Toward a Scalable Synthesis and Process for EMA401, Part III: Using an Engineered Phenylalanine Ammonia Lyase Enzyme to Synthesize a Non-natural Phenylalanine Derivative

Article abstract of DOI:10.1021/acs.oprd.0c00217

A process using an engineered phenylalanine ammonia lyase (PAL) enzyme was developed as part of an alternative route to a key intermediate of olodanrigan (EMA401). In the first part of this report, the detailed results from a screening for the optimal reaction conditions are presented, followed by a discussion of several workup strategies investigated. In the PAL-catalyzed reaction, 70-80% conversion of a cinnamic acid derivative to the corresponding phenylalanine derivative could be achieved. The phenylalanine derivative was subsequently telescoped to a Pictet-Spengler reaction with formaldehyde, and the corresponding tetrahydroisoquinoline derivative was isolated in 60-70% yield with >99.9:0.1 er. On the basis of our screenings, carbonate/carbamate-buffered ammonia at an NH3 concentration of 9-10 M and pH 9.5-10.5 was found to be optimal. Enzyme loadings down to 2.5 wt % (E:S = 1:40 w/w) could be achieved, and substrate concentrations between 3-9 v/w (1.17-0.39 M) were found to be compatible with the reaction conditions. A temperature gradient was applied in the final process: A pre-equilibrium was established at 45 °C, before making use of the temperature dependence of the entropy term with subsequent cooling to 20 °C to achieve maximum conversion. This temperature gradient also allowed balancing of the enzyme stability (low at 45 °C, high at 20 °C) with the activity (high at 45 °C, low at 20 °C) in order to achieve optimal conversion (low at 45 °C, high at 20 °C). From the various workup operations investigated, a sequence consisting of denaturation of the enzyme, NH3/CO2 removal by distillation, acidification, and telescoping to the subsequent Pictet-Spengler cyclization was our preferred approach. The process presented in this study is a more sustainable, shorter, and more cost-effective alternative to the previous process.

Full text of DOI:10.1021/acs.oprd.0c00217

Subscriber access provided by CORNELL UNIVERSITY LIBRARY
Communication
Towards a scalable synthesis and process for EMA401.
Part III: Using an engineered phenylalanine ammonia lyase
enzyme to synthesize a non-natural phenylalanine derivative
Leo Albert Hardegger, Pascal Beney, Dominique Bixel, Christian Fleury, Feng Gao, Alexandre
Grand-Guillaume Perrenoud, Xingxian Gu, Julien Haber, Tao Hong, Roger Humair, Andreas
Kaegi, Michael Kibiger, Florian Kleinbeck, Van Tong Luu, Lukas Padeste, Florian A. Rampf,
Thomas Ruch, Thierry Schlama, Eric Sidler, Anikó Udvarhelyi, Bernhard Wietfeld, and Yao Yang
Org. Process Res. Dev., Just Accepted Manuscript • DOI: 10.1021/acs.oprd.0c00217 • Publication Date (Web): 20 Aug 2020
Downloaded from pubs.acs.org on August 21, 2020
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 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 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.
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 45
Organic Process Research & Development
1
2
3
4
5
6
7
8
9
Towards a scalable synthesis and process for
EMA401. Part III: Using an engineered
phenylalanine ammonia lyase enzyme to synthesize
a non-natural phenylalanine derivative
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
Leo A. Hardegger†*, Pascal Beney†, Dominique Bixel†, Christian Fleury†, Feng Gao‡,
Alexandre Grand-Guillaume Perrenoud†, Xingxian Gu‡, Julien Haber†, Tao Hong‡,
Roger Humair†, Andreas Kaegi†, Michael Kibiger†, Florian Kleinbeck†, Van Tong Luu†,
Lukas Padeste†, Florian A. Rampf†, Thomas Ruch†, Thierry Schlama†, Eric Sidler†≠,
Anikó Udvarhelyi¥, Bernhard Wietfeld†, Yao Yang‡
In memoriam Michael Kibiger
†Chemical and Analytical Development, Novartis Pharma AG, 4002 Basel, Switzerland.
¥Pharmaceutical and Analytical Development, Novartis Pharma AG, 4002 Basel,
Switzerland.
ACS Paragon Plus Environment
1

Organic Process Research & Development
Page 2 of 45
1
2
3
4
5
‡Suzhou Novartis Technical Development Co., Ltd., Changshu City, 215537, China
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
For table of contents only
ACS Paragon Plus Environment
2

Page 3 of 45
Organic Process Research & Development
1
2
3
4
5
6
7
8
9
ABSTRACT A process using engineered phenylalanine ammonia lyase (PAL)
enzymes was developed as part of an alternative route to a key intermediate of
olodanrigan (EMA401). In the first part of the manuscript, the detailed results from a
screening for the optimal reaction conditions are presented, followed by the discussion
of several work-up strategies investigated. In the PAL catalyzed reaction, 70–80%
conversion of a cinnamic acid derivative to the corresponding phenylalanine derivative
could be achieved. The phenylalanine derivative was subsequently telescoped to a
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
Pictet-Spengler
reaction
with
formaldehyde
and
the
corresponding
tetrahydroisoquinoline derivative was isolated in 60–70% yield with >99.9:0.1 er. Based
on our screenings, carbonate/carbamate buffered ammonia at 9–10 M NH3
concentration and pH 9.5–10.5 were found as the optimal conditions. Enzyme loadings
down to 2.5wt% (E:S 1:40 w/w) could be achieved and substrate concentrations
between 3–9 v/w (1.17–0.39 M) were found to be compatible with the reaction
conditions. A temperature gradient was applied in the final process: a pre-equilibrium
was established at 45 °C, before making use of the temperature-dependence of the
entropy term with subsequent cooling to 20 °C and achieving maximum conversion.
ACS Paragon Plus Environment
3

Organic Process Research & Development
Page 4 of 45
1
2
3
4
5
6
7
8
9
This temperature gradient also allowed balancing enzyme stability (low at 45 °C, high at
20 °C) with activity (high at 45 °C, low at 20 °C) in order to achieve optimal conversion
(low at 45 °C, high at 20 °C). From the various work-up operations investigated, a
sequence consisting of denaturation of the enzyme, followed by NH3/CO2 removal by
distillation, acidification and telescoping to the subsequent Pictet-Spengler cyclization
was our preferred approach. The process presented in this study is a more sustainable,
shorter and more cost effective alternative to the previous process.
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
KEYWORDS. EMA401, phenylalanine ammonia lyase, scale-up, process development,
biocatalysis, phenylalanine derivative
Introduction
In a route scouting exercise for the synthesis of EMA401 (1, olodanrigan), an
angiotensin II type 2 antagonist used for the treatment of postherpetic neuralgia and
neuropathic pain, the synthetic approach via the hydroamination of cinnamic acid 2 to
amino acid 3 as the key step was identified as an attractive alternative to the existing
chemocatalytic route (Scheme 1).1 The chemocatalytic route relied on an asymmetric
ACS Paragon Plus Environment
4

Page 5 of 45
Organic Process Research & Development
1
2
3
4
5
6
7
8
9
hydrogenation step catalyzed by a chiral rhodium complex. The common late stage
intermediate for both, the biocatalytic and chemocatalytic, routes is
tetrahydroisoquinoline 4, while the common starting material for both routes is
benzylated ortho-vanillin 5. For the preparation of amino acid 3 via the chemocatalytic
route, aldehyde 5 was treated with Horner-Wadsworth-Emmons reagent 6 to obtain
enamide 7, which was subjected to asymmetric hydrogenation conditions to afford
amide 8, followed by functional group interconversion to carbamate 9 and finally
deprotection to amino acid 3. The synthetic approach using the biocatalytic route
described in this publication follows a synthetic route requiring only two stages to the
key intermediate phenylalanine 3 starting from benzylated ortho-vanillin 5, which was
reacted to cinnamic acid 2 and subsequently subjected to the PAL-catalyzed
hydroamination to amino acid 3.2
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
5

Organic Process Research & Development
Page 6 of 45
1
2
3
This publication: biocatalytic route
4
5
O
O
PAL
6
7
8
NH3
CH2O
HO
OH
O
O
O
O
O
O
O
O
O
O
O
OH
OH
OH
O
(S)
(S)
NH2
NH
9
5
2
3
4
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
O
P
MeO
HN
OMe
O
HCl
chemocatalytic route
O
6
O
Na
O
O
1) Boc2O
2) NaOH
O
O
O
H2, RhL*
O
O
O
O
O
O
O
O
(S)
OH
O
O
O
O
O
O
O
N
HN
HN
HN
7
8
9
1
Scheme 1. Comparison of two synthetic routes to EMA401 (1) via the key intermediate
amino acid 3 and tetrahydroisoquinoline 4, either following the biocatalytic route (red) or
the chemocatalytic route (black).
The key step of the biocatalytic route, the hydroamination to amino acid 3, is
catalyzed by a phenylalanine ammonia lyase (PAL, EC 4.3.1.24), an enzyme class
which has attracted a lot of attention during the last decade in the academic and
industrial community.3,4 The advantages of the biocatalytic route are obvious: It is three
steps shorter than the chemocatalytic route, uses readily available reagents, avoids the
use of heavy metals, and is environmentally more sustainable.
ACS Paragon Plus Environment
6

Page 7 of 45
Organic Process Research & Development
1
2
3
4
5
6
7
8
9
PALs have been shown to catalyze the hydroamination reaction of various non-natural
substrates,5–14 and some examples for successfully scaled-up processes have been
reported.15,16 Two different mechanisms for the 4-methylideneimidazole-5-one (MIO)-
dependent hydroamination with PALs have been proposed and appear to be
operational depending on the electronic nature of the substrate.6,8,14,17–19 Our own
calculations of Gibbs free energies G in the gas phase between cinnamic acid
derivatives and amino acid derivatives indicated a strong correlation between the
electronic nature of the aromatic ring and the equilibrium conversion, with smaller G
values predicted for electron-poor systems, which corresponds to an equilibrium
position favoring product formation (Figure 1).20–22 A similar trend was found when
running the calculations simulating an aqueous environment (see Supporting
Information (SI)). Our calculations indicated that both the enthalpy and entropy term
were negative. Hence, the Gibbs free energy became negative at low temperatures
only. In order to shift the equilibrium as much as possible on the amino acid product
side, we aimed at applying a temperature gradient to lower temperatures towards the
end of the reaction and applying Le Chatelier’s principle by decreasing S (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
ACS Paragon Plus Environment
7

Organic Process Research & Development
Page 8 of 45
1
2
3
4
5
6
7
8
9
O
O
OH
OH
NH2
X
X
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. Correlation of calculated Gibbs free energy G for the equilibrium between
cinnamic acid derivatives and amino acid derivatives with the Hammett constant (p) in
the gas phase. The G values of electron-poor aromatic systems are smaller than those
of electron-rich aromatic systems (see SI for more discussion on G).
While whole cells were used in most of the previously published cases, we aimed at
employing the isolated, lyophilized enzyme powders as the catalyst. Even though no
conversion of cinnamic acid 2 could be detected in initial screening experiments with
wild type PAL enzymes, presumably due to the bulky and electron-rich substituents on
the substrate, we were convinced that the PAL-catalyzed route was the best option for
ACS Paragon Plus Environment
8

Page 9 of 45
Organic Process Research & Development
1
2
3
4
5
6
7
8
9
the synthesis of EMA401 (1) and initiated an extensive enzyme engineering program for
the directed evolution of an enzyme starting from an engineered variant from anabaena
variabilis specifically for the conversion of cinnamic acid 2 to amino acid 3.23 Enyzme
engineering was performed at Codexis using the CodeEvolver® technology. After
several rounds of enzyme evolution and more than 20 amino acid mutations from the
starting point of evolution, engineered enzyme variants were obtained which could
successfully be used in lab experiments and scale-up (vide infra, see Section 8 in the
SI).
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
Results and discussion
Screening of reaction parameters
For the preparation of amino acid 3, we initiated process development by screening the
most important parameters (ammonia concentration and counter ion of ammonia buffer
system, pH, substrate concentration, temperature profile of the reaction, enzyme
loading and stability, and the use of co-solvents or additives), while enzyme evolution
ACS Paragon Plus Environment
9

Organic Process Research & Development
Page 10 of 45
1
2
3
4
5
6
7
8
9
was still ongoing in parallel. Most of the parameters screened were not strongly
influenced by enzyme evolution, with the obvious exception of the enzyme loading,
enzyme stability, and the use of co-solvents.
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
(CH2O)n
10 M NH3, PAL
H2O, pH 10
45 -> 20 °C
1-3 d
O
O
O
O
O
O
50 °C
16 h
60-70% (2 steps)
O
O
O
OH
OH
OH
NH2
NH
2
3
4
O
NH2
O
OH
O
O
O
OH
OH
NH
10
11
Scheme 2. Details of the synthetic route from cinnamic acid 2 to amino acid 3 and
tetrahydroisoquinoline 4. -Amino acid 10 was observed as a by-product in the reaction
to -amino acid 3. As a result of benzyl cleavage under the strongly acidic conditions of
the Pictet-Spengler cyclization to tetrahydroisoquinoline 4, phenol 11 was observed as
the major impurity.
ACS Paragon Plus Environment
10

Page 11 of 45
Organic Process Research & Development
1
2
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
b)
a)
c)
d)
e)
f)
Figure 2. a) Dependence of conversion to amino acid 3 on the ammonia concentration;
conditions: pH 9, 20 v/w, 40 °C, >40 h reaction time. Higher ammonia concentrations
led to higher conversions; b) dependence of conversion on pH: the optimal pH range for
ACS Paragon Plus Environment
11

Organic Process Research & Development
Page 12 of 45
1
2
3
4
5
6
7
8
9
the conversion to amino acid 3 was found to be 9.0–10.0; conditions: 5wt% enzyme
PAL-130, 40 °C, 10 v/w, >40 h reaction time; c) influence of the substrate concentration
on the conversion to amino acid 3 obtained in screenings; conditions: 5wt% enzyme
PAL-130, 40 °C, pH 10, 10.7 M NH3, >40 h reaction time; d) conversion to amino acid 3
observed as a function of the substrate concentration in experiments on more than 10 g
scale with at least 60 h total reaction time. The result from scale up on 2 kg of cinnamic
acid 2 is highlighted in red; e) temperature-dependent conversion to amino acid 3, a
result of the temperature dependent entropy term; conditions: 5wt% enzyme PAL-130,
10 v/w, pH 9, 10 M NH3, >40 h reaction time; f) temperature gradient employed during a
standard PAL reaction (blue line), and respective conversion to amino acid 3 of a
selected experiment with 2.5wt% PAL-131 enzyme (green line).
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
Ammonia concentration and pH: As expected for reactions driven against the
thermodynamically favored direction, an excess of ammonia led to higher conversion of
cinnamic acid 2 (Le Chatelier’s principle). Our screenings confirmed that the higher the
ACS Paragon Plus Environment
12

Page 13 of 45
Organic Process Research & Development
1
2
3
4
5
6
7
8
9
ammonia concentration, the higher the conversion to amino acid 3 (Figure 2a).22,24–28
The ammonia concentration was, however, limited by the solubility of the buffer system
required due to the reduced stability of the enzyme above pH 10.5. At pH values below
8.5 (Figure 2b), little to no conversion was observed, presumably either due to the lack
of free ammonia present in solution, which is required for the attack on the MIO group to
initiate the catalytic cycle, or due to reduced solubility of the enzyme. As a result of our
screening efforts, we selected a concentration of 10 M NH3 at a pH = 10±0.5. The pH
was found to be stable over the entire course of the reaction. Dosing of additional
ammonia at different time points to replace the ammonia consumed by the formation of
amino acid 3 did not have a significant influence on the equilibrium conversion (Figure
SI3).
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
Ammonia source: In agreement with reports in the literature, carbonates and
carbamates were found to be the optimal counter ions for the preparation of the buffer
system used in the conversion of cinnamic acid 2 to amino acid 3 (Scheme 2).22,28,29
The ammonia buffer with carbonates/carbamates could be prepared either by
ACS Paragon Plus Environment
13

Organic Process Research & Development
Page 14 of 45
1
2
3
4
5
6
7
8
9
dissolution of ammonium carbamate in water, by the addition of gaseous CO2 to
aqueous ammonia, or by the addition of aqueous ammonia to ammonium carbonate.
While starting from ammonium carbamate was the simplest option in the lab for buffer
preparation, its price and the low decomposition temperature of 35 °C did not make this
the ideal reagent for handling, transport and scale-up. The addition of CO2 to aqueous
ammonia was a time-consuming process on scale, and preventing off-gassing during
the addition of CO2 to maintain full control over the composition of the buffered
ammonia solution proved to be difficult. The preferred protocol consisted in the
preparation of a suspension of ammonium carbonate in water and subsequent addition
of concentrated aqueous ammonia to afford a solution at the desired concentration and
pH. According to Raman studies found in the literature, an equilibrium of the
carbonate/carbamate counter ion is established rapidly irrespective of the reagents
used for its preparation.30,31 Indeed, the results of the reaction were found to be
independent of the method used for the preparation of the buffered ammonia solution.
As a result of sublimation, ammonium salts were observed as a white precipitate at cold
parts in the headspace of the reactors in lab reactions. Alternative counter ions tested
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
14

Page 15 of 45
Organic Process Research & Development
1
2
3
4
5
6
7
8
9
delivered lower conversion as compared to the carbonate/carbamate system. We
reasoned that a weak counter ion was required to maintain enzyme stability. The use of
unbuffered, diluted ammonia was tested with the penultimate PAL enzyme generation
PAL-130 at a loading of 50wt% enzyme, providing a satisfactory conversion of 68% at
25 °C, pH 12.6 and 6.3 M ammonia concentration (Table SI3). The conversion dropped
to 19% when the enzyme loading was reduced to 5wt% with respect to the cinnamic
acid substrate 2 in unbuffered conditions. We hypothesized that the enzyme stability
was reduced at higher pH as shown in Figure 2b for buffered systems.
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
Substrate concentration: Screenings proved that satisfactory conversion as well as
acceptable stirring properties could be achieved when reactions were concentrated to 3
v/w. It is important to note that cinnamic acid 2 in our case did not dissolve in the
ammonia buffer, but formed an emulsion upon addition of the buffered ammonia
solution. This behavior was observed at all dilutions of interest for commercial
production and did not appear to have a detrimental influence on the outcome of the
reaction. The conversion was found to be proportional to the dilution of the substrate in
ACS Paragon Plus Environment
15

Organic Process Research & Development
Page 16 of 45
1
2
3
4
5
6
7
8
the range of 3–9 v/w, corresponding to 1.17–0.39 M, as evidenced in screenings (Figure
2c) and reactions with more than 10 g substrate (Figure 2d).24,27–29
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
Temperature profile: The desired reaction was thermodynamically not favored (vide
supra). We used the temperature dependence of the entropy term S, as confirmed in
screenings at different temperatures (Figure 2e), following investigations of the reaction
kinetics and applying a temperature gradient as shown in Figure 2f.32 The mixture was
stirred at 45 °C for 8 h, to establish a pre-equilibrium (criterion: less than 2mol% change
in conversion per hour). Subsequently the mixture was cooled to 20 °C over 8 h. Kinetic
modelling indicated that the nature of the cooling ramp does not have a major influence
on the equilibrium established at 20 °C after 24 h reaction time, with a conversion of 65–
70% achieved in 9 v/w dilution. When applying this gradient, higher enzyme activity and
lower enzyme stability at 45 °C was balanced with lower enzyme activity and higher
enzyme stability at 20 °C.
Enzyme loading: The enzyme loading could be significantly optimized throughout the
enzyme evolution. For the latest enzyme generation PAL-131, 2.5wt% with respect to
ACS Paragon Plus Environment
16

Page 17 of 45
Organic Process Research & Development
1
2
3
4
5
6
7
8
9
the substrate cinnamic acid 2 was used in standard experiments to achieve more than
60% conversion after 8 h at 45 °C. Similar results were obtained for PAL-130, the
enzyme generation mainly used for development, at 5wt% loading. Screening
experiments showed that conversion deteriorated at loadings of less than 3wt% for
PAL-130 (Table SI4). When increasing the enzyme loading to 25wt%, a conversion of
up to 68% amino acid 3 could be achieved within 2 h at 40 °C (Table SI4). For simplified
enzyme handling and charging in the scale-up, the lyophilized enzyme powders were
charged as a colloidal solution in a mixture of the buffered ammonia solution and water.
The engineered PAL enzymes proved to be stable for several hours when stored in
solution at room temperature. Expectedly, the reduction of the enzyme loading with
every new enzyme generation led to faster clear filtrations for the enzyme removal.
Clear filtrations could also be expedited with some of the co-solvents tested, for
example DMSO. First attempts of immobilization of the PAL enzymes were in our case
only successful with epoxy-substituted beads, which led to a significant reduction of the
enzyme activity and processing issues with the epoxy beads due to long wetting times.
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
17

Organic Process Research & Development
Page 18 of 45
1
2
3
4
5
6
7
8
9
Co-solvents: An extensive screening of co-solvents, surfactants and additives did not
indicate any hit with a significant influence on the equilibrium conversion to amino acid
3. Some of the solvents tested, however, were found to increase the initial reaction rate.
While for early generations of the engineered PAL enzymes, DMSO or glycols were
required to achieve good conversion,25–27 co-solvents were no longer necessary for the
latest generations, PAL-130 and PAL-131.
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
Impurities: A streamlined manufacturing process for the manufacture afforded the
substrate cinnamic acid 2 with around 4–5mol% morpholine content.2 Use-tests and
screenings showed that morpholine in cinnamic acid 2 was not critical, and up to
10mol% morpholine did not have an impact on the conversion (Table SI5). The only by-
product observed by HPLC in the PAL-catalyzed conversion of cinnamic acid 2 to -
amino acid 3 was the corresponding -amino acid 10 in amounts below 0.5a% (Scheme
2). As the formation of the -amino acid 10 by a background 1,4-addition was not
observed when the substrate was subjected to the standard reaction conditions in the
ACS Paragon Plus Environment
18

Page 19 of 45
Organic Process Research & Development
1
2
3
4
5
absence of an enzyme, the impurity is potentially generated via a phenylalanine amino
6
7
8
mutase (PAM) pathway catalyzed by the PAL enzyme.18,21
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
In summary, the screening results of the key parameters for the PAL-catalyzed
hydroamination of cinnamic acid 2 to -amino acid 3 resulted in the following conditions
for the standard PAL-catalyzed reaction: The reaction was run in 9–10 M buffered
ammonia solution at pH 10±0.5, at a dilution of 9 v/w and with 5wt% (PAL-130) or
2.5wt% (PAL-131) enzyme loading for 24–96 h, applying the temperature gradient in
Figure 2f. Under these conditions, an equilibrium conversion of 70–80% was achieved
on lab scale.33 The conditions were then successfully scaled up to a batch size of 2 kg
of cinnamic acid 2, resulting in 81% conversion to amino acid 3 after 71 h with 100 g of
enzyme PAL-130.
Work-up of the PAL-catalyzed reaction
ACS Paragon Plus Environment
19

Organic Process Research & Development
Page 20 of 45
1
2
3
4
5
6
7
8
9
With the reaction conditions established, we focused our attention on different work-up
strategies. Applying the above conditions, approximately 1–1.5 kg of ammonia had to be
used for each kg of substrate 2; and 20–30% of the substrate 2 could not be converted
to the desired product. For a multi ton product such as EMA401 (1), this would have
resulted in significant amounts of ammonia and cinnamic acid 2 waste streams. During
design of the work-up of the PAL reaction, we therefore evaluated the generated waste
streams and thoroughly assessed the potential for recycling of ammonia, cinnamic acid
2 and enzyme. Some of the work-up strategies are summarized in Figure 3 and
described in more detail 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
ACS Paragon Plus Environment
20

Page 21 of 45
Organic Process Research & Development
1
2
3
Reaction mixture
Work-up Options
4
5
Distillation of NH3/CO2
3
Amino acid
H2
Addition of MeOH to
denaturate enzyme
Distillation of NH3
CO2 and MeOH
,
Filtration to remove
PAL enzyme and 2
amino acid 3 collected
in mother liquor
pH adjusted to <1.2
O
6
7
PAL enzyme
Substrate 2
NH3, CO2
H2
O
Recycling of NH3 and
CO2 from distillate
Recycling of substrate
2
from filter cake
8
9
Quench with HCl
3
Amino acid
Addition of
aqueous HCl
Clear filtration to
remove PAL enzyme
Azeotropic drying and
addition of heptane
Amino acid 3 isolated
as HCl salt
O
O
Addition of 2-MeTHF
Phase separation
NH3
O
OH
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
CO2
PAL enzyme
Substrate 2
3
Amino acid from
mother liquor /
aqueous acidic layer
telescoped to
Release of CO2 in
acidic quench
Denaturated enzyme
cannot be recycled
Recycling of substrate
2
Substrate
NH4Cl generated
2
from mother liquor
Pictet-Spengler cyclization
Quench with H2SO4
3
Amino acid
NH3
Addition of
aqueous H2SO4
Clear filtration to
remove PAL enzyme
amino acid 3 collected
in aqueous layer
Addition of toluene
Addition of TBME
Phase separation
O
O
CO2
Substrate
2
PAL enzyme
O
H2O
OH
NH2
Realease of CO2 in
acidic quench
Denaturated enzyme
cannot be recycled
Recycling of substrate
2
from organic layer
3
Amino acid
Extractions from ammonia layer
2
Amino acid 3
amino acid 3 collected
Substrate
PAL enzyme
NH3
Addition of nBuOH/
xylenes 7:3 (w/w)
Clear filtration to
remove PAL enzyme
Distillation of nBuOH;
Addition of H2SO4
Phase separation
Phase separation
in aqueous layer
3
Amino acid
2
Substrate
PAL enzyme
NH3, CO2
H2O
H2
O
Denaturated enzyme
cannot be recycled
Re-cycled ammonia
shows conversion
Recycling of substrate
CO2
2
from organic layer
Tangential flow filtration of the enzyme
3
Amino acid
2
H2
O
Substrate
Tangential flow
filtration
Distillation of NH3
CO2 and MeOH
,
Filtration to remove
substrate 2
amino acid 3 collected
in mother liquor
Legend:
pH adjusted to <1.2
Full re-cycling possible
Partial re-cycling possible
No re-cycling possible
H2O
3
NH3, CO2
Amino acid
PAL enzyme
Substrate 2
NH3, CO2
H2O
Re-cycling of enzyme
possible
Recycling of NH3 and
CO2 from distillate
Recycling of substrate
2
from filter cake
Figure 3. Work-up strategies investigated for the PAL-catalyzed reaction. For details,
see the respective sections of the manuscript.
Distillation of NH3/CO2: Distillation allowed for the recycling of ammonia and CO2 by
condensation into a second reactor for the next batch. As the hydroamination reaction is
reversible, it was of obvious importance to remove or denaturate the enzyme before
distillation to avoid the back reaction. As mentioned above, the use of immobilized
enzymes as a straightforward way to remove the enzyme from the reaction mixture prior
to distillation was not yet practical in our case. We thus evaluated enzyme denaturation
ACS Paragon Plus Environment
21

Organic Process Research & Development
Page 22 of 45
1
2
3
4
5
6
7
8
9
by heat, taking into consideration that the chiral stability of amino acid 3 had previously
been found to deteriorate at temperatures above 70 °C under the reaction conditions.
As an additional challenge, the stability of the enzyme had been concomitantly
improved over the course of the extensive protein engineering, and we found that full
denaturation could not be ensured at temperatures below 70 °C, posing the risk of a
back reaction upon distillation of ammonia. We therefore investigated the suitability of
denaturating via chaotropic reagents. For a first scale-up, we selected MeOH to
denaturate the PAL enzymes, as MeOH could be removed in the subsequent
distillations. Experiments had shown that the PAL-130 enzyme retained activity after
addition of 20v% of MeOH, and 50v% MeOH was required to ensure full denaturation of
the enzyme (Figure SI4). Even though this approach did not allow for the recycling of
ammonia due to contamination with MeOH, denaturation with MeOH was found to be
acceptable as a temporary solution, in line with the phase-dependent development
targets. Under the applied reaction conditions, the reaction mixture was stirred for 1 h
after addition of MeOH to ensure close to full denaturation of the enzyme, followed by
subsequent removal of ammonia, CO2 and MeOH by distillation. The resulting
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
22

Page 23 of 45
Organic Process Research & Development
1
2
3
4
5
6
7
8
9
suspension was then acidified with aqueous H2SO4 to pH below 1.2 at 40 °C. To ensure
stirrability of the resulting suspension thus obtained, an anchor stirrer was required. The
suspension was particularly thick around pH 5.9, corresponding to the iso-electric point
of amino acid 3, and pH 4.2, corresponding to the pKa of cinnamic acid 2 (Figure 4). At
lower pH values, amino acid 3 started to dissolve again, while the enzyme and
substrate 2 were suspended and removed via clear filtration. The use of filter aids led to
significant losses of amino acid 3 in the clear filtration. Collection of the substrate in the
filter cake gave the potential for recycling with a simple process. Residual amino acid 3
in the filter cake was extracted by re-slurrying with 1 M aqueous H2SO4. The mother and
wash liquors were combined and subsequently telescoped to the Pictet-Spengler
cyclization to tetrahydroisoquinoline 4 (vide infra). This work-up procedure was the
preferred option for the scale-up and could be implemented successfully at a batch size
of 2 kg of cinnamic acid 2. Distillations ran smoothly without observation of major
sublimation of ammonium salts. An Apovac pump was used in the distillations to trap
ammonia. An agitated filter drier with a diameter of 20 cm was used to re-slurry the filter
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
23

Organic Process Research & Development
Page 24 of 45
1
2
3
4
5
6
7
8
cake, leading to filtration times of 2 h for the clear filtration and 1 h each for the
filtrations after re-slurry.
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
O
O
O
O
O
O
O
OH
2
pKa = 4.2
O
OH
NH2
3
pKa = 9.1, 2.5
Isoelectric point = 5.9
O
OH
NH
4
pKa = 8.9, 1.9
Isoelectric point = 5.4
Figure 4. pH-dependent solubility of cinnamic acid 2, amino acid 3 and
tetrahydroisoquinoline 4 in water determined with a Sirius T3 instrument.
Quench with acids: Very high or very low pH is known to denaturate many enzymes. In
the PAL-catalyzed hydroamination reaction, the addition of NaOH led to the formation of
by-products and was thus abandoned. When acids were added, full denaturation of the
PAL enzyme could be ensured, however at the cost of a strong exotherm, the violent
ACS Paragon Plus Environment
24

Page 25 of 45
Organic Process Research & Development
1
2
3
4
5
6
7
8
9
release of large amounts of CO2 and hence concomitant foaming. In addition, recycling
of the thus formed ammonium salts was not as straightforward as compared to the
distillation of the ammonium carbonate/carbamate system.
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
When the reaction mixture was quenched with concentrated aqueous HCl, 2-MeTHF
was added before the quench to suppress foaming during the addition of HCl, and to
extract the resulting hydrochloride salt of amino acid 3 along with cinnamic acid 2 into
the organic layer. Though low, the solubility of the hydrochloride salt of amino acid 3
exceeded the solubility of the zwitterion, and large amounts of 2-MeTHF were required
for the extraction. A clear filtration to remove the enzyme buffer layer prior to phase
separation was not implemented due to major losses of amino acid 3 in the filter cake,
which could only be recovered with larger amounts of 2-MeTHF. The enzyme buffer
layer made the phase separations tedious, leading to the presence of residual enzyme
in the organic layer and the need for clear filtration of the organic layer. The
hydrochloride salt of amino acid 3 was isolated from the organic layer after azeotropic
ACS Paragon Plus Environment
25

Organic Process Research & Development
Page 26 of 45
1
2
3
4
5
distillation and addition of heptane. Cinnamic acid 2 was collected in the mother liquor
6
7
8
and could be isolated by concentration.
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
In an alternative process, the buffered ammonia solution obtained at the end of the
reaction was quenched with aqueous H2SO4. In contrast to the quench with aqueous
HCl, we aimed at retaining amino acid 3 in the aqueous acidic layer. This approach
allowed to extract cinnamic acid 2 in the organic layer and perform a clear filtration
before the phase separation. Though the quench was again strongly exothermic, heat
generation and CO2 release occurred dose-controlled. An organic solvent had to be
added to suppress foaming. The solvent of choice was toluene, as other solvents (for
example TBME) were either not compatible with the process conditions, or led to
racemization in the subsequent step to tetrahydroisoquinoline 4 (for example THF and
2-MeTHF).34 After the acidic quench, TBME was added to the toluene layer to solubilize
cinnamic acid 2 in the organic layer. A clear filtration allowing the removal of the
enzyme was implemented, followed by a phase separation . This approach allowed
extraction of amino acid 3 into the aqueous acidic layer, while cinnamic acid 2 could be
ACS Paragon Plus Environment
26

Page 27 of 45
Organic Process Research & Development
1
2
3
4
5
6
7
8
9
recycled from the organic layer. In addition, the ammonium sulfate generated during the
quench ensured complete denaturation of the enzyme. The aqueous layer was
subsequently treated with para-formaldehyde, and amino acid 3 was telescoped to the
subsequent transformation to tetrahydroisoquinoline 4 (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
Extractions from ammonia buffer: As the quench with acids did not allow for the simple
recycling of ammonia, we also investigated an approach relying on extraction of the
basic ammonia buffer, potentially even providing a handle to further drive the
equilibrium between cinnamic acid 2 and amino acid 3. A first round of solvent
screening indicated that alcoholic solvents were capable of extracting amino acid 3 from
the reaction mixture. In a subsequent screening of more than 20 alcohols or solvent
mixtures with alcohols, a mixture of nBuOH/xylenes 7:3 (w/w) was chosen to extract
amino acid 3 from the ammonia buffer, along with cinnamic acid 2. To our dismay, none
of the solvents or solvent mixtures tested favored the extraction of amino acid 3 over the
extraction of cinnamic acid 2 from the buffered ammonia layer. Addition of
nBuOH/xylenes 7:3 (w/w) to the reaction mixture resulted in a biphasic mixture with an
ACS Paragon Plus Environment
27

Organic Process Research & Development
Page 28 of 45
1
2
3
4
5
6
7
8
9
enzyme buffer layer, which was removed by clear filtration. The organic layer containing
the cinnamic acid 2 and amino acid 3 was then separated. Attempts to run a continuous
extraction with a perforator were stopped as precipitation at the inlet of the organic
solvent into the aqueous layer led to clogging. Using nBuOH/xylenes led to the co-
extraction of significant amounts of the aqueous ammonia buffer into the organic layer,
and titrations indicated that the concentration of the aqueous ammonia buffer dropped
to only 3.4 M NH3 when starting from 10 M NH3. Hence, recycling of ammonia proved
difficult with this work-up strategy. The ammonia buffer thus recovered was
nevertheless treated with substrate 2 and enzyme PAL-130, leading to 28% conversion
in 27 h (Figure SI5). The organic nBuOH/xylenes phase was concentrated by distillation
to remove residual water, ammonia and nBuOH, resulting in a suspension of amino acid
3 and cinnamic acid 2 in xylenes. Aqueous H2SO4 was added to dissolve amino acid 3
while cinnamic acid 2 remained dissolved in xylenes at higher temperature. This
allowed re-cycling of cinnamic acid 2 from xylenes after phase separation. The aqueous
acidic layer containing amino acid 3 was telescoped to the subsequent Pictet-Spengler
cyclization to tetrahydroisoquinoline derivative 4 (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
ACS Paragon Plus Environment
28

Page 29 of 45
Organic Process Research & Development
1
2
3
4
5
6
7
8
9
Tangential flow filtration of the enzyme: In parallel to the traditional work-up strategies
presented in this study, implementation of an ultrafiltration was investigated to separate
the enzyme from the small molecules cinnamic acid 2 and amino acid 3, taking
advantage of the significant size difference between the enzyme (61.5 kDa/subunit of
the tetramer), product and residual substrate. The main drivers for the investigations
were a potential cost reduction, achieved by recycling of the enzyme, process
intensification as a result of increased enzyme loading (vide supra) and simplified
downstream work-up due to the absence of the enzyme (see discussion of the NH3/CO2
distillation). In trial reactions, the reaction was run with PAL-130 for 8 h at 45 °C,
followed by cooling to 25 °C. Subsequently, the ultrafiltration with a cellulose membrane
(Ultracel from Millipore 10 kDa, 45 cm2) was investigated and filtration experiments were
run for 7–8 h. As a proof of concept, three subsequent batches were successfully run on
50 mL scale, affording 52% yield and complete collection of cinnamic acid 2 and amino
acid 3. In between the individual batches, half of the initial amount of enzyme had to be
replaced due to its relatively short half-life of 10 h at 45 °C (three days at 25 °C). While
the process was technically robust, the main challenge was the short half-life of the
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