Evaluation Only. Created with Aspose.PDF. Copyright 2002-2021 Aspose Pty Ltd.
2. Result and Discussion
Another crucial point in scheme 1 is the amination of 2, we
tried to utilize Gabriel Synthesis with potassium phthalimide
transforming 2 to an intermediate. Although this progress was
available and effective, unfortunately, it was hard to cleavage the
intermediate to get product 3 regardless of the acid condition or
base condition and the cleavaged part had a large proportion of
racemization. Delepine reaction is another method which is
usually used for this reaction. The Delepine reaction between 2
and hexamethylenetetramine made a success with 97% yield and
easy separation.
According to the retrosynthetic analysis, a mixture of
eliglustat and its stereoisomers (5) can be obtained in four steps
(Scheme 1). The synthesis commenced with a commercially
available 1,4-benzodioxan-6-yl methyl ketone (1) which was
subjected to sequentially bromination, amination, and acylation
to produce the intermediate 4 with 93% yield. Mannich reaction
of 4 with paraformaldehyde and pyrrolidine, and then reduction
with NaBH4 led to the formation of mixture 5. The ratio of
diastereosiomers A/B was about 1:1 which was determined by
HPLC. However, direct separation of diastereosiomers A from
crude 5 via chromatography was found to be a challeng due to
the little polarity differences. Further optimization lent an
interesting twist that treatment of crude 5 with HCl in water led
to the spontaneous crystallization of diastereomer A (Eligustat
and its enantiomer) hydrochloride salt which can get nearly 50%
total free diastereomer A by further alkalizing with NaHCO3.
The main byproduct of the synthesis presents to be the
byproduct of Mannich reation transforming compound 4 to
compound
5 by self-elimination of mannich base and
hydroxymethylated elimination of compound 4 (see Scheme S3,
in the supporting information). Because the Mannich base is hard
to separate, we reduct it directly without purification. The
byproduct of Mannich reaction can be also reduced by sodium
borohydride and it is easy to separate from the main product.
O
O
O
O
O
O
O
NH2
HCl
With the successful preparation of diastereomer A (rac-
eliglustat), we further studied the chiral resolution of the desired
(1R, 2R) enantiomer eliglustat. Obviously, a suitable chiral
resolving reagent is supposed to be crucial to this separation.
Classical resolution procedures involving the use of well-known
resolving agents such as chiral camphor-10-sulphonic acid,
tartaric acid, and mandelic acid have been available for a long
time. Rene Imhof etal previously described the use of binaphthyl
phosphoric acids in the resolution of 7-phenylquinolizidines.10 In
2002, Periasamy etal reviewed many novel methods of resolving
racemic diols and amino alcohols.11 Inspired by these
encouraging works, a series of commercially available and
commonly methods used acidic resolving agents L1-L6 (Table 2)
were screened for the enantiomer resolution. Firstly,
diastereomers A and resolving agents were mixed and dissolved
in acetone or ethanol for crystallization (Table 2, entries 1-8).
L1-L2 could not yield any precipitated products both in acetone
and ethanol. As for L3, no resolution effect was detected. While,
upon further optimization, a good enantioselectivity (88.5:11.5 dr)
and moderate yield of eliglustat salt was obtained by using bulky
L4 as the resolution regent. To our delight, the yield of product
could be increased to 78% with 95.6:4.3 dr when sterically
bulkier L5 was employed. These results indicated that steric
hinderance effect may be favorable for the resolution of
diastereomers A. Indeed, further optimization with L6 as
resolution regent in acetone showed an excellent resolution effect
Pyridinium tribromide
(CH2)6N4
O
O
Br
HCl, ethanol
1
3
2
OH
*
O
H
N
(2)
O
(1)
H
N
O
O
n-C7H15
O
C7H15 1)(CH2O)n, Pyrrolidine
2) NaBH4
Cl
*
O
O
N
O
CH3COONa
4
5
mixture of (1R,2R)(1S,2S)(1R,1S)(2S,2R)
OH
H
1)
O
O
N
C7H15
HCl
spontaneous
crystallization
O
N
eliglustat
(1R,2R)
2) NaHCO3
+
~ 14% total yield
OH
H
O
N
C7H15
(1S,2S)
O
O
N
Scheme 1. The preparation of rac-eliglustat with four isomers
To get monobrominated 2-bromine-3’,4’-(ethylenedioxy)-
acetophenone, we firstly used bromine as bromide reagent at
room temperature, however it only generated a mixture which
contained 57.5% monobrominated, 20% dibrominated and some
multibrominated product, monitored by LC-MS. And the
mixture can be purified by crystallization in ethyl acetate. Some
experiments have been made to promote the yield such as adding
different equivalent HOAc and reacting at a lower temperature
but they all failed without any meaningful results. Furthermore,
we tried different bromination reagents. NBS had a low activity
of the bromination with trace product. Pyridinium tribromide
showed an almost perfect behavior in this bromination process
with 98% yield (Table 1).
O
O
OH
OH
OH
OH
HO
OH
HO
O
OH
O
HO
OH
, R-Mandelic acid
OH
L2
, D-(-)-Quinic acid
Table 1.The bromination of 3’,4’-(ethylenedioxy)-acetophenone
L1
L3
, D-(+)-Tartaric acid
Entry Reagenta T ℃ HOAc
Solvent
DCM
Yield %
57.5%
59.1%
63.0%
61.2%
5.2%
O
O
O
O
O
O
Ph
OH
O
OH
O
1
2
3
4
5
6
Br2
Br2
25
0
-
O
O
O
P
-
DCM
OH
HO
Ph
HO
O
O
O
O
Br2
0
0.1 eq
DCM
Br2
0
1 eq
DCM
DCM+HOAcb
L4
, D-(+)-DBTA
L5
, D-(+)-DTTA
L6
, (R)-(-)-BNDHP
(95:5). Upon switching to mixed solvents acetone/dimethyl
sulphoxide(2:1), the yield was increased to 65.2%, but a slightly
diminished enantioselectivity was obtained (7.6:92.4). Pleasingly,
the enantioselectivity and yield could be further improved to
97.7:2.3 and 78.7%, respectively, when ethanol was used as
solvent. Changing the amount of L6 to 0.75 equiv or 1.5 equiv
led to lower enantioselectivity and yield both in ethanol or
acetone.
Br2
0
-
-
NBS
NBS
C5H6Br3N
25
25
25
DCM
1.2%
Acetonitrile
MeOH+DCMc 98.1%
5%
7
-
a The reagent equilvalent is 1.05eq.
b HOAc is used to dissolve the sub strate 3’,4’-(ethylenedioxy)-acetophenone.
DCM is dichloromethane. HOAc is acetic acid
c Pyridinium tribromide(C5H6Br3N) can be dissolved well in
MeOH(methanol).
Table 2. Resolution of racemic eliglustat with resolving agentsa