The use of glycidyl ethers involving aziridinium intermediates and other methodology for the preparation of enantiomerically pure drug candidates

Article abstract of DOI:10.1021/op200310y

An enantiospecific 1,2-amine migration process through an aziridinium intermediate involving ring-opening with potassium phthalimide derivatives to produce precursors to drug candidates was developed. The precursor amine derivatives were readily available by epoxide opening of simple glycidyl ether derivatives. The regioselectivity of the process was shown to provide approximately 85% of the desired rearranged product with subsequent conversion to the desired drug candidate occurring with excellent purity. An alternative approach using the same glycidyl ether derivatives as starting materials that overcame this regiochemical limitation was subsequently demonstrated.

Full text of DOI:10.1021/op200310y

Article
pubs.acs.org/OPRD
The Use of Glycidyl Ethers Involving Aziridinium Intermediates and
Other Methodology for the Preparation of Enantiomerically Pure
Drug Candidates
Timothy A. Ayers,* Timothy J. N. Watson, Witold Subotkowski, John Daniel, and Mark Webster
Molecular Innovative Therapeutics, Sanofi, U.S. Route 202-206, P.O. Box 6800, Bridgewater, New Jersey 08807, United States
S
* Supporting Information
ABSTRACT: An enantiospecific 1,2-amine migration process through an aziridinium intermediate involving ring-opening with
potassium phthalimide derivatives to produce precursors to drug candidates was developed. The precursor amine derivatives
were readily available by epoxide opening of simple glycidyl ether derivatives. The regioselectivity of the process was shown to
provide approximately 85% of the desired rearranged product with subsequent conversion to the desired drug candidate
occurring with excellent purity. An alternative approach using the same glycidyl ether derivatives as starting materials that
overcame this regiochemical limitation was subsequently demonstrated.
benzylserine derivative via diazotization chemistry and
subsequent elaboration. Thus, removal of the serine nitrogen
followed by the incorporation of a new nitrogen moiety makes
this process inefficient and costly owing to the use of the D-
isomer of serine. Finally, reduction of the ester to alcohols 12
and 13 followed by reaction with a phthalimide nucleophile
provided access to key intermediates 14 and 15. It was
anticipated that opening of aziridinium intermediate 5 with the
appropriate phthalimide nucleophile (phth = phthalimido)
would provide a more direct route for the preparation of
compounds 14 and 15. We focused our attention on the benzyl
glycidol derivative 3 (R′ = Bn), because it would provide access
to intermediates that were well characterized from the kilo
synthesis.
INTRODUCTION
■
Drugs for the treatment of psychiatric disorders such as
obsessive compulsive disorder and attention deficit disorder are
continuing targets of the pharmaceutical industry. Indazole 1
and benzisoxazole 2 bearing an isoindoline appendage1 were
advanced as potential drug candidates, and kilogram supplies of
active pharmaceutical ingredient (API) were required to
support their development. Recently, we described process
improvements in the syntheses of isoindolines 1 and 2 leading
Initial studies were conducted with indazole 16 in order to
avoid complications with an unprotected indazole in both the
epoxide reaction and in the subsequent formation and ring-
opening of the aziridinium intermediate. Indazole 16 was
prepared in suitable quantities for the aziridinium studies by
reaction of the Boc-piperazine with chloroimidate 17,2
cyclization and removal of the Boc-group (Scheme 4). Ring-
opening of glycidyl ethers 3 with indazoles 8 or 16 occurred
smoothly to provide the desired amino alcohols (Scheme 5).
For example, reaction of N-Ts piperazine 16 with either benzyl
or trityl glycidyl ether in refluxing ethanol provided the
corresponding secondary alcohols 18a and 18b, respectively.
The use of glycidyl butyrate ester was problematic as cleavage
of the ester group was a major side reaction especially when
EtOH was the solvent. However, the use of THF as the solvent
and Yb(OTf)3 in catalytic amounts provided the best
conversion to the desired product 18c. It was subsequently
shown that the unprotected indazole 8 reacts smoothly with
benzyl glycidyl ether to provide alcohol 18d. Isoxazoles 19 were
prepared in a similar fashion (Scheme 6) from piperidine 9.
to the preparation of kilogram (kilo) quantities of each of these
compounds.2
With the increasing availability of glycidol derivatives 33
owing to the Jacobsen kinetic resolution process,4 we
anticipated that these materials could be readily elaborated to
a precursor 4 that would provide a suitable aziridinium
intermediate 5 under the proper conditions. Aziridinium 5
might then undergo selective ring-opening5−7 upon addition of
a nucleophile to provide product 6 for a new route for the
preparation of 1 and 2 (Scheme 1). We now report our studies
showing that these aziridinium intermediates are viable and do
indeed rearrange in an enantiospecific manner.
DISCUSSION OF RESULTS
■
The previously reported key step to install the stereocenter in
high optical purity for the kilo synthesis of 1 and 2 involved the
SN2 displacement of triflate 7 with the appropriate amine
nucleophile 8 or 9 to give 10 and 11, respectively (Schemes 2
and 3).2,8 The major drawback to this process is that the triflate
was prepared in three steps from the corresponding D-
Received: October 31, 2011
Published: December 14, 2011
© 2011 American Chemical Society
141
dx.doi.org/10.1021/op200310y | Org. ProcessRes. Dev. 2012, 16, 141−147

Organic Process Research & Development
Article
Scheme 1
Scheme 2
Scheme 3
Scheme 4
than the primary mesylate used in the kilo synthesis (Scheme
3), as the primary mesylate decomposed above 0 °C and was
prepared in situ.1 However, if mesylate 20 is stored at room
temperature, slow decomposition is observed after several
hours. Thus, 20 was prepared and used immediately. The
reaction of mesylate 20 with PhthNK (Phth = phthalimido)
was usually performed at 60 °C in MeCN as solvent and was
complete after 4 h; whereas room temperature experiments
required at least 18 h. When other solvents such as THF,
CH2Cl2 or toluene were used, little or very slow reaction was
observed, presumably arising from the poor solubility of
PhthNK in these solvents. Unfortunately, the aziridinium
intermediate 21 was not regioselectively opened because an
87:13 mixture of 14a and 22a was obtained, as is often the case
Finally, the enantiomeric compounds were prepared for
establishment of the enantiomeric purity.9
The aziridinium approach was usually conducted by first
preparing and isolating the mesylate intermediate for
simplification of the workup of the phthalimide displacement
step (Scheme 5). The secondary mesylate 20 is more stable
142
dx.doi.org/10.1021/op200310y | Org. ProcessRes. Dev. 2012, 16, 141−147

Organic Process Research & Development
Article
Scheme 5
Scheme 6
for these reactions.4−6 The ratio of these two products was
confirmed by the 1H NMR spectra10 and HPLC analysis. It was
also found that the use of MsCl was detrimental by producing
chloride 23 as an impurity. Mesyl anhydride (Ms2O) was used
in subsequent preparations of the mesylate to avoid this
complication. However, in the MsCl case, all three compounds
were isolated and their structures were confirmed by extensive
NMR analyses. Determination of the enantioselectivity of the
migration process was accomplished by the synthesis of both
enantiomers and HPLC analysis using a chiral column
(Chiralpak AD). The ee of phthalimide 14a was shown to be
>99%, which demonstrates that the migration proceeds in an
enantiospecific fashion. Reaction of trityl ether 20b with
PhthNK also proceeded in an enantiospecific manner, however
the regioselectivity of the pthalimide displacement is slightly
lower as an 84:16 mixture of 14b and 22b was obtained.
Counterion effects were explored briefly and had no observable
impact, as the use of the cesium salt of phthalimide did not alter
the 84:16 selectivity.
primary alcohol 12 provided only slightly better regioselectivity
as a 91:9 mixture of isomers 14d and 22d is obtained.2 Of
course, in this reaction a mixture of N-H indazole and N-Ms
indazole is obtained as well (this was not problematic as the
methanesulfonamide is cleaved in the subsequent LAH
reduction). For direct comparison, the free indazole derivative
18d was also treated with Ms2O followed by PhthNK and once
again an 87:13 mixture of regioisomers 14d, 14e and 22d, 22e
was obtained. All products were isolated to clearly establish the
intermediacy of an aziridinium species for the primary mesylate
derivative. These results demonstrate that both reactions are
proceeding mostly through the aziridinium intermediate, and
perhaps, 31% of a direct SN2 displacement process occurs in the
primary mesylate reaction.11
With this understanding of the aziridinium process in hand,
we now only needed to demonstrate that the 87:13 mixture of
regioisomers in the aziridinium route could be purified at the
end by formation of the ditosylate salt. It was decided to use the
N-Ts indazole for the route verification, although the N-H
indazole should also be suitable. Thus, 15 g of piperazine 16
was converted into 10 g of 1 using the aziridinium process to
It was also established during the development of the
aziridinium approach that the kilo process (Scheme 1) using
143
dx.doi.org/10.1021/op200310y | Org. ProcessRes. Dev. 2012, 16, 141−147

Organic Process Research & Development
Article
key intermediate 14a (>99% ee, but as an 87:13 mixture with
22a) as shown in Scheme 2. HPLC analysis of the final product
showed >96% chemical purity and >99.5% enantiomeric purity.
Futhermore, the overall yield (34%) compares well with the
kilogram-scale route (28%). More significantly, this process
avoids the use of serine and the necessary transformations of
the serine to the hydroxy ester and is much more convergent
and more cost effective as the required glycidol derivative is
now commercially available in production quantities.
Following the demonstration of the aziridinium approach
with initial drug candidate 1, backup compound 2 was selected
for further development. The aziridinium approach was also
investigated and shown to be an attractive alternative. In this
case, the opposite enantiomer of the desired drug substance
was prepared to demonstrate this process and to provide
analytical data with a reference standard for their use. Now
conversion of alcohol 19 to the corresponding mesylate and
reaction with 5-FPhthNK occurred via aziridinium intermediate
24 to provide an 84:16 mixture of 15 and 25 (Scheme 6) which
is similar to the indazole case. However, formation of the HCl
salt removes some of the undesired regioisomer and provided a
97:3 mixture of 15 and 25. Conversion to the enantiomer of
the drug substance 2 was accomplished by reduction of the
phthalimide 15 with LAH, removal of the benzyl group with
BBr3, and formation of the dimesylate salt to give 58 g of the
enantiomer of 2. The overall yield of this process from
piperidine 9 was 34% and is a significant improvement
compared to that of the kilo synthesis.
we focused on this alternative route using the fluoro-substituted
phthalimide. Thus, generation of triflate 27 under standard
conditions and subsequent reaction with piperidine 9 occurred
cleanly to provide the desired adduct with >99% ee and 100%
regiochemistry as expected in 32% overall yield (unoptimized
from glycidol). Once again, subsequent conversion to the
desired drug candidate 2 was done as previously described.
However, one can quickly envision the use of other
nucleophiles and facile cleavage of the simple unsubstituted
phthalimide group to provide ready access to a plethora of
optically pure 1-amino-3-hydroxy-derivatives from the now
readily available glycidyl compounds.
In conclusion, we have demonstrated ready application of
glycidyl ethers 3 as key precursors to aziridinium intermediates
that can be converted to potential drug candidates. The
migration and ring-opening occur in an enantiospecific fashion
to provide the desired product with >99% ee. In addition, the
aziridinium routes (Schemes 5 and 6) avoided the necessary
three-step synthesis of triflate 7 and subsequent LAH reduction
(Schemes 1 and 2), thereby shortening the synthesis of both 1
and 2. A limitation often observed with these aziridinium
processes, the nonideal regioselectivity, was not problematic in
our case as the unwanted regioisomer is removed in the
downstream processing. We subsequently overcame the
regioselectivity limitation by avoiding the aziridinium inter-
mediate; however, we still used glycidyl ethers as the key
enantiomeric pure precursors for synthesis of the desired
enantiomer. This alternative route (Scheme 7) involving
phthalimide opening of the glycidyl ether was also shorter,
but maintained a potentially unfavorable cost with the use of
triflic anhydride. Even though both of the alternative routes
were only demonstrated on lab scales, we did not anticipate any
road blocks to further scale-up of either route.12
Finally, we were not satisfied with the regioselectivity of the
aziridinium process and were prompted to pursue other
applications of glycidol ethers 3 in an alternative process. It
was anticipated that, if a nonmigratory amino group could be
prepared from glycidyl ethers 3 followed by subsequent
conversion of the hydroxy functionality to a leaving group,
then direct SN2 dispacement with the piperidine 9 could be
achieved. Thus, we reacted glycidyl ethers 3 with PhthNK in
order to produce alcohol 26 (Scheme 7). In the initial
EXPERIMENTAL SECTION
■
General. HPLC conditions used were as follows. Reverse
phase separations were conducted on a Waters Symmetry C18,
3.9 mm × 150 mm; wavelength = 214 nm; flow rate = 1 mL/
min; solvent A: 25% ACN/75% buffer; solvent B: 45% ACN/
55% buffer; buffer = 0.05 M NaH2PO4, pH = 3.0; gradient
100% A to 100% B over 30 min. Chiral separations were
conducted on a Daicel Chiralpak AD, 250 mm × 4.6 mm, 10
μm packing; IPA and heptane mixture as stated with 0.1%
Et2NH. Additional experimental and spectral data are in the
Supporting Information.
Scheme 7
6-Fluoro-3-(1-piperazinyl)-1-(4-methylphenyl)-
sulfonyl-1H-indazole (16). To a solution of Boc-piperazine
(105 g, 0.56 mmol) in THF (1.5 L) at room temperature under
a nitrogen atmosphere was added chloroimidate 17 (100 g, 0.28
mol) portionwise over 2 h. After 2 h, the mixture was
concentrated. NMP (1 L) and powdered K2CO3 (100 g, 0.72
mol) were added, and this mixture was heated to 125 °C. After
2 h, the mixture was cooled to room temperature overnight.
EtOH (1 L) and water (4 L) were added along with a few seed
crystals. After 1 h, the solid was collected and air-dried to give
111 g (85%) of Boc-16: 1H NMR (CDCl3) δ 7.85 (m, 1), 7.74
(m, 2), 7.53 (m, 1), 7.19 (m, 2), 6.99 (m, 1), 3.56 (m, 4), 3.40
(m, 4), 2.37 (s, 3), 1.23 (s, 9); 19F NMR (CDCl3) δ −111.4; IR
(KBr) 1697, 1437, 1417, 1174 cm−1; MS (APCI) m/z (relative
intensity) 419 (M + H) (100), 264 (36). Anal. Calcd for
C23H27FN4O4S: C, 58.21; H, 5.74; N, 11.81. Found: C, 58.22;
H, 5.73; N, 11.56.
experiments, we attempted to open the epoxide with only
PhthNK in refluxing ethanol and found that the resulting
potassium ethoxide reacted with the phthalimido group to give
the opened ester/amide derivative. In order to avoid this
complication, we found that reaction of 10−50 mol % of
PhthNK and 1 equiv of PhthNH (to scavenge the alkoxide and
complete the conversion) with the glycidyl ethers smoothly
provided the desired alcohols 26 (Scheme 7).
With the kilo-scale synthesis having been completed for both
1 and 2, and with only 2 still being pursued as a drug candidate,
144
dx.doi.org/10.1021/op200310y | Org. ProcessRes. Dev. 2012, 16, 141−147

Organic Process Research & Development
Article
To a solution of Boc-16 (111 g, 234 mmol) in EtOAc (1 L)
and THF (0.2 L) cooled to 0−5 °C was bubbled HCl gas for
0.5 h until saturated. Sat. NaHCO3 (1 L) and 15% NaOH
(∼1.5 L) were carefully added until pH = 8. The organic phase
was washed with brine (0.5 L), dried with MgSO4 (10 g), and
concentrated. Plug chromatography on 3.5 L of gravity silica gel
in a 4-L fritted funnel using a MeOH/CH2Cl2 gradient gave 60
Calcd. for C21H25FN4O2: C, 65.61; H, 6.55; N, 14.57. Found:
C, 65.56; H, 6.59; N, 14.27.
2-(3-Benzyloxy-2S-{4-[6-fluoro-1-(4-methylphenyl)-
sulfonyl-1H-indazol-3-yl]piperazin-1-yl)propyl-isoin-
dole-1,3-dione (14a). To a solution of 18a (22 g, 40 mmol)
and i-PrNEt2 (8.0 g, 62 mmol) in THF (200 mL) cooled below
−10 °C was added Ms2O (10.5 g, 60 mmol) portionwise over
15 min. After 1.5 h, sat. NaHCO3 (50 mL) and EtOAc (50 mL)
were added. The organic phase was washed with brine (20
mL), dried with MgSO4 (2 g), and concentrated (30 °C, 30
Torr). MeCN (100 mL) was added, and the solution was
concentrated (30 °C, 30 Torr). The crude oil was dissolved in
MeCN (200 mL), and PhthNK (9.3 g, 50 mmol) was added.
The mixture was heated to 60 °C for 4 h. The solids were
removed by filtration. The filtrate was concentrated. Toluene
(200 mL) was added, and this mixture was filtered and
concentrated. HPLC analysis showed an 87:13 mixture of 14a
(>99% ee) and 22a. [Note: Reaction of the R-enantiomer
(enant-14a) under the same conditions gave an 87:13 mixture
of enant-18a and enant-22a to demonstrate the optical purity of
14a to be (>99% ee.) The absolute configuration of 22a was
not determined but was shown to be (>99% ee)]. Further
conversion of this solution to the drug candidate 1 was
performed as before (Scheme 2) and is described in the
Supporting Information.
1
g (68%) of 16: H NMR (CDCl3) δ 7.85 (m, 1), 7.74 (m, 2),
7.56 (m, 1), 7.20 (m, 2), 6.90 (m, 1), 3.43 (m, 4), 3.03 (m, 4),
2.36 (s, 3); 19F NMR (CDCl3) δ −115.4; IR (KBr) 1614, 1594,
1529, 1372, 1190 cm−1; MS (APCI) m/z (relative intensity)
375 (M + H) (100), 220 (33). Anal. Calcd. for C18H19FN4O2S:
C, 57.74; H, 5.11; N, 14.96. Found: C, 57.50; H, 5.04; N, 14.73.
(2R)-3-Benzyloxy-1-{4-[6-fluoro-1-(4-methylphenyl)-
sulfonyl-1H-indazol-3-yl]piperazin-1-yl}propan-2-ol
(18a). A solution of piperazine 16 (15 g, 40 mmol) and (R)-
benzylglycidyl ether (6.7 g, 40.9 mmol) in EtOH (150 mL) was
heated to reflux for 18 h. The solution was cooled to room
temperature and concentrated. Toluene (100 mL) was added,
and the solution was concentrated to give 22 g (100%) of 18a:
1H NMR (CDCl3) δ 7.82 (m, 1), 7.78 (m, 2), 7.40−7.28 (m,
6), 7.20 (m, 2), 6.99 (m, 1), 4.60 (s, 2), 3.97 (m, 1), 3.51 (m,
5), 2.82−2.21 (m, 6), 2.38 (s, 3); IR (neat) 3493, 1614, 1594,
1536, 1175 cm−1; MS (APCI) m/z (relative intensity) 539 (M
+ H) (100), 384 (16).
In a separate experiment using mesyl chloride, HPLC
analysis showed a 79:11:10 mixture of 14a, 22a, and 23 was
obtained. These compounds were separated by flash
A sample of the S-enantiomer (enant-18a) was prepared
using the (S)-benzylglycidyl ether. HPLC separation of the
enantiomers was achieved using a Chiralpak AD column as
follows: mobile phase (heptane:IPA, 75:25 + 0.1% Et2NH),
flow rate (1 mL/min) and UV detection at 254 nm.
1
chromatography. Compound 14a showed: H NMR (CDCl3)
δ 7.82 (m, 3), 7.74 (m, 4), 7.44 (m, 1), 7.36−7.21 (m, 5), 7.18
(m, 2), 6.94 (m, 1), 4.50 (s, 2), 3.97 (m, 1), 3.70−3.57 (m, 3),
3.28−3.20 (m, 5), 3.01 (m, 2), 2.61 (m, 2), 2.35 (s, 3); 19F
NMR (CDCl3) δ −109.8 (m); IR (KBr) 1771, 1712 cm−1; MS
(CI, CH4) m/z (relative intensity) 668 (M + H) (26), 546
(95), 507 (100); [α]D (MeOH) = −54.51. Undesired
(2R)-3-Triphenylmethoxy-1-{4-[6-fluoro-1-(4-
methylphenyl)sulfonyl-1H-indazol-3-yl]piperazin-1-yl}-
propan-2-ol (18b). A solution of piperazine 16 (10 g, 26.7
mmol) and (R)-triphenylmethylglycidyl ether (8.7 g, 27.5
mmol) in EtOH (100 mL) was heated to reflux for 16 h. The
solution was cooled to room temperature and a solid was
observed. The mixture was concentrated to ∼50% of the
volume. Heptane (50 mL) was added, and the solid was
collected, washed with heptane, and oven-dried (70 °C, 20
1
regioisomer 22a showed: H NMR (CDCl3) δ 7.82 (m, 3),
7.74 (m, 4), 7.51 (m, 1), 7.33−7.20 (m, 5), 7.18 (m, 2), 6.95
(m, 1), 4.73 (m, 1), 4.50 (m, 2), 4.00 (m, 1), 3.79 (m, 1), 3.30
(m, 4), 3.06 (m, 1), 2.69 (m, 2), 2.64 (m, 1), 2.48 (m, 2), 2.33
(s, 3); 19F NMR (CDCl3) δ −109.7 (m); IR (KBr) 1773, 1709
cm−1; MS (CI, CH4) m/z (relative intensity) 668 (M + H)
1
Torr) to give 13.8 g (75%) of 18b: H NMR (CDCl3) δ 7.86
(m, 1), 7.75 (m, 2), 7.52 (m, 1), 7.49−7.41 (m, 5), 7.24−7.16
(m, 12), 6.95 (m, 1), 3.96 (m, 1), 3.44 (m, 4), 3.22 (m, 1), 3.12
(m, 1), 2.76 (m, 2), 2.62−2.44 (m, 4), 2.36 (s, 3); 19F NMR
(CDCl3) δ −111.6 (m); MS (APCI) m/z (relative intensity)
691 (M + H) (100), 243 (75).
A sample of the S-enantiomer (enant-18b) was prepared
using the (S)-triphenylmethylglycidyl ether. HPLC separation
of the enantiomers was achieved using a Chiralpak AD column
as follows: mobile phase (heptane:IPA, 75:25 + 0.1% Et2NH),
flow rate (1 mL/min) and UV detection at 254 nm.
(2R)-3-Benzyloxy-1-[4-(6-fluoro-1H-indazol-3-yl)-
piperazin-1-yl]propan-2-ol (18d). A mixture of 6-fluoro-3-
(1-piperazinyl)-1H-indazole (8) (5.0 g, 23 mmol) and (R)-
benzylglycidyl ether (3.8 g, 23 mmol) in EtOH (50 mL) was
heated at reflux for 18 h. Upon cooling to room temperature,
complete crystallization had occurred. Heptane (25 mL) was
added, and the solid was collected. After oven-drying (60 °C,
100 Torr) for 10 h, 6.2 g (70%) of 18d was obtained: 1H NMR
(CDCl3) δ 9,43 (br s, 1), 7.58 (m, 1), 7.40−7.26 (m, 5), 6.94
(m, 1), 6.82 (m, 1), 4.58 (s, 2), 4.14 (m, 1), 3.58 (m, 7), 3.04−
60 (m, 6); 19F NMR (CDCl3) δ −115.8; IR (KBr) 3266, 1629,
1520, 1443, 1148, 1124 cm−1; MS (APCI) m/z (relative
intensity) 385 (M + H) (100). [α]D (MeOH) = +2.32. Anal.
1
(26), 512 (84), 387 (100). Chloride 23 showed: H NMR
(CDCl3) δ 7.82 (m, 3), 7.76 (m, 4), 7.55 (m, 1), 7.40−7.22 (m,
5), 7.19 (m, 2), 6.97 (m, 1), 4.60 (m, 2), 4.17 (m, 1), 3.80−
3.64 (m, 2), 3.42 (m, 4), 2.83−2.60 (m, 6), 2.38 (s, 3); 19F
NMR (CDCl3) δ −109.6 (m); IR (KBr) 1739 cm−1; MS (CI,
CH4) m/z (relative intensity) 557 (M + H) (10), 541 (14), 401
(100), 387 (92), 219 (79).
Reaction of Mesylate 20b and Phthalimide Nucleo-
philes. A mixture of 20b mesylate (384 mg, 0.5 mmol) and
PhthNK (120 mg, 0.65 mmol) in MeCN (3 mL) was heated to
60 °C for 4 h. The mixture was filtered through Celite and
concentrated. 1H NMR and HPLC showed an 84:16 mixture of
14b and 22b.
In a separate experiment 20b (100 mg, 0.13 mmol), PhthNH
(30 mg, 0.16 mmol) and Cs(CO3)2 (75 mg, 0.23 mmol) in
1
MeCN (2 mL) were heated to 60 °C. H NMR and HPLC
showed an 84:16 mixture of 14b and 22b. No reaction is
observed if only PhthNH is used.
Reaction of 18d with Mesyl Anhydride and Potassium
Phthalimide. To a solution of 18d (2.0 g, 5.2 mmol) and i-
Pr2NEt (1.4 g, 10.8 mmol) in MeCN (20 mL) and THF (15
mL) (note: solubility problems with only MeCN) cooled to
145
dx.doi.org/10.1021/op200310y | Org. ProcessRes. Dev. 2012, 16, 141−147

Organic Process Research & Development
Article
−10 °C was added Ms2O (1.45 g, 8.3 mmol) in portions over
20 min. After 2 h, TLC showed complete conversion to
mesylate. Potassium phthalimide (2.9 g, 15.5 mmol) was added,
and the mixture was heated to 60 °C. After 4 h, the mixture was
cooled to room temperature and concentrated. HPLC analysis
showed a 73.2:10.9:13.9:1.9 mixture of 14d, 14e, 22d, and 22e.
Flash chromatography on silica gel using 20% EtOAc/heptane
as eluant gave samples of pure compounds. Compound 14d
showed: 1H NMR (CDCl3) δ 9.64 (s, 1), 7.82 (m, 2), 7.74 (m,
2), 7.58 (m, 1), 7.43−7.22 (m, 5), 6.94 (m, 1), 6.78 (m, 1),
4.52 (m, 2), 4.00 (m, 1), 3.73−58 (m, 3), 3.40−3.20 (m, 4),
3.10 (m, 2), 2.71 (m, 2), 2.64 (m, 1); 19F NMR (CDCl3) δ
−116.1 (m); MS (CI, CH4) m/z (relative intensity) 514 (M +
H) (100). Compound 14e showed: 1H NMR (CDCl3) δ 7.88−
7.63 (m, 7), 7.36−7.24 (m, 4), 7.08 (m, 1), 4.52 (m, 2), 3.97
(m, 1), 3.73−3.58 (m, 3), 3.38 (m, 5), 3.09 (m, 2), 3.03 (s, 3),
2.65 (m, 2); 19F NMR (CDCl3) δ −111.3 (m); MS (CI, CH4)
m/z (relative intensity) 592 (M + H) (100). Compound 22d
showed: 1H NMR (CDCl3) δ 7.83−7.57 (m, 5), 7.36−7.22 (m,
5), 7.01−6.80 (m, 2), 4.75 (m, 1), 4.51 (m, 2), 4.00 (m, 1),
3.81−2.56 (m, 12); 19F NMR (CDCl3) δ −115.9 (m); MS (CI,
CH4) m/z (relative intensity) 514 (M + H) (100). Compound
22e showed: 1H NMR (CDCl3) δ 8.01−7.61(m, 9), 7.36−7.22
(m, 2), 7.01 (m, 1), 4.73 (m, 1), 4.50 (m, 2), 4.00 (m, 1), 3.81
(m, 1), 3.74−3.59 (m, 2), 3.38 (m, 4), 3.15−3.01 (m, 4), 2.81−
51 (m, 3); 19F NMR (CDCl3) δ −111.2 (m); MS (CI, CH4)
m/z (relative intensity) 592 (M + H) (100).
Comparison of these samples with a retained sample of the
crude reaction mixture from the kilo synthesis (Scheme 2:
conversion of 12 to 14) confirmed that an aziridinium
intermediate is involved in this reaction as a 52.9:38.3:5.0:3.6
mixture of 14d, 14e, 22d, and 22e was generated.
3-Benzyloxy-2-(S)-[4-(6-fluorobenzo[d]isoxazol-3-yl)-
piperidin-1-yl]-1-propanol (19). A solution of piperdine 9
(63 g, 0.29 mol) and (S)-benzyl glycidyl ether (48 g, 0.29 mol)
in EtOH (0.5 L) was heated to reflux for 18 h. The solution was
concentrated to a thick oil. MeCN (100 mL) was added, and
the solution was concentrated to give 112 g (>100%) of 19. A
sample of the (R)-enantiomer (enant-19) was prepared in an
analogous fashion.
2-{3-Benzyloxy-2-(S)-[4-(6-fluorobenzo[d]isoxazol-3-
yl)piperidin-1-yl]propyl}-5-fluoro-isoindole-1,3-dione
Hydrochloride (15). To a solution of 19 (109 g, 0.29 mol)
and i-PrNEt2 (61 g, 0.47 mol) in MeCN (1.1 L) cooled to 0 °C
was added Ms2O (75 g, 0.43 mol) portionwise over 0.5 h. After
3 h, potassium 4-fluorophthalimide (67 g, 0.33 mol) was added
in one portion. The mixture was heated to 60 °C for 4 h. After
cooling to room temperature, the mixture was filtered through
Celite. HPLC analysis showed an 86:14 mixture of regioisomers
15 and 25. The filtrate was concentrated. The crude product
was dissolved in EtOAc (200 mL), and HCl gas was bubbled
through this solution for 15 min until saturated. TBME (100
mL) was added. The solid was collected and air-dried to give
102 g (66%) of 15 (>99% ee) as the hydrochloride salt. HPLC
analysis showed a 97:3 mixture of regioisomers 15 and 25.
(Note: A sample of the (R)-enantiomer (enant-15) was
prepared in an analogous fashion for optical purity determi-
nation.) Compound 15 showed: 1H NMR (d6-DMSO) δ 10.80
(br s, 1), 8.21 (m, 1), 7.98 (m, 1) 7.81 (m, 1), 7.76−7.60 (m,
2), 7.40−7.22 (m, 6), 4.58 (s, 2), 4.22−3.69 (m, 9), 2.60−2.2
(m, 5); 19F NMR (d6-DMSO) δ −103.8 (m, 1), −109.8 (m, 1);
IR (KBr) 1779, 1719, 1617, 1397 cm−1; MS (APCI) m/z
(relative intensity) 532 (M + H) (100). [α]D (MeOH) = +6.7.
Further conversion of this solution to the enantiomer of drug
candidate 2 was performed as before (Scheme 3) and is
described in the Supporting Information
2-[3-Triphenylmethoxy-2-(S)-2-hydroxyl]-isoindole-
1,3-dione (26a). A mixture of (R)-triphenylmethylglycidyl
ether (1.0 g, 3.2 mmol), PhthNH (0.50 g, 3.4 mmol) and
PhthNK (0.60 g, 3.2 mmol) in EtOH (50 mL) was heated to
reflux for 20 h. The mixture was cooled to room temperature
and concentrated. Flash chromatography on silica gel using
1
20% EtOAc/heptane gave 1.27 g (87%) of 26a. H NMR
(CDCl3) δ 7.80 (m, 2), 7.64 (m, 2) 7.42 (m, 6), 7.24−7.16 (m,
9), 4.16 (br s, 1), 3.96−3.73 (m, 2), 3.12 (m, 2), 3.06 (m, 1);
LCMS (ESI) m/z (relative intensity) 486 (M + Na+) (18), 243
(100).
In a separate experiment using only PhthNK, formation of
the desired product and cleavage of the phthalimide to generate
the mixed amide/ester was observed.
2-[3-Benzyloxy-2-(S)-2-hydroxyl]-5-fluoro-isoindole-
1,3-dione (26b). A mixture of (R)-benzylglycidyl ether (4.00
g, 24.4 mmol), 5-FPhthNH (4.30 g, 26.1 mmol), and 5-
PhthNK (4.30 g, 16.6 mmol) in EtOH (110 mL) was heated to
reflux for 18 h. The mixture was cooled to room temperature,
filtered to remove solids and concentrated. Flash chromatog-
raphy on silica gel using 20% EtOAc/heptane gave 5.65 g
(71%) of 26b.
Conversion of 26b to Enantiomer of Intermediate 15.
A solution of 26b (6.0 g, 18.2 mmol) and i-Pr2NEt (2.42 g, 3.27
mmol) in toluene (60 mL) cooled to 0 °C was added a solution
of Tf2O (5.16 g, 18.3 mmol) in toluene (60 mL) while keeping
the reaction temperature below 5 °C. After 1.5 h, the solution
was allowed to warm to room temperature. After 1 h, the
mixture was filtered, washing the solids with toluene (3 × 20
mL). The combined filtrates were concentrated. Flash
chromatography using 25% EtOAc/heptane as eluant gave
4.79 g (57%) of triflate 27 as a white powder.
A solution of 9 (2.0 g, 9.1 mmol) and i-Pr2NEt (1.51 g, 14.9
mmol) in MeCN (100 mL) was added dropwise to a solution
of 27 (4.78 g, 10.4 mmol) in MeCN (90 mL). After 36 h,
HPLC showed 85.3% enant-15 and 14.7% 27 with none of the
regioisomer 25 being observed. Work-up and flash chromatog-
raphy gave 3.85 g (79%) of enant-15.
ASSOCIATED CONTENT
■
S
* Supporting Information
1H NMR spectra of most compounds are provided. Additional
experimental details for the final conversion of the key
intermediates to 1 and enant-2 are also included. This material
is available free of charge via the Internet at http://pubs.acs.org.
AUTHOR INFORMATION
■
Corresponding Author
timothy.ayers@sanofi.com.
ACKNOWLEDGMENTS
■
We thank Rick Nevill, Denis Prat, Christian Richard, and
Philippe Mackiewicz for their intellectual contributions and
passionate discussions. We also acknowledge the extensive
NMR experiments performed by Dirk Friedrich for structure
elucidation.
146
dx.doi.org/10.1021/op200310y | Org. ProcessRes. Dev. 2012, 16, 141−147

Organic Process Research & Development
Article
REFERENCES
■
(1) For a review of isoindole chemistry, see: Bonnett, R.; North, S.
A. The Chemistry of Isoindoles. In Advances in Heterocyclic Chemistry;
Katritzky, A. R.; Boulton, A. J., Eds.; Academic Press: New York, 1981;
Vol. 29, pp 341−399.
(2) Watson, T. J.; Ayers, T. A.; Shah, N.; Wenstrup, D.; Webster, M.;
Freund, D.; Horgan, S.; Carey, J. P. Org. Process Res. Dev. 2003, 7, 521.
See also: Hendrix, J. A.; Shimshock, S. J.; Shutske, G. M.; Tomer, J. D.
IV; Kapples, K. J.; Palermo, M. G.; Corbett, T. J.; Vargas, H. M.; Kafka,
S.; Brooks, K. M.; Laws-Ricker, L.; Lee, D. K. H.; de Lannoy, I.;
Bordeleau, M.; Rizkalla, G.; Owolabi, J.; Kamboj, R. K. ChemBioChem
2002, 3, 999. Leroy, V.; Lee, G. E.; Lin, J.; Herman, S. H.; Lee, T. B.
Org. Process Res. Dev. 2001, 5, 179. Strupczewski, J. T.; Bordeau, K. J.
3-(1-Substituted-4-piperazinyl)-1H-indazoles. U.S. Patent 4,954,503,
Sept 4, 1990.
(3) For reviews of glycidol derivatives, see: Pena, P. C. A.; Roberts,
S. M. Curr. Org. Chem. 2003, 7, 555. Hanson, R. M. Chem. Rev. 1991,
91, 437.
(4) For leading references, see: Nielsen, L. P. C; Stevenson, C. P.;
Blackmond, D. G.; Jacobsen, E. N. J. Am. Chem. Soc. 2004, 126, 1360.
Schaus, S. E.; Brandes, B. D.; Larrow, J. F.; Tokunaga, M.; Hansen, K.
B.; Gould, A. E.; Furrow, M. E.; Jacobsen, E. N. J. Am. Chem. Soc.
2002, 124, 1307. Tokunaga, M.; Larrow, J. F.; Kakiuchi, F.; Jacobsen,
E. N. Science 1997, 277, 936.
(5) For migrations involving aziridinium intermediates with epoxides,
see: Liu, Q.; Marchington, A. P.; Rayner, C. M. Tetrahedron 1997, 53,
15729. Liu, Q.; Marchington, A. P.; Boden, N.; Rayner, C. M. Synlett
1995, 1037.
(6) For leading references on migrations involving aziridinium
intermediates, see: Metro, T.-X.; Duthion, B.; Pardo, D. G.; Cossy, J.
Chem. Soc. Rev. 2010, 39, 89. Harrington, P. J.; Brown, J. D.; Foderaro,
T.; Hughes, R. C. Org. Process Res. Dev. 2004, 8, 86. Andrews, D. R.;
Dahanukar, V. H.; Eckert, J. M.; Gala, D.; Lucas, B. S.; Schumacher, D.
P.; Zavialov, I. A. Tetrahedron Lett. 2002, 43, 6121. O’Brien, P.;
Towers, T. D. J. Org. Chem. 2002, 67, 304. Chuang, T.-H.; Sharpless,
K. B. Org. Lett. 1999, 1, 1435. Draper, R. W.; Hou, D.; Iyer, R.; Lee, G.
M.; Liang, J. T.; Mas, J. L.; Tormos, W.; Vater, E. J. Org. Process Res.
Dev. 1998, 2, 175. Lee, J.; Hoang, T.; Lewis, S.; Weissman, S. A.; Askin,
D.; Volante, R. P.; Reider, P. J. Tetrahedron Lett. 2001, 42, 6223.
Knapp, S.; Morriello, G. J.; Doss, G. A. Tetrahedron Lett. 2003, 44,
2645. Anderson, S. R.; Ayers, J. T.; DeVries, K. M.; Mendenhall, D.;
Vanderplas, B. C. Tetrahedron: Asymmetry 1999, 10, 2655. Harden, R.
C.; Hodgkinson, T. J.; McKillop, A.; Prowse, W. G.; Urquhart, M. W.
J. Tetrahedron 1997, 53, 21. Picq, D.; Cottin, D.; Anker, D.; Pacheco,
H. Tetrahedron 1983, 39, 1797. Zhao, S.; Ghosh, A.; D’Andrea, S. V.;
Freeman, J. P.; VonVoigtlander, P. F.; Carter, D. B.; Smith, M. W.;
Blinn, J. R.; Szmuszkovicz, J. Heterocycles 1994, 39, 163. Morimoto, S.;
Adachi, T.; Watanabe, Y.; Omura, S. Heterocycles 1990, 31, 305. Picq,
D.; Cottin, M.; Anker, D.; Pacheco, H. Tetrahedron Lett. 1983, 24,
1399.
(7) For leading references on migrations involving azirdines, see:
Hu, X. E. Tetrahedron 2004, 60, 2701. Concellon, J. M.; Riego, E. J.
Org. Chem. 2003, 68, 6407.
(8) For the preparation of 9, see: Strupczewski, J. T.; Allen, R. C.;
Gardner, B. A.; Schmid, B. L.; Stache, U.; Glamkowski, E. J.; Jones, M.
C.; Ellis, D. B.; Huger, F. P.; Dunn, R. W. J. Med. Chem. 1985, 28, 761.
Strupczewski, J. T.; Gardner, B. A.; Allen, R. C. 3-(4-Piperidyl)-1,2-
benzisoxales. U.S. Patent 435,5,037, Oct 19, 1982.
(9) The ee’s were obtained using a Chiralpak AD HPLC column.
(10) The methinyl proton adjacent to the phthalimido group in 22 is
clearly separated at δ 4.73 ppm as a useful characteristic signal.
(11) In order to obtain 9% of 22, only 69% needs to go through the
aziridinium intermediate, although the reaction was performed at −20
°C instead of 60 °C and the ratio could be influenced by the reaction
temperature.
(12) Appropriate process saftey review of the chemistry including the
epoxide-opening reaction would be fully addressed before scaling-up.
147
dx.doi.org/10.1021/op200310y | Org. ProcessRes. Dev. 2012, 16, 141−147