Process Research and Development of an Enantiomerically Enriched Allyic Amine, One of the Key Intermediates for the Manufacture of Synthetic Tetracyclines

Article abstract of DOI:10.1021/acs.oprd.5b00274

A robust, cost-effective, and high yielding manufacturing process for enantiomerically enriched (S)-allylic amine 3, a key intermediate for fully synthetic tetracyclines have been developed. Two novel and scalable asymmetric vinylations resulting in high-to-excellent stereoselectivity have been developed for the key step. The final product is purified by an efficient crystallization of a l-tartaric salt. The process described has been used to manufacture ～350 kg of the tartaric salt of 3 with 99.0% ee in 8 steps (35% overall yield) from cheap and readily available dimethyl maleate.

Full text of DOI:10.1021/acs.oprd.5b00274

Subscriber access provided by EPFL | Scientific Information and Libraries
Full Paper
Process R & D of an Enantiomerically Enriched Allyic Amine, One of
the Key Intermediates for the Manufacture of Synthetic Tetracyclines
Wu-Yan Zhang, Philip Hogan, Chi-Li Chen, John Niu, Zhimin Wang,
Danny G Lafrance, Olga Gilicky, Nicholas Dunwoody, and Magnus Ronn
Org. Process Res. Dev., Just Accepted Manuscript • Publication Date (Web): 30 Sep 2015
Downloaded from http://pubs.acs.org on September 30, 2015
Just Accepted
“Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted
online prior to technical editing, formatting for publication and author proofing. The American Chemical
Society provides “Just Accepted” as a free service to the research community to expedite the
dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts
appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been
fully peer reviewed, but should not be considered the official version of record. They are accessible to all
readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered
to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published
in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just
Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor
changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers
and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors
or consequences arising from the use of information contained in these “Just Accepted” manuscripts.
Organic Process Research & Development is published by the American Chemical
Society. 1155 Sixteenth Street N.W., Washington, DC 20036
Published by American Chemical Society. Copyright © American Chemical Society.
However, no copyright claim is made to original U.S. Government works, or works
produced by employees of any Commonwealth realm Crown government in the course
of their duties.

Page 1 of 39
Organic Process Research & Development
1
2
3
4
5
6
7
8
9
Process R & D of an Enantiomerically Enriched
Allyic Amine, One of the Key Intermediates for the
Manufacture of Synthetic Tetracyclines
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Wu-Yan Zhang*, Philip C. Hogan, Chi-Li Chen, John Niu, Zhimin Wang, Danny Lafrance, Olga
Gilicky, Nicholas Dunwoody and Magnus Ronn
Tetraphase Pharmaceuticals, Inc., Process Chemistry R&D, 480 Arsenal Street, Suite 110,
Watertown, MA, 02472
* To whom correspondence should be addressed: wzhang@tphase.com
ACS Paragon Plus Environment
1

Organic Process Research & Development
Page 2 of 39
1
2
3
4
Table of Contents Graphic
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
ACS Paragon Plus Environment
2

Page 3 of 39
Organic Process Research & Development
1
2
3
4
ABSTRACT
5
6
7
8
9
A robust, cost effective and high yielding manufacturing process for enantiomerically enriched
(S)ꢀallylic amine 3, a key intermediate for fully synthetic tetracyclines have been developed.
Two novel and scalable asymmetric vinylations resulting in high to excellent stereoselectivity
have been developed for the key step. The final product is purified by an efficient crystallization
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
of a Lꢀtartaric salt. The process described has been used to manufacture ~350 kg of the tartaric
salt of 3 with 99.0% ee in 8 steps (35% overall yield) from cheap and readily available dimethyl
maleate.
Key Words: allylic amine, eravacycline, tetracycline, asymmetric vinylation, isoxazole,
sulfonamide
ACS Paragon Plus Environment
3

Organic Process Research & Development
Page 4 of 39
1
2
3
4
Introduction
5
6
7
8
9
As part of our ongoing program to develop an efficient manufacturing approach to enone 1, a
key intermediate in the manufacture of fully synthetic tetracyclines, including our lead candidate
eravacycline,1 we have devoted significant efforts to develop manufacturing routes to
intermediates 2 and 3. These two key intermediates are used as building blocks for the
convergent synthesis of enone 1 (Figure 1).2 Here we present process research and development
of intermediate 3.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 1
Manufacture of eravacycline from 2 and 3
H3C
H
CH3
H3C
CH3
N
O
O
N
O
5 Steps
O
O
H
N
N
OCH3
OBn
O
OBn
OTBS
2
3
1
H3C
CH3
F
N
H
H
OH
NH2
O
N
N
OH
OH
H
OH
O
O
O
Eravacycline
The original MyersꢀBrubaker route towards 3 was initially developed into a fit for purpose
process to provide material for the initial supply of enone 1 (shown in Scheme 1).2d The cost of
the starting material, acetylene dicarboxylic acid methyl ester (4), was high (>$1000/kg) and the
ACS Paragon Plus Environment
4

Page 5 of 39
Organic Process Research & Development
1
2
3
4
5
6
7
8
9
yield of the isoxazole forming step to provide 5 was modest (~50%) on laboratory scale when
attempted at Tetraphase. The second step, benzylation of the hydroxyl group of intermediate 5,3
gave the desired benzyl ether 6 but was contaminated with varying amounts of the regioisomeric
NꢀBnꢀisoxazole (10ꢀ20%, vide infra) that could only be removed by chromatography. The
reduction of the ester 6 to the aldehyde 7 performed well on small scale, but increasing the scale
led to the formation of overꢀreduced material (the corresponding alcohol) as well as substantial
amounts of unconsumed 6, necessitating chromatographic purification to isolate 7. The reported
asymmetric addition of vinylzinc to the aldehyde within 7 provided 8 with ~90% yield and 92%
ee, but manufacturing 8 on a significant scale was complicated by the large amounts of solids
formed during the preparation of divinyl zinc. The filtration to remove the formed solids was
extremely slow, and if the reaction was done in the presence of these solids, essentially racemic
alcohol 8 was obtained. The mesylation of the alcohol 8 and subsequent substitution of the allylic
mesylate 9 with dimethylamine to provide 3 produced a mixture of regioisomeric substitution
products as well as another significant side product (vide infra).
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Scheme 1
Initially reported route to the isoxazole 3
ACS Paragon Plus Environment
5

Organic Process Research & Development
Page 6 of 39
1
2
3
4
5
6
Results and Discussion
Formation of isoxazole 5
7
8
9
A literature search revealed a reported preparation of the hydroxyl isoxazole 5 from dimethyl
2,3ꢀdibromosuccinate (10, see Scheme 2) using hydroxyurea (11) as the hydroxylamine source
promoted by sodium methoxide.4 We were intrigued by this report as we imagined that 10 would
be easily manufactured by bromination of dimethyl maleate or dimethyl fumarate.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Scheme 2
Preparation of 5 from dimethyl dibromo succinate (10)
Experimentation revealed that 10 could be simply prepared by refluxing a solution of dimethyl
maleate (12, see Scheme 3, chosen due to its cost relative to dimethyl fumarate) in
dichloromethane (DCM) with molecular bromine (1.1ꢀ1.3 equiv.) over 2ꢀ3 days. In the
laboratory, the reaction times could be shortened to as little as 24 h by the inclusion of azoꢀ
bis(isobutyronitrile) (AIBN). Further investigations led to the discovery that the bromination
proceeded smoothly in the presence of a small amount of water and was significantly accelerated
under light radiation from a regular incandescent light bulb. Using a combination of AIBN, water
and light, this reaction is normally completed in 7 h in a plant setting. We observed a significant
exotherm at the beginning of the reaction as the addition of bromine commenced. Analysis of the
1
reaction mixture by HPLC and HꢀNMR after a small amount of bromine was added indicated a
nearly complete isomerization of 12 into dimethyl fumarate (13), and we attributed the observed
ACS Paragon Plus Environment
6

Page 7 of 39
Organic Process Research & Development
1
2
exotherm to this phenomenon (Scheme 3).5 After the consumption of both 12 and 13 by HPLC,
excess bromine was carefully quenched by the addition of aqueous solutions of inorganic
reducing reagents, typically sodium thiosulfate, sodium sulfite or sodium bisulfite. During
laboratory runs, the dibromide 10 was typically isolated in quantitative yield as a solid comprised
of two diastereomers. During one of manufacturing campaigns, a significant amount of
13(~20%) was observed after the quench. Investigation showed that if solutions of 10 are left in
prolonged contact with concentrated solutions of these inorganic reducing reagents at room
temperature (~20 °C), the reduction of significant amounts of 10 to produce 13 readily occurs
over the course of a few hours. The debromination of vicinalꢀdibromide to the corresponding
(E)ꢀalkenes has been well documented.6 The rate of this reverse reaction was dependent on both
the concentration of the reducing reagent in the aqueous layer as well as the temperature during
the quench and phase cut. Experimentation revealed sodium bisulfite to be the mildest of the
reagents screened, and that if a dilute solution (5 wt%) was used at colder temperatures (0ꢀ5 °C)
10 was stable for several hours. Typically, the plant yields were 90ꢀ92% after isolation of the
product by precipitation with heptane. Since 10 is a skin irritant, a solvent switch from DCM to
methanol was performed after work up and 10 was directly used in the next step without
isolation during manufacturing.
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Scheme 3
Bromination of dimethyl maleate (12)
ACS Paragon Plus Environment
7

Organic Process Research & Development
Page 8 of 39
1
2
3
4
5
6
7
8
9
We next investigated the conversion of 10 to isoxazole 5, which has been reported in the
presence of hydroxyurea and sodium methoxide at ambient temperatures using methanol as a
solvent.4 In our development work, the best results were obtained by the addition of sodium
methoxide in methanol (4 equiv.) to a suspension of 10 (1 equiv.) and hydroxyurea (11, 1.05
equiv.) in methanol while keeping the temperature below 5 °C. Sodium methoxide could also be
replaced by potassium tꢀbutoxide. Isoxazole 5 was rapidly formed along with a significant sideꢀ
product (~25%) whose structure is tentatively assigned as that of 14 based on LC/MS and NMR
(Figure 2). This sideꢀproduct is consumed upon warming the reaction to 20ꢀ30 °C. However,
yield assays for 5 before and after warming were similar, suggesting that 14 is not a productive
intermediate. This was further confirmed by following the area % integration of the product and
side product by HPLC against an internal standard of chlorobenzene as the reaction was warmed
from 0 °C to room temperature. Results demonstrated a slow disappearance of the side product
while the concentration of 5 remained constant throughout the process. Significant effort was
devoted to lower the initial ratio of the sideꢀproduct 14. However, the formation of 14 proved to
be surprisingly resilient in a wide range of reaction conditions but fortunately compound 14
decomposes when the reaction mixture is warmed up and ends up being purged during the
precipitation of product 5 from the aqueous solution (see discussion below).
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
The isoxazole 5 is quite stable in the basic methanolic solutions as long as moisture is
excluded. Direct aqueous acidic quench of the reaction mixture led to varying amounts of
hydrolysis of the methyl ester within 5 to give the carboxylic acid 15 (Figure 2) even if the
quench is performed rapidly. On the other hand, addition of the reaction mixture containing 5 to
an aqueous acidic solution led to a smooth quench without the formation of 15. On laboratory
scales, after quench and removal of some inorganics by filtration, the methanol was distilled and
ACS Paragon Plus Environment
8

Page 9 of 39
Organic Process Research & Development
1
2
3
4
5
6
7
8
9
5 could be isolated by extraction from the resulting aqueous mixture using MTBE or 2ꢀ
methyltetrahydrofuran (MeTHF), the latter performing more efficiently due to better solubility.
On larger scales, two observations were made. At very low pH (<2) a substantial amount of 15
was formed even after just a few hours at distillation temperature (30ꢀ40 °C). A pH range of 3ꢀ4
provided some additional stability. And at an even higher pH (6ꢀ7) the process appeared to be
quite robust during the long distillation times required to remove the methanol in the plant.
Another observation was that 5 precipitated from acidic aqueous solutions once the methanol had
been removed by distillation. The product obtained from this precipitation was >95% pure and it
could be recovered directly in good yield (~70%). For manufacturing, the reaction mixture is
quenched into a mixture of aqueous HCl/H3PO4 targeting a final pH of 3 to 6. The pH of the
mixture is adjusted to 6ꢀ7 and the methanol is distilled to give an aqueous solution of 5. The
hydroxyl proton within 5 is acidic enough to form a salt at this pH range. Reꢀacidification of this
mixture results in precipitation of 5 which can be collected by filtration.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 2
Side products formed en route to 5.
Benzylation of 5
The regioselective (O vs. N, 91:9 ratio) benzylation of the hydroxyl group of 5 using benzyl
bromide in acetone has been described,3 and was successful in our hands at small scale.
However, modest scale up of the heterogeneous reaction conditions provided the undesired NꢀBn
derivative 16 in 5ꢀ 20% yield depending on the scale (Scheme 4).
ACS Paragon Plus Environment
9

Organic Process Research & Development
Page 10 of 39
1
2
3
4
5
6
Scheme 4
Benzylation of 5
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
This benzylation was examined in several solvents (i.e., THF, acetone, toluene, DMF or
NMP), with bases such as K2CO3, Na2CO3, DBU, or DIPEA and at temperature ranging from
room temperature up to 60 °C. The best results were achieved by adding benzyl bromide (1.1ꢀ1.2
equiv.) slowly (3ꢀ4 h) into a mixture of 5 and K2CO3 (2 equiv.) in 5 volumes of NMP at 60 °C.
These reaction conditions resulted in acceptable yields of the desired product with ~2% of 16
present. Recrystallization and reꢀslurry experiments were performed in an attempt to remove 16,
but proved unsuccessful due to the high solubility of 6 in most organic solvents and led to
significant losses of product to the mother liquor. Recrystallization with aqueous systems was
sometimes successful, but oiling out of 6 was often a problem. Finally it was found that a
filtration of the toluene extract through a small pad of silica gel could remove the more polar 16
as well as other baseline impurities. The final solution of the product in toluene could then be
telescoped directly into the subsequent reduction step. This procedure was used for initial
manufacturing campaigns with reasonable success with varying amounts of the side products 17
and 18 observed. The formation of 17 is also exacerbated during the work up if nonꢀacidic
aqueous mixtures are added directly to the reaction mixture, particularly if done slowly. The
carboxylic acid 17 is easily removed into basic aqueous washes and any 18 formed is recovered
with 6.
ACS Paragon Plus Environment
10

Page 11 of 39
Organic Process Research & Development
1
2
3
4
5
6
7
8
9
After the initial campaigns, we reꢀvisited using DBU as base. When DBU was used in
conjunction with benzyl bromide, only 4ꢀ5% of 16 was observed, however DBU was rapidly
converted to a quaternary salt with benzyl bromide leading to unproductive consumption of both
reagents. The use of benzyl chloride led to a significantly cleaner reaction with nearly
undetectable amounts of 16 and with the quaternization reaction being much slower. The
reaction could be completed using an acceptable excess of both DBU and BnCl. Additionally, as
long as the moisture levels are low at the start of the reaction (KF<0.1%), very little 17 and 18
are observed. For current manufacturing campaigns the solid 5 is dissolved in NMP (4 volumes).
DBU (~1.25 equiv.) and BnCl (~1.28 equiv.) are charged and the resulting mixture is stirred for
24 h at 5ꢀ10 °C. When the reaction is complete, the mixture is added to aqueous HCl and toluene
to mitigate the formation of 17. The toluene solution obtained can be used directly in the next
step.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Preparation of the aldehyde 7 from 6
As described previously, ester 6 can be reduced to aldehyde 7 using DIBAL.3 However, the
product was frequently contaminated with unacceptable amount of alcohol 19 necessitating a
chromatographic purification. Since the downstream products resulting from the processing of 19
were difficult to remove, our focus shifted to a twoꢀstep reduction oxidation sequence that uses
the alcohol 19 as an intermediate in the preparation of 7 from 6 (Scheme 5).
The reduction of 6 to produce 19 generated excellent yields using NaBH4 in ethanol.
Additionally, the alcohol 19 was a solid that could be recrystallized to high purity. The oxidation
of 19 to produce 7 proved more complex than initially anticipated. We initially used a DMSOꢀ
SO3*pyridine based system which we had successfully used in another program. This reagent
generated good yield of the desired aldehyde in reactions up to 50 g scale, but in larger reactions
ACS Paragon Plus Environment
11

Organic Process Research & Development
Page 12 of 39
1
2
3
4
5
6
7
8
9
the conversion was not complete and the reaction stalled at 50ꢀ70% product. Addition of more
reagents did not result in complete conversion; instead an increase in impurities was observed,
the two most problematic being tentatively assigned as the methyl thioꢀmethyl ether 20 and the
sulfate ester 21. These two impurities were formed in 3ꢀ10% and up to 25%, respectively.
Despite efforts to optimize this reaction by changing solvents, amount of reagents, addition times
and addition order we were not successful in developing a robust oxidation with this reagent.
Additionally, we screened a number of oxidation conditions including activated MnO2, several
conditions using TEMPO and various Swernꢀlike conditions. The TEMPOꢀbased conditions
invariably led to substantial over oxidation of 19 to the carboxylic acid 17. The best results were
achieved using the cryogenic Swern condition. This system was selected for further
development. The reaction proceeded by adding a solution of 19 to the reagent obtained from
DMSO (2 equiv) and oxalyl chloride (1.3 equiv) at ꢀ70 to ꢀ60 °C for 1 h. Triethylamine (3
equiv.) was added and the mixture was warmed up slowly to produce, after workup, 7 in 91%
yield in a manufacturing campaign. Critical to obtaining a good yield of 7 was the hold time at
colder temperature after addition of triethylamine as the consumption of 19 took approximately 1
h at ꢀ70 to ꢀ60 °C. If the reaction was allowed to warm up too soon, the yield was significantly
reduced due to the formation of deleterious side products. This twoꢀstep reductionꢀoxidation
sequence was used to produce sufficient 7 to support both discovery and early development.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
ACS Paragon Plus Environment
12

Page 13 of 39
Organic Process Research & Development
1
2
3
4
Scheme 5
Preparation of 7 by a two-step sequence from 6 or direct reduction of 6
5
6
7
O
O
DIBAL
-78 oC
8
9
O
O
H
MeO
N
N
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
7
OBn
OBn
6
HO
O
N
19
OBn
Py-SO3
DMSO, base
O
MeS
O
-O3SO
O
H
O
N
O
N
+
N
20
21
7
OBn
OBn
OBn
We were disappointed that the cryogenic condition could not be avoided in the twoꢀstep
approach and reexamined the direct DIBAL reduction of 6 to 7. Further investigations revealed
the key factors necessary for a successful scale up in the manufacturing facility. We used toluene
as solvent to allow for telescoping of the material from the previous step. On larger scales the
starting material in the reaction mixture precipitated when cooling to <ꢀ70 °C and we believed
this precipitation was responsible for the formation of significant amounts of alcohol 19 as well
as the incomplete consumption of 6. We found that inclusion of DCM increased the solubility of
6 under the reaction condition and thereby avoiding the precipitation. After the reduction is
complete, methanol is slowly added while maintaining the internal temperature below ꢀ70 °C to
quench the remaining reagent. The low temperature is critical to avoid reduction of 7 during the
ACS Paragon Plus Environment
13

Organic Process Research & Development
Page 14 of 39
1
2
3
4
5
6
quench. To date, yields in the manufacturing facility have been consistently high (89ꢀ99%) with
>92% HPLC purity.
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Enantioselective vinyl addition to 7
A chiral vinyl addition to aldehyde 7 has been developed by Myers and Brubaker (Scheme 6)
using the putative stoichiometric complex 24 derived from the norephedrine derivative 22. The
stereochemical model 25a was proposed to explain the sense of induction observed.2d
Scheme 6
Myers-Brubaker vinyl addition to 7
H3C
O
OH
O
Ph
N
n-BuLi
O
N
Li
L
Zn
Zn
CH3
22
24
23
7
H3C
Ph
O
L
N
H3C
O
O
Ph
N
Li
O
O
Zn
N
O
H
Li
O
OBn
Zn
O
H
N
25b
25a
OBn
(R)-8
(S)-8
The use of divinylzinc (23) as reported provided 8 in high yield and ee in our laboratory.
Divinylzinc was prepared by the addition of vinylmagnesium bromide to a solution of ZnCl2 in
ACS Paragon Plus Environment
14

Page 15 of 39
Organic Process Research & Development
1
2
3
4
5
6
7
8
9
diethyl ether. After addition of 1,4ꢀdioxane, a large amount of magnesium halide salts formed
and was removed by a slow and impractical filtration. Attempt to use 23 without filtration
resulted in the formation of essentially racemic 8. Use of ZnI2 or ZnBr2 as the zinc source while
omitting the filtration resulted in improved, but still unacceptable, ee of 8 (60% ee and 66% ee
respectively). From earlier experiments where ZnCl2 was used to produce the divinyl zinc it was
clear that removing inorganic salts was necessary to obtain 8 with acceptable optical purity. We
hypothesized that the excess magnesium salts generated interfered with a highly organized
transition state for asymmetric induction and this prompted us to think about ways to generate an
organized complex absent of excessive amounts of Lewis acidic salts. We thought this could be
achieved by incorporating all the Lewis acid salts in the system into the transition complex. We
investigated the use of alkyl zinc reagents both as a base to deprotonate amino alcohol ligands as
well as a zinc source that could participate in an organized complex7 to deliver a vinyl group to 7
in a highly enantioselective manner. After initial screen, a study was performed using the
combination of different alkyl zinc reagents, vinyl magnesium bromide and amino alcohol 26,
which can be easily prepared from ephedrine, shown to be a privileged additive for this substrate
(Scheme 7).
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
The reaction of a dialkylzinc (1 equiv.) with 26 (2 equiv.) followed by the addition of
vinylmagnesium bromide (2 equiv.) provided a complex that reacted with 7 (1 equiv.) at ~ ꢀ78
°C to provide 8 with acceptable ee. In this procedure the magnesium atom is part of the
transition complex and filtration is avoided. As can be seen from Table 1 below the reaction is
relatively insensitive to the solvent used (Entries 1ꢀ6). However, the use of only 1 equiv. of
vinylmagnesium bromide led to loss of selectivity (Entry 7).
ACS Paragon Plus Environment
15

Organic Process Research & Development
Page 16 of 39
1
2
3
4
Scheme 7
New enantioselective vinylation of 8
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Table 1
Optimization of second generation vinylation of 7
Alkyl zinc
(1eq.)
vinylMgBr
Temperature
(°C)
Entry
1
Solvent
Toluene
ee (%)
76
(eq.)
2
Et2Zn
ꢀ78
2
3
4
5
6
Et2Zn
Me2Zn
Me2Zn
Me2Zn
Me2Zn
2
2
2
2
2
THF
Toluene
ꢀ78
ꢀ78
ꢀ78
ꢀ78
ꢀ78
84
77
73
80
81
MTBE
Diethyl ether
1,4ꢀdioxane
7
Et2Zn
1
THF
ꢀ75
ꢀ2
Our mechanistic hypothesis is shown in Scheme 8. The diethylzinc reacts with 26 to provide
the Zn (II) alkoxide 27. Addition of vinylmagnesium bromide (2 equiv.) to this complex would
formally lead to 28 (2 equiv.) and divinylzinc (1 equiv.). Complexation of divinylzinc with 28
analogous to 24 shown above produces 29. Reaction of 7 with 29 followed by delivery of the
vinyl group as shown in 30, where the aldehyde oxygen is chelated with magnesium atom and
the vinyl group is transferred via a sixꢀmembered transition state to exert the aldehyde reꢀface
ACS Paragon Plus Environment
16

Page 17 of 39
Organic Process Research & Development
1
2
3
4
5
6
selectivity. This could be an extension of the Oppolzer system where the lithium atom is replaced
by magnesium.8
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Scheme 8
Mechanistic hypothesis for second generation vinylation
On multiple occasions, the use of diethylzinc led to the appearance of a byꢀproduct derived
from the addition of ethyl group to the aldehyde within 7. We attributed this side reaction to the
incomplete reaction of diethyl zinc with 26. The most reproducible results were obtained when
26 and diethylzinc were heated at 60 °C for several hours to complete the consumption of
ACS Paragon Plus Environment
17

Organic Process Research & Development
Page 18 of 39
1
2
3
4
5
6
7
8
9
diethylzinc, followed by cooling and the addition of vinylmagnesium bromide (2 equiv.). A
solution of 7 was added to the resulting reagent at ꢀ78 °C to reproducibly provide 8 in good
yields (~90%) and ee (89% to 92%) for up to 0.5 kg scale of intermediate 7. After work up,
dimethyl ephedrine 26 can be easily recovered by acid/base extraction and reꢀused. This process
was transferred to a manufacturing facility where 60 kg of the desired alcohol was produced
using a 1500 L reactor in multiple batches with an average ee of 86.4% and 88% yield. We
discovered that the formation of 27, which required high temperature, and vinylation of 7, which
required cryogenic conditions, must be conducted in one reactor to avoid loss of selectivity.
Switching flasks/reactors after the formation of 27 resulted in lower ee both at lab scale and in a
manufacturing facility. Because the cryogenic condition requires a different heat transfer media
from heating in a manufacturing facility setting, a significant amount of engineering planning is
necessary to convert the reactor for heating into a cryogenic vessel in a short timeframe. This
prompted us to develop an alternative vinylation approach (vide infra).
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Conversion of 8 to 3
The mesylation of 8 proceeded quickly and cleanly, but the following displacement led to loss
of yield mainly due to two major impurities, tentatively assigned as vinylmesylate 31 and linear
amine 32 (Scheme 9). To minimize the two impurities, the displacement should initiate at low
temperature (ꢀ20 to ꢀ15 °C) followed by a gradual warm up to 0 °C to drive the reaction to
completion. We initially screened different leaving groups (i.e., tosyl, 4ꢀ and 2ꢀ
nitrophenylsulfonyl and trifluoro methylsufonyl) in an attempt to suppress these impurities but
we concluded that mesylate was optimal. We investigated several sources of dimethylamine
along with various solvents (Table 2). In all cases, the mesylate was formed at ꢀ20 °C,
dimethylamine was added at the same temperature and then warmed up to 0 °C and quenched.
ACS Paragon Plus Environment
18

Page 19 of 39
Organic Process Research & Development
1
2
3
4
5
6
For all conditions the amount of linear amine 32 was approximately 5%. DCM was the best
solvent and that dimethyl amine in EtOH gave the least amount of vinylmesylate 31 (Table 2).
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Scheme 9
Mesylation and dimethylamine displacement
Table 2
Optimization of mesylation-substitution to form 3
Entry
Solvent
DCM (10v)
DCM (10v)
Toluene (5v)
Toluene (10v)
2ꢀMeTHF (10v)
MTBE (10)
EOAc (10v)
DCM (20v)
DCM (10v)
THF (10v)
Amine solution (6eq)
THF
31:3
6.7:1
8.3:1
3.6:1
4.3:1
1.7:1
3.8:1
2.6:1
12.5:1
6.7:1
1
2
EtOH
3
EtOH
4
EtOH
5
EtOH
6
EtOH
7
EtOH
9
EtOH
10
12*
aqueous
aqueous
* In this reaction, most of the starting material 8 was recovered.
ACS Paragon Plus Environment
19

Organic Process Research & Development
Page 20 of 39
1
2
3
4
Salt formation of allylic amine
5
6
7
8
The allylic amine, which is an oil, is first purified by an acid/base extraction. It could be
further purified by column chromatography but we wanted to investigate a salt formation for the
purpose of purity upgrade as well as an upgrade of chiral purity in addition to a preference for
crystalline solid during large scale manufacture, storage and shipping. A screen using 12
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
different chiral acids (i.e.,
toluoylꢀ ꢀtartaric acid, ꢀ(+)ꢀtartaric acid,
acid, Nꢀacetylꢀ ꢀleucine, ꢀ(ꢀ)ꢀquinic acid,
L
ꢀ(ꢀ)ꢀ10ꢀcamphorsulfaonic acid, (+)ꢀ10ꢀcamphorsulfaonic acid, diꢀpꢀ
ꢀtartaric acid, ꢀ(ꢀ)ꢀmandelic acid, ꢀ(+)ꢀmandelic
ꢀpyroglutanic acid, ꢀ(ꢀ)ꢀmalic acid and ꢀ(+)ꢀmalic
D
L
D
D
L
L
D
L
L
D
acid) was initiated using 7 different solvents for each one (i.e., toluene, ethyl acetate, MTBE,
IPA, acetone, water and acetonitrile). From the screen we found that amine 3 formed crystalline
salts with diꢀpꢀtoluoylꢀ
D
ꢀtartaric acid (DTTA),
ꢀtartaric acid did not improve the ee, DTTA improved the ee by 4% only in
ꢀtartaric acid improved the ee by approximately 5% in all solvents except in
Lꢀtartaric acid did not dissolve in toluene. NMR revealed that the ratio of amine
Lꢀ(+)ꢀtartaric acid and Dꢀtartaric acid with good
recovery. While
acetonitrile, and
D
L
toluene since
and DTTA in the salt was always 1:1, regardless if 1 equiv. or 0.5 equiv. of acid was added.
Larger scale (2 g) experiments were carried out to compare these two acids and a variety of
solvent system (Table 3) with allylic amine 3 having an ee of 89%. Best results were achieved
with Lꢀ(+)ꢀtartaric acid using acetone/toluene as solvents, which was selected for further scaleꢀup
and manufacturing. The ratio of acetone and antiꢀsolvent toluene can be adjusted depending on
the ee of the crude product and the target ee of the final product.
ACS Paragon Plus Environment
20

Page 21 of 39
Organic Process Research & Development
1
2
3
4
Table 3
Further optimization of salt formation of 3
ee of 3 in ee of 3 in the
5
6
7
8
Entry
Acid
Solvent
Yield
the Salt
95.2%
Mother Liquor
14.2%
Acetone 10 mL
Toluene 10 mL
L
L
L
ꢀ(+)ꢀTartaric
9
1
2
3
4
93%
81%
93%
78%
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
acid
IPA 10 mL
Toluene 28 mL
ꢀ(+)ꢀTartaric
acid
98.0%
90.3%
93.6%
44.2%
35.0%
81.8%
ACN 10 mL
Toluene 12 mL
ꢀ(+)ꢀTartaric
acid
ACN 10 mL
Water 15 mL
Diꢀpꢀtoluoylꢀ
tartaric acid
Dꢀ
After this development, an inꢀhouse manufacturing campaign was successfully performed to
evaluate robustness at scale. The procedure was then transferred to a manufacturing facility that
produced approximately 48 kg of the tartaric acid salt. The results for this first generation
manufacturing of 3 are summarized in Scheme 10.9
ACS Paragon Plus Environment
21

Organic Process Research & Development
Page 22 of 39
1
2
3
4
Scheme 10 First generation route to 3 from dimethyl maleate 12
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Second Generation route to 3
Despite the process described above was successfully scaled up, we were not satisfied with the
complicated operations of the vinyl addition and the modest yield of the dimethylamine
displacement. With an adequate route to aldehyde 7 in hand, we explored the possibility of a
chiral sulfinamide auxiliary10ꢀ based approach (Scheme 11). Initial screen of both (R)ꢀ and (S)ꢀ2ꢀ
methylpropylsulfinamide (33) revealed that (S)ꢀ33 led to desired chirality. Condensation of 7
with (S)ꢀ33 in toluene in the presence of copper (II) sulfate at 40 °C proceeded smoothly to
produce the imine 34. The copper sulfate was removed first by filtration then by extraction using
aqueous citric acid. The toluene was switched to 2ꢀpropanol and 34 was then crystallized by the
addition of water. In the laboratory, yields of the condensation/crystallization sequence to
produce 34 were as high as 95%, but in the manufacturing facility yield dropped slightly to a still
ACS Paragon Plus Environment
22

Page 23 of 39
Organic Process Research & Development
1
2
3
4
5
6
satisfactory ~85% likely due to some oiling of the material. Improvements for this isolation are
ongoing.
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Scheme 11 Second generation route to 3 from 711
With the imine 34 in hand we extensively investigated the use of the commercially available
vinylmagnesium bromide and vinylmagnesium chloride for the nucleophilic addition of a vinyl
group to the imine 34 to provide the amine 35. Our results are summarized in Table 4.
Table 4 Initial Screening of the vinylation of 34
Solvent
(10V)
Temp
(°C)
dr of 35 (1Hꢀ
Entry
Nucleophile (2eq)
Additive
NMR)
VinylMgCl (1.6 M)
in THF
VinylMgBr (1.0 M)
in THF
VinylMgBr (1.0 M)
in THF
VinylMgBr (1.0 M)
in THF
1
2
3
4
DCM
DCM
ꢀ78
ꢀ78
ꢀ78
ꢀ78
ꢀ
ꢀ
ꢀ
ꢀ
5.4:1
9.6:1
2.8:1
4.7:1
Toluene
THF
ACS Paragon Plus Environment
23

Organic Process Research & Development
Page 24 of 39
1
2
3
4
5
6
7
8
9
VinylMgCl (1.6 M)
In THF
VinylMgBr (1.0 M)
in THF
VinylMgBr (1.0 M)
in THF
5
6
THF
CPME
DCM
ꢀ78
ꢀ78
ꢀ78
ꢀ78
ꢀ78
ꢀ78
2:1
ꢀ
3.1:1
1:2.6
15.5:1
6.5:1
14.7:1
7
TMEDA
AlMe3
AlMe3
AlMe3
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
VinylMgBr (1.0 M)
in THF
VinylMgBr (1.0 M)
in THF
8
DCM
9
Toluene
Toluene
VinylMgBr in
DCM
10
Although only relatively modest diastereoselectivity was achieved by the use of the
commercially available vinylmagnesium halides (Table 4, entries 1 to 6), we did use the best
conditions (Table 4, entry 2) successfully for a manufacturing campaign. Subsequent
deprotection and reductive amination provided crude 3 (vide infra). Our optimal conditions for
the tartaric acid salt formation of 3 produced a 63% yield with 94.2% ee. Attempts to obtain the
tartaric salt of 3 in higher yield invariably led to a lower than acceptable ee. After the first
manufacturing, further screens found that addition of AlMe3 (3 equiv.) could improve the
diastereoselectivity (Table 4, entries 8 to 10).
Scheme 12 Reduction product of 34 and resultant downstream product
Improving the diasteroselectivity of the vinyl addition was clearly critical to improving the
overall yield and the use of vinylmagnesium chloride was preferable due to the lower cost of this
ACS Paragon Plus Environment
24

Page 25 of 39
Organic Process Research & Development
1
2
3
4
5
6
7
8
9
reagent and its stability. The use of older bottles vinylmagnesium bromide (3ꢀ6 months old)
often led to the formation of the reduction product 37, which in turns led to the formation of a
downstream impurity 38 that was difficult to remove (Scheme 12). To date only traces of 37
have been formed using vinylmagnesium chloride (VMC), even when older lots (>6 months old)
have been used. The use of VMC THF solution (1.5 equiv.) in conjunction with additives in
different solvents was explored (Table 5).
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Table 5
Additive screening in vinylation of 34
dr of 35
Entry Solvent Temp (°C)
Additive (equiv.)
SM 34
37
(1HꢀNMR)
1
2
3
4
5
6
7
8
DCM
THF
DCM
THF
ꢀ48
ꢀ78
ꢀ78
ꢀ78
ꢀ78
ꢀ78
ꢀ78
0
ZnCl2 (1.1)
AlMe3 in toluene (1.7)
Dimethylzinc (1.7)
Dimethylzinc (1.7)
Diethylzinc (1.7)
MgBr2 etherate (1.1)
Ti(OEt)4 (1.1)
> 90%
1.2 %
nd
2:1
10:1
9:1
6.5:1
4:1
trace
trace
nd
40%
nd
nd
THF
10%
1.4%
64%
> 50%
DCM
DCM
THF
5:1
4:1
nd
nd
ZnCl2 (1.1)
nd: not detected
Encouraged by the promising dr obtained in entries 3 and 4, we studied literature reports using
mixed zincate ligands between vinylmagnesium halides and dimethyl zinc to affect nucleophilic
transfer with high diastereoselectivites.12 In addition to the pyrophoric nature of dimethylzinc,
the use of this reagent on scale in stoichiometric amounts was considered costꢀprohibitive. On
the other hand, methyllithium (MeLi) and ZnCl2 are relatively inexpensive, and we speculated
that dimethylzinc could be generated in situ by mixing these reagents in a 2:1 stoichiometry.
When a solution of ZnCl2 (1.5 equiv.) in diethylether or 2ꢀMeꢀTHF was mixed with MeLi in
diethoxymethane (3 equiv.) and VMC (1.5 equiv.) followed by the addition of a solution of 34 (1
ACS Paragon Plus Environment
25

Organic Process Research & Development
Page 26 of 39
1
2
3
4
5
6
equiv.) in THF at ꢀ78 °C, we were pleased to observe that 35 was formed in > 99% dr with 10%
residual 34 remaining (Table 6, entry 1).
7
8
9
Table 6
Entry
Optimization of Zinc catalyst used in the vinylation of 34
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
ZnCl2
equiv.
MeLi Vinyl MgCl LiCl
dr of 35
37
(%)
34
(%)
equiv.
equiv.
equiv.
(1HꢀNMR)
1.5 in
ether
1
2
3
4
5
6
7
3
1.5
0
0.3
0.2
0.64
0.2
ndb
1.23
nd
99.6:0.4
99.4:0.6
97.9:2.1
99.3:0.7
97.3:2.7
75:25
10
nd
1.5 in
ether
3
2.25
1
1.7 + 0.25
0
0
1.5 in
ether
2.25
1.5
1.5
1.5
2
36
0.5 in
ether
0
nd
0.1 in
ether
0.2
0
2.25
2.2
0
15.3
nd
0
0.275 in
MeTHF
0
90:10
nd
0.275
(solid)
8
0
2
3
nd
90:10
nd
9a
0.275 in
MeTHF
0.5
0.5
1.5
1.5
0
nd
99.8:0.2
99.0:1.0
nd
10
0.275 in
MeTHF
0
nd
nd
a
The solvent THF used in this reaction was intentionally spiked with 250 ppm water and 250
ppm BHT
b nd: not determined or not detected
The use of additional VMC to drive the reaction to completion did not lead to a significant loss
of dr (Entry 2). Lowering the ratio of MeLi to ZnCl2 did however lead to a slight drop in dr
(Entry 3). We also observed that when maintaining a 2:1 MeLi/ZnCl2 ratio, the amount of ZnCl2
used could be lowered to 25ꢀ50 mol% (Entries 4 and 9) without loss of dr. When only 10 mol %
ZnCl2 was used, it did lead to some dr erosion (Entry 5). Several other conditions were also
ACS Paragon Plus Environment
26

Page 27 of 39
Organic Process Research & Development
1
2
3
4
5
6
7
8
9
examined as control experiments. Omission of MeLi and ZnCl2, with the inclusion of LiCl
provided a modest 3:1 dr (Entry 6). Omission of MeLi and LiCl provided a 9:1 dr (entry 7). And
finally the omission of MeLi with the inclusion of LiCl also provided a 9:1 dr (entry 8). Previous
experiment (Table 5, entry 4) showed that dimethylzinc alone provided 9:1 dr. We concluded
that the methyl ligands on Zn as well as the presence of LiCl, both formed in situ, are critical for
achieving the highest diastereomeric ratios observed. These results are in accordance with the
observations of Hatano et al13 that the presence of LiCl is critical for the turnover of zinc in the
dialkylzinc catalyzed addition of organomagnesiums to carbonyl systems. A possible catalytic
cycle is depicted in Figure 3. The vinyl group is delivered via an in situ generated zincate which
is more reactive towards the imine compared to the vinyl magnesium chloride. In addition, the
imine is also activated by the lithium/magnesium complexation. Therefore, under the more
sensitive environment, the stereochemical effect from the chiral auxiliary group is amplified
leading to better stereoselectivity. Finally, the steric hindrance of the triorganozincate relative to
vinylmagnesium may contribute to the enhancement as well. It is worth noting that no
significant difference in diastereoselectivity was observed when the additional sequence was
changed (i.e., 34 was added to a mixture of MeLi, ZnCl2 and VMC vs. VMC was added to a
mixture of MeLi, ZnCl2 and 34). A significant advantage with this is that if the reaction is not
complete more VMC can be added. The highly diasteroselective vinylation of the imine 34 has
been implemented in a manufacturing setting with a high degree of success (typically 99:1 dr).
To the best of our knowledge, this is the first observation where the stereoselectivitiy of a
vinylation is significantly improved by dimethyl zinc and LiCl, particularly when only catalytic
amounts are required. In addition, compared to other alkyl zinc, the use of dimethyl zinc avoids
the risk of addition of the alkyl group to the imine (e.g., the addition of ethyl group to 7, vide
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
ACS Paragon Plus Environment
27

Organic Process Research & Development
Page 28 of 39
1
2
3
4
supra). And the in situ formation of dimethyl zinc and LiCl from ZnCl2 and MeLi provides a safe
5
6
and costꢀ effective approach for scale up.
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 3
Proposed catalytic cycle
Cleavage of the chiral auxiliary within 35 and reductive methylation to 3
The tꢀbutylsulfinyl group of 35 was smoothly cleaved to the primary amine 36 under a variety
of conditions. Our initial conditions used HCl/THF, where the primary amine was isolated as the
free base after extractive workup. The use of Eschweiler–Clarke conditions (HCO2H, H2CO, 65
°C) provided only modest yields. Initially, the use of sodium triacetoxyborohydride (STAB) was
explored, but in our hands was generally not efficient, needing ~5 equiv. of reducing reagent to
provide an acceptable conversion of 36 into 3. Additionally, on larger laboratory scales, thick
ACS Paragon Plus Environment
28

Page 29 of 39
Organic Process Research & Development
1
2
3
4
5
6
7
8
9
emulsions were observed during aqueous workups. We were intrigued by the use of other boron
based reducing reagents for this transformation and examined the pyridineꢀborane family as
these have recently gained attention as useful reagents for reductive aminations, particularly in
polar protic solvents including water and methanol.14 Attempted reductive aminations with
pyridineꢀborane produced 3 but also generated significant amounts of unidentified further
reduced products by LC/MS. The use of the commercially available 2ꢀmethylꢀ5ꢀethylꢀpyridineꢀ
borane complex provided an excellent chemical transformation, but unfortunately the removal of
the 2ꢀmethylꢀ5ꢀethylpyridine byꢀproduct proved to be problematic in downstream operations.
The use of 2ꢀpicolineꢀborane as a hydride source provided the same selective and highꢀyield
reaction as 2ꢀmethylꢀ5ꢀethylpyridine, with the added advantage that the byꢀproduct (2ꢀpicoline)
was easily removed by azeotropic distillation with toluene after workup to provide a solution of
crude 3 for the saltꢀforming step. Further optimization led to a streamlined process where the two
steps (cleavage of the chiral auxiliary of 35 and reductive methylation of 36) were executed in
one pot using methanol as solvent to produce 3 without any significant sideꢀproducts. Best
results were achieved when the pH of the reaction mixture is adjusted using NaOAc before the
addition of picolineꢀborane. With this optimized procedure, 80ꢀ90% yield from 34 and ~99% ee
were achieved including tartaric salt formation. We have observed a 2% to 3% drop of the ee
from 35 to crude 3 irrespective of the scale. Stress studies were performed to ensure this slight
loss of chiral purity would not be exacerbated with longer reaction time at larger scale.15
Conclusion: We have developed a robust and stereoselective process for the manufacture of 3
starting from readily available dimethyl maleate with an overall yield of up to 45%. This
approach was demonstrated in our kilo laboratory and in a manufacturing facility making ~ 450
kg of tartaric salt (see experimental section). Two scalable stereoselective vinylation methods
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
ACS Paragon Plus Environment
29