCl solution (69 kg), maintaining
the temperature below 30 °C. The resulting biphasic mixture
was filtered through a pad of Celite. A 32-kg toluene filter
rinse was combined with the filtrate. The phases were separated.
The aqueous phase was extracted with toluene (48 kg). The
combined organic phases were washed with a 2.3% aqueous
NaOH (26.2 kg, 15.0 mol) and a 5.3% aqueous NaCl (26.4
kg). The resulting organic solution was dried by azeotropic
water removal via atmospheric distillation to a volume of 145
L (ca. 55 L of distillate was removed). The concentrated solution
was cooled to 20 °C and filtered through a pad of silica gel
(10.8 kg). The silica gel pad was rinsed with ethyl acetate (75
kg), and the rinse was combined with the main filtrate. The
solution was concentrated by atmospheric distillation to a
volume of ca. 32 L. The product crystallized out of the filtrate
on cooling to -5 °C over 5 h. The precipitated solid was filtered
at -5 °C, washed with heptane (30 kg), and dried on the filter
at 20-25 °C in the stream of nitrogen over 16 h to afford 3.56
kg of the title product (72% yield) as a beige solid. Mp 123
Method B. A reactor was charged with D-(-)-DIPT (13.0 g
46 mmol), 4-Å molecular sieves (90 g), and dichloromethane
(4.00 L) and purged with nitrogen. This was cooled to -15
°C. Titanium isopropoxide (12.19 g, 43 mmol) was added
rapidly to the reaction mixture via the addition funnel, and the
reaction mixture was further cooled to -20 °C. A solution of
allylic alcohol 3 (127 g, 0.854 mol) in CH2Cl2 (380 mL) was
added to the reaction mixture via the addition funnel at a rate
to keep the temperature below -20 °C. The resulting mixture
was allowed to stir at -20 °C for 10 min. A solution of TBHP
in CH2Cl2 (4.5 M, 380 mL, 1.71 mol)16 was added to the
reaction mixture via the addition funnel at a rate to maintain
the temperature below -20 °C and above -25 °C (addition
rate: 7 mL/min). The reaction mixture was stirred at -20 °C
for 4 h, then transferred from the reactor into a solution of FeSO4
× 7H2O (356 g, 1.28 mol), citric acid monohydrate (93 g, 0.39
mol), and deionized water (to the total volume of 1.0 L) chilled
to 0 °C. The rate of transfer was adjusted to maintain the
temperature of the mixture below 10 °C. The resultant mixture
was stirred for 25 min. The organic layer was separated and
filtered through a pad of Celite. The aqueous phase was
extracted with MTBE (2 × 300 mL). Combined organic
solutions were cooled to 0 °C. A 30% solution of NaOH in
brine (100 mL, prepared by dissolving 5 g of NaCl in a solution
of 30.0 g of NaOH in 90 mL of water) was cooled to 0 °C and
added to the combined organic phases. The resulting mixture
was stirred for 2 h at 0 °C. Water (400 mL) was added, and
the layers were separated. The aqueous layer was extracted with
MTBE (2 × 250 mL). The combined organic layers were dried
over Na2SO4 (300 g) and concentrated in vacuum. The residue
was mixed with 700 mL of toluene, and the solvent was
removed in vacuum to afford 125.9 g of the title product as an
oil (88% yield). Chiral purity: 89% ee.
1
°C. H NMR (CDCl3, 300 MHz) δ: 7.34-7.15 (m, 3H),
7.05-6.85 (m, 4 H), 5.53 (s, 1 H), 4.73 (dd, J ) 5, 5 Hz, 1 H),
3.67 (d, J ) 5 Hz, 2 H), 1.45 (s, 3 H), 1.39 (s, 3 H). HRMS
(for MH+) calc.: 348.1406; found: 348.1407.
7-Fluoro-1-[(1S,2R)-1-(3-fluorophenyl)-2-hydroxy-3-(meth-
ylamino)propyl]-3,3-dimethylindolin-2-one Hydrochloride (1).17
To a solution of diol 6 (53 g, 94% strength, 0.144 mol) in
MeCN (400 mL) was added Bu2SnO (0.36 g, 1.44 mmol, 0.01
equiv) and p-toluenesulphonyl chloride (28.8 g, 0.151 mol, 1.05
equiv). Triethylamine (29 g, 0.288 mol, 2 equiv) was added
dropwise at 0-5 °C. The mixture was stirred for 1 h at 0-5
°C (HPLC analysis indicated completion of tosylation). A
solution of NaOH (58 g, 0.72 mol, 5 equiv) in water (400 mL)
was added over 20 min at 0-5 °C. The mixture was stirred at
5 °C for 1 h. Toluene (800 mL) and NaCl (25 g) in water (150
mL) were added to form a biphasic mixture. The layers were
separated. The organic layer was washed with 2N HCl (300
mL) followed by brine (300 mL). The organic layer was diluted
with toluene (700 mL) and concentrated to a volume of ca.
900 mL. The resultant solution was filtered through silica gel
(200 g) plug. The silica gel plug was eluted with toluene (1.5
L). The combined filtrate was concentrated under vacuum to
ca. 300 mL. Methylamine solution in EtOH (33 wt %, 287 mL,
2.3 mol, 16 equiv) was added to the toluene solution, followed
by 400 mL of EtOH. The reaction mixture was stirred at 40-45
°C for 10 h, concentrated via vacuum distillation to ca. 200
mL, diluted with MTBE (500 mL) and washed with water (500
mL). To the organic layer was added 3 N HCl (500 mL). The
7-Fluoro-1-[(1S,2S)-1-(3-fluorophenyl)-2,3-dihydroxypro-
pyl]-3,3-dimethyl-1,3-dihydro-2H-indol-2-one (6). Toluene
(12.2 kg), DMF (2.63 kg) and 7-fluoro-3,3-dimethyloxindole
5 (2.56 kg, 14.3 mol) were charged to reactor A. The
temperature of the mixture was adjusted to 5 °C, and a 1 M
solution of LiHMDS in toluene (36.7 kg, 42.7 mol) was added
over 50 min, maintaining the temperature in the 5-20 °C range.
The mixture was warmed up to 20-25 °C, held at that
temperature range for 25 min, and cooled to 4 °C.
To a separate reactor (B), was charged a solution of epoxide
4 in a toluene-decane mixture (12.7 kg, 25.5% assayed
(17) Optimization of preparation of 1 was finalized after the first kilo-lab
campaign, in preparation for the second one. The project was
deprioritized before the second campaign. Optimized lab-scale demo-
run procedure is described.
(16) Hansen, R. M.; Sharpless, K. B. J. Org. Chem. 1986, 51, 1922.
886
•
Vol. 13, No. 5, 2009 / Organic Process Research & Development

1
mixture was stirred for 5 min. The phases were separated. The
aqueous layer was washed with MTBE (500 mL). To the acidic
aqueous layer was charged MTBE (500 mL), the mixture was
cooled to 0-5 °C and basified with NaOH (50% w/w, 150 g,
100 mL) at 0-20 °C. The mixture was stirred for 20 min. The
phases were separated. The aqueous phase was extracted with
MTBE (500 mL). The combined MTBE solution was washed
with 15% NaCl (170 mL) and concentrated to ca. 250 mL via
atmospheric distillation. To the MTBE concentrate was added
EtOH (2B) (150 mL) followed by HCl (37% aqueous solution,
25.5 g). The mixture was stirred at 20-25 °C for 20 min, cooled
to 0-5 °C over 1 h, and stirred at that temperature for 22 h.
The precipitate was filtered, washed with MTBE (100 mL),
and dried in vacuum at 40 °C for 18 h to afford 33 g of the
title product as a white solid (58% yield). Mp 209-212 °C.
[R]2D5° ) +10.7°. 1H NMR (D2O, 400 MHz) δ: 7.40-7.25 (m,
3H), 7.16-6.97 (m, 4H), 5.47-5.25 (2H, broad m), 3.27-3.20
(2H, broad m), 2.76 (s, 3H), 1.37 (s, 3H), 1.24 (broad s, 3H).
ES+ MS, m/z 361 (MH+). Anal. Calc’d for C20H23ClF2N2O2:
C, 60.53; H, 5.84; N, 7.06. Found: C, 60.43; H, 5.69; N, 6.84.
Sn content: <1 ppm. Enantiomeric purity: 99.1% ee. Chiral SFC
analysis conditions: column: Chiralcel OF 250 mm × 4.6 mm;
mobile phase: 30% ethanol, 0.4% diethylamine in CO2; detec-
tion wavelength: 254 nm; 2 mL/min, 40 °C.
(2S,3S)-3-((E)-7-Fluoro-3,3-dimethyl-2-(methylimino)in-
dolin-1-yl)-3-(3-fluorophenyl)propane-1,2-diol (10). To a
solution of diol 6 (11.4 g, 32.9 mmol) in dichloromethane (70
mL), were added Bu2SnO (0.164 g, 0.66 mmol, 2 mol %) and
triethylamine (13.7 mL, 98.7 mmol, 3 equiv). A solution of
p-toluenesulphonyl chloride (6.27 g, 32.9 mmol) in dichlo-
romethane (30 mL) was added over 15 min at 20 °C. The
reaction mixture was stirred at ambient temperature for 3 h, at
which point HPLC analysis indicated 80:20 areas ratio of
monotosylate: diol. Diisopropylethylamine (5 mL, 28.7 mmol)
was added, the reaction mixture was kept at room temperature
for 18 h, diluted with acetonitrile (150 mL), and stirred at 40
°C for 7.5 h. Methylamine (41 mL of 8 M EtOH solution, 328
mmol) was added at 10 °C. The reaction mixture was allowed
to warm up to room temperature, stirred for 24 h, and then
concentrated in vacuum. MTBE (100 mL) was added, followed
by 2 N HCl (100 mL). The phases were separated. The aqueous
phase was washed with 50 mL of MTBE. The acidic aqueous
phase was cooled to 10 °C. MTBE (100 mL) was added,
followed by 5 N NaOH (40 mL). The precipitated solids were
filtered to afford 4.11 g of crude product. The solids were
slurried in 12 mL of acetonitrile at 81 °C, cooled to room
temperature, and filtered to afford 2.89 g of the title product.
The remaining biphasic aqueous NaOH/MTBE system was
further processed to afford a second crop of product. The MTBE
phase was separated, the aqueous phase was extracted with 50
mL of MTBE. The combined MTBE phase was dried over
Na2SO4 and concentrated to yield 2.81 g of crude product, which
was slurried in 5 mL of acetonitrile at 81 °C, cooled to room
temperature, and filtered to afford 1.608 g of the title product
(38% combined yield). H NMR (400 MHz, CD3CN, 55 °C)
δ: 7.30-7.19 (m, 3 H), 6.98-6.90 (m, 2 H), 6.86-6.76 (m, 2
H), 5.67 (br, 1 H), 4.40 (br, 1 H), 3.58 (dd, J ) 11, 5 Hz, 1 H),
3.46-3.38 (m, 1 H), 3.43 (s, 3 H), 1.66 (s, 3 H), 1.59 (s, 3 H).
For detailed 2D NMR studies see Supporting Information. MS
(positive ESI, for M + H): m/z 361.
Preparation of 7-Fluoro-1-[(1S,2R)-1-(3-fluorophenyl)-
2-hydroxy-3-(methylamino)propyl]-3,3-dimethylindolin-2-
one (1) from Amidine 10. To a mixture of 10 (180 mg, 0.5
mmol), triethylamine (0.21 mL, 1.5 mmol), DMAP (1.2 mg,
0.01 mmol), Bu2SnO (2.5 mg, 0.01 mmol), and THF (4 mL),
was added a solution of mesitylenesulphonyl chloride (110 mg,
0.5 mmol) in THF (1 mL) at 0 °C over 5 min. The mixture
was allowed to warm up to room temperature and stirred for
4 d. The precipitated solids were filtered to afford 183 mg of
the intermediate iminium salt. THF (2 mL) was added to the
isolated solid, followed by 2 mL of 1 N NaOH at 0 °C. The
resultant mixture was allowed to warm up to room temperature,
stirred for 1 h, partially concentrated in vacuum to remove THF,
and extracted with 3 × 5 mL of MTBE. Combined MTBE
extracts were dried over Na2SO4 and concentrated to afford 68
mg of the title product (38% yield).
7-Fluoro-1-((1S,2R)-1-(3-fluorophenyl)-3-hydroxy-2-(meth-
ylamino)propyl)-3,3-dimethylindolin-2-one (16). To a solution
of 10 (1 g, 2.78 mmol) in dichloromethane (15 mL), was added
a solution of SOCl2 (0.63 mL, 8.3 mmol) in dichloromethane
(5 mL) at -70 °C over 10 min. The mixture was stirred at
-70 °C for 1 h, then allowed to warm up to room temperature
overnight. The solution was added to 30 mL of 1 N aqueous
NaOH at 5 °C over 10 min. The biphasic system was stirred at
room temperature for 1 h. Phases were separated. The aqueous
phase was extracted with 2 × 10 mL of dichloromethane. The
combined organic phase was dried over Na2SO4 and concen-
trated to afford 0.856 g of crude products mixture. Flash
chromatography separation (silica gel, dichloromethane-
MeOH, 97.5:2.5, saturated with conc. NH4OH) afforded 0.7 g
of the title product (70% yield), along with 0.09 g of 1 (9%).
1H NMR (300 MHz, CDCl3) δ: 7.47-7.2 (m, 3 H), 7.04-6.89
(m, 4 H), 5.36 (br, 1 H), 4.25 (br, 1 H), 3.55 (br d, J ) 10 Hz,
1 H), 3.19 (br, 1 H), 2.4 (s, 3 H), 1.41 (s, 3 H), 1.31 (s, 3 H).
For detailed 2D NMR studies see Supporting Information. MS
(positive ESI, for M + H): m/z 361.
Acknowledgment
We thank M. Young and E. Muczynska for help with chiral
HPLC analysis.
Supporting Information Available
Details of 2D NMR analyses of compounds 10 and 16. This
material is available free of charge via the Internet at
http://pubs.acs.org.
Received for review June 5, 2009.
OP900141R
Vol. 13, No. 5, 2009 / Organic Process Research & Development
•
887