Proteases screening for the kinetic resolution of amines with N-acyl α-amino acid trifluoromethyl esters: automated docking approach of binding energies using Subtilisin Novo as a prototype for serine proteases

Article abstract of DOI:10.1016/j.tetasy.2009.11.013

The screening of a panel of 17 proteases resulted in the selection of 4 serine proteases for the resolution of 3-amino-1-phenylbutane. The latter were used to determine the best acyl donor from a series of N-acyl α-amino acid trifluoroethyl esters selecte

Full text of DOI:10.1016/j.tetasy.2009.11.013

Tetrahedron: Asymmetry 20 (2009) 2823–2834
Contents lists available at ScienceDirect
Tetrahedron: Asymmetry
journal homepage: www.elsevier.com/locate/tetasy
Proteases screening for the kinetic resolution of amines with N-acyl
a-amino acid trifluoromethyl esters: automated docking approach of
binding energies using Subtilisin Novo as a prototype for serine proteases
b
b
a,
*
Anne-Lise Bottalla a, Séverine Queyroy c, , Nadia Azzi-Schue , Nicolas Vanthuyne , Stéphane Gastaldi
,
*
b,
Michèle P. Bertrand a, , Gérard Gil
*
*
a Laboratoire de Chimie Moléculaire Organique, LCP UMR 6264, Boite 562, Université Paul Cézanne, Aix-Marseille III, Faculté des Sciences St Jérôme,
Avenue Escadrille Normandie-Niemen, 13397 Marseille Cedex 20, France
b Laboratoire de Stéréochimie Dynamique et Chiralité, ISM2, UMR 6263, Université Paul Cézanne, Aix-Marseille III, Faculté des Sciences St Jérôme,
Avenue Escadrille Normandie-Niemen, 13397 Marseille Cedex 20, France
c Laboratoire de Chimie Théorique et Modélisation, LCP UMR 6264, Université de Provence, Aix-Marseille I, Faculté des Sciences St Jérôme, Avenue Escadrille Normandie-Niemen,
13397 Marseille Cedex 20, France
a r t i c l e i n f o
a b s t r a c t
Article history:
The screening of a panel of 17 proteases resulted in the selection of 4 serine proteases for the resolution of
Received 19 October 2009
Accepted 13 November 2009
Available online 6 January 2010
3-amino-1-phenylbutane. The latter were used to determine the best acyl donor from a series of N-acyl
a
-amino acid trifluoroethyl esters selected as peptide mimetics (E factor up to 99). The results were cor-
related to an automated docking determination of their binding affinities for Subtilisin Novo.
Ó 2009 Elsevier Ltd. All rights reserved.
1. Introduction
further to select the most suitable acyl donors among a series of
new peptide mimetics. A correlation to automated docking calcu-
lations using Subtilisin Novo7 as the model is discussed.
Our group is currently investigating the kinetic resolution of
amines with the prospect of devising dynamic kinetic resolution
(DKR) of amines in which the stereocenter is directly adjacent to
the amine moiety.1 (R)-Selective chemoenzymatic DKR was
achieved via the association of a radical racemization process to
CAL-B-mediated kinetic resolution.2 In order to make the proce-
dure versatile, that is, enabling an alternative selective conversion
of racemic amines into enantiomerically pure (S)-amides, prote-
ase-mediated kinetic resolution3 had to be optimized.1b,4 Prelimin-
ary results led to the design of a one-pot three-step procedure
involving alternated sequential kinetic resolution with alkaline
protease and photochemically induced racemization5 mediated
with trifluoroethane thiol at room temperature.6 However, optimi-
zation was still necessary to achieve a true DKR process involving a
kinetic resolution by a protease.
Herein, we report the screening of a large panel of commercially
available proteases for the resolution of 3-amino-1-phenylbutane.
This amine was selected as the model compound because it led to
intermediate E-values in our preliminary experiments, thus allow-
ing us to appreciate any improvement.1b
Alkaline protease was taken as the reference regarding both the
rate of reaction and the enantioselectivity. This led to the selection
of three additional potentially interesting proteases that were used
2. Results and discussion
Endopeptidases catalyze both the hydrolysis and the formation
of amide links. They are divided into four classes based on their
catalytic mechanism: serine proteases, aspartic proteases, cysteine
proteases, and metalloproteases.8 The screening of 17 different
proteases was achieved in 3-methyl-3-pentanol at room tempera-
ture using N-octanoyl glycine trifluoroethylester 3a as the acyl do-
nor, that is, the optimal experimental conditions already tested for
alkaline protease-mediated kinetic resolution of racemic 3-amino-
1-phenylbutane.1b Proteases are generally less reactive, less selec-
tive, and less stable in organic media than lipases.9 All enzymes
were coated according to the procedure previously described for
subtilisin.10 The results are reported in Table 1. Entries 1–17 follow
the order of improvement in the E-factor.11 Entries 1–9 correspond
to non-selective enzymes or enzymes with very low selectivity. In
most cases, the rate of conversion was low (even for aspartic pro-
teases Protex 50FP and 51FP that were the fastest in this series, en-
tries 2 and 3).
Entries 5 and 9 correspond to metalloproteases, the activity of
which depends on the presence of bound divalent cations.12 Unlike
the other enzymes in the series, the latter belongs to a class of en-
zymes which are known to work in thermodynamically controlled
syntheses.13
* Corresponding authors.
E-mail address: michele.bertrand@univ-cezanne.fr (M.P. Bertrand).
0957-4166/$ - see front matter Ó 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.tetasy.2009.11.013

2824
A.-L. Bottalla et al. / Tetrahedron: Asymmetry 20 (2009) 2823–2834
Table 1
Screening of proteases for the kinetic resolution of 3-amino-1-phenylbutane 1a at 21 °C
O
H
N
n-C7H15
Protease,
NH2
HN
3a
O
Ph
(R)-1
Ph
(S)
3-methyl-3-pentanol, rt
2a
1
Enzyme
Type
—
Time
1 eeb (%)
2a eec (%)
Cd (%)
Ee
1
2
3
4
5
6
7
8
Protex 26L
24 h
8 h
8 h
24 h
24 h
24 h
24 h
20 h
1.2
0
0
5.7
0.8
3.1
9.1
15.1
23.0
37.4
39.9
68.2
52.2
98.1
87.4
99.5
88.9
1.0
0
0
0.8
4.9
3.3
18.4
42.2
46.7
68.1
75.0
78.4
83.3
75.2
90.4
81.0
92.5
64.0f
>75g
>69g
47.0f
48.0f
47.0f
54.0f
26.4
68.0f
35.4
34.7
46.5
38.5
56.6
49.2
55.1
49.0
1h
1
1
Protex 50FP
Protex 51FP
Validase AFP1000L
Protex 14L
3.4.23.18 (Aspergillopepsin)
3.4.23.18 (Aspergillopepsin)
—
3.4.24.27 (Thermolysin)
—
1h
1h
1h
1h
3
Protex 15L
a-Chymotrypsin
3.4.21.1 (Chymotrypsin)
—
Validase FP 500
Protex 7L
Protex 30L
Protex 89L
Protex 6L
Subtilisin A (novo)
Protex 40 XL
Protex 40L
Alkaline protease
Properase 1600L
9
Metallo neutral endopeptidase
3.4.21.62 (Subtilisin)
3.4.21.62 (Subtilisin)
3.4.21.62 (Subtilisin)
3.4.21.62 (Subtilisin)
3.4.21.62 (Subtilisin)
3.4.21.62 (Subtilisin)
3.4.21.62 (Subtilisin)
3.4.21.62 (Subtilisin)
24 h
2
8
10
11
12
13
14
15
16
17
20 min
20 min
4 min
1 h
10
17
19
32
55
55
77
2 h
45 min
45 min
45 min
a
Standard procedure: reactions were performed on a 0.5 M solution of 1 (0.125 mmol) in 3-methyl-3-pentanol at room temperature, with 6 mg of coated enzyme (co-
lyophilized in different buffers corresponding to optimal pH with BrijÒ56 and n-octyl
a
,b-
D
-glucopyranoside, w/w) and 1 equiv of acyl donor.
b
Determined by GC after derivatization.
Determined by HPLC.
Calculated according to C = ees/(ees + eep) unless otherwise stated.
c
d
e
Enantioselectivity factors were calculated according to E = ln[(1 ꢀ C)(1 ꢀ ees)]/ln[(1 ꢀ C)(1 + ees)].
Determined by 1H NMR after work-up from the ratio of remaining acyl donor to amide.
f
g
h
Based on isolated yield.
1 < E < 1.5.
Entries 10–12 correspond to enzymes which give rise to rapid
R1
R2
conversion of amine 1 into amide 2. However, their enantioselec-
tivity was rather low. It should be noted that Protex 6L (entry
12) distinguished itself from the other enzymes due to the fast rate
of acylation of 1.
The rates of reaction were slower for the reactions mediated
with Subtilisin A and Protex 40XL (entries 13 and 14), and the E
factor was moderate.
Protex 40L led to data very close to those obtained with alkaline
protease (entries 15 and 16). Properase 1600L turned out to be
slightly superior to the latter regarding the enantioselectivity with
this acyl donor.
3a
3b
3c
3d
3e
3f
3g
3h
3q
COn-C7H15
COCH3
COPh
COn-C7H15
COCH3
COn-C7H15
COn-C7H14Ph
Ph
H
H
H
CH2Ph
CH2Ph
CH3
H
H
H
O
H
N
(S)
R1
O
CF3
R2
COn-C8H16Ph
Figure 1. Acyl donors.
The enzymes used in entries 12, 15, and 17 were therefore se-
lected [the selection criteria being either the rate of reaction alone
(Protex 6L) or the rate of reaction associated to good enantioselec-
tivity (Protex 40L and Properase 1600L)], and compared to alkaline
protease (entry 16) for the screening of peptide mimetics other
than N-octanoyl glycine trifluoroethylester.14
All the selected enzymes belong to the subtilisin family. Since
the reaction rate is often determined mainly by the specificity of
the enzyme toward the acyl donor, the search for acyl donors bet-
ter recognized at the S1 site of the enzyme channel15,16 led us to
version was determined from the signals of trifluoromethyl groups
of the donor and of released trifluoroethanol, respectively). Due to
the reaction time needed to reach approximately 50% conversion in
the enzymatic acylation, less than 2% of non-catalyzed reaction
could interfere with the enzymatic conversion.20
Our goal was to determine the enzymatic reaction with the
highest possible enantioselectivity coupled to a fast rate of acyla-
tion, that is, the most suitable conditions to further associate this
enzymatic acylation with sulfanyl radical-mediated racemization.
Regarding the E-values, it is obvious that enantio- or diastereose-
lectivity increases in the order: GLY < PHE < ALA for all the en-
zymes tested, but this is only true as far as the acyl substituent
on nitrogen is a long lipophilic chain. The best selectivities were
obtained with 3f (entries 6, 15, 24, 33), followed by 3d (4, 40, 13,
22, 31), and then 3a (entries 1, 10, 10, 100, 19, 28). An average E fac-
tor of 20 was registered with donors 3b (entries 2, 20, 11, 20, 29,
29’), 3c (3, 12, 21, 30), and 3e (5, 14, 23, 32) whatever is the nature
of R2.
test a series of acylated a-aminoesters.
The interest in using N-acyl aminoesters as acyl donor resides in
the binding affinity of proteases for these donors. Their structure is
closely related to the proteases natural substrates.17,18 The N-acyl
amino acid trifluoroesters 3a–g shown in Figure 1 were used to
check whether the enantioselectivity of these enzymes could be
improved. Donor 3h is the unique non-acylated derivative in the
series. The results are given in Table 219 (the results concerning
substrate 3q will be discussed later on).
For alkaline protease and Protex 6L, the time required to reach
50% conversion roughly increased with the enantio- or diastereose-
lectivity. Protex 40L exhibited much longer times of reaction with
It should be noted that the non-catalyzed reaction of amine 1
with acyl donors 3a, 3d, and 3f was followed by 19F NMR (the con-

A.-L. Bottalla et al. / Tetrahedron: Asymmetry 20 (2009) 2823–2834
2825
Table 2
for subtilisin was reported to be TYR, PHE > LEU, MET, LYS > HIS,
ALA, GLN, SER ꢁ GLU, GLY.21 Perona and Craik concluded that
hydrophobicity of the S1 site was the main driving force for subtil-
isin specificity, provided steric repulsions do not predominate.22
Subtilisins are reputed to show little preference for aminoacids
at the S2 and S3 sites, but are known to favor large hydrophobic
residues at the S4 site.15,16 Specificity toward a substrate residue
distant from the scissile bond is a characteristic feature of the sub-
tilisin family. Some data suggest that the interaction between the
substrate and the S4 binding pocket is at least as important for
substrate preference as interactions within the S1 region. The ami-
noacids at the S4 site are TYR104, ILE107, and LEU126 which form
a large hydrophobic pocket.22 The pronounced preference of subtil-
isin BPN’ for aromatic groups in position P4 was assigned to align-
ment of the aromatic group of TYR104 parallel to the aromatic
group of the substrate.23 However, it can be noted that an addi-
tional hydrophobic phenyl substituent at the end of the 8 carbons
acyl chain did not bring any additional selectivity but conversely
rather induced a dramatic drop of the E-factor (see 3g, entries 7,
16, 25, 34).
Screening of acyl donors for the kinetic resolution of 3-amino-1-phenylbutane 1 at
21 °Ca
O
H
N
Protease,
NH2
HN
(S)
R1
R2
acyl donor 3a-h or 3q
Ph
(R)-1
Ph
3-methyl-3-pentanol, rt
2a-h or 2q
1
Entry Enzyme
Acyl
donor
Time
(min)
1 eeb
(%)
2 ee or dec
C
(%)
E
(%)
1
10
2
20
3
4
4’
5
6
7
8
9
Alkaline
protease
3a
3a
3b
3b
3c
3d
3d
3e
3f
35
45
10
25
180
40
60
35
90
120
25
83.2
91.3
81.0
88.0
60.4
83.7
94.0
86.0
76.4
94.3
50.6
77.5
82.1
47.7 57
55.1 55
32.3 24
62.2 22
48.3 27
44.4 72
53.6 79
51.8 19
47.8 99
99.5
42.0
>99.5
78.2
75.0
>99.5
82.2
86.5
84.1
84.0
76.0
3g
3h
3q
62.4
8
52.0 21
48.1 23
60
The comparison of L-phenylalanine and L-alanine derivatives to
N-acylated glycines was expected to parallel the scale of affinity of
related serine proteases for these natural substrates. However, the
data gathered in Table 2 could not be that simply correlated to the
specificity for the acyl donor, since the rate of enzyme acylation
may not be the rate-limiting step, and the enantiomeric ratio is
determined in the second step of the enzymatic process.
It should be noted that experiments achieved with Alkaline pro-
tease and racemic N-octanoyl alanine 3f confirmed that only the
natural (S)-isomer was recognized by the enzyme. The double res-
olution led only to the (S,S)-isomer of 2f.24 The same result was ob-
tained with acyl donor 3d.
10
10’
11
12
13
14
15
16
17
18
Protex 40L
3a
3a
3b
3c
3d
3e
3f
3g
3h
3q
45
60
30
300
280
120
60
30
45
60
87.4
>99.5
78.8
42.1
79.7
75.6
59.5
96.5
43.0
68.0
90.4
83.3
76.1
64.0
90.3
79.1
95.2
51.1
44.8
82.4
49.2 55
54.4 64
50.9 17
39.7
7
46.9 48
48.9 19
38.5 72
65.4 11
49.0
4
45.2 21
19
20
21
22
23
24
25
26
26’
27
Properase
1600L
3a
3b
3c
3d
3e
3f
3g
3h
3h
3q
45
30
420
75
60
120
40
60
40
60
88.9
79.0
65.8
80.2
57.0
83.8
88.9
96.2
63.2
77.9
92.5
79.0
74.5
92.0
83.1
93.4
86.8
84.9
92.7
93.7
49.0 77
50.0 20
46.9 13
46.6 58
40.7 19
47.3 77
50.6 42
53.1 48
40.5 52
45.4 73
A better understanding of ligand/protein interactions was ex-
pected from the automated docking calculations that were per-
formed on coated Subtilisin Novo. Therefore,
a series of
comparative experiments were undertaken with this pure enzyme
and the same acyl donors. The results obtained for the resolution of
amine 1 are given in Table 3. Subtilisin Novo roughly exhibited the
same trend as the previously selected enzymes. The highest enan-
tio- or diastereoselectivities were observed for peptide mimetics
bearing a long hydrophobic chain, that is, 3a, 3d, 3f (entries 1, 10,
4, 40, 6, 60). However, it was far less selective with donor 3a derived
28
29
29’
30
31
32
33
34
35
36
Protex 6Ld
3a
3b
3b
3c
3d
3e
3f
3g
3h
3q
10
15
35
45
120
45
50
75
25
60
67.6
70.3
>99.5
76.2
98.7
61.3
90.1
72.7
69.0
94.3
77.7
86.8
69.2
69.9
85.0
72.2
91.0
66.2
74.9
80.8
46.5 16
44.8 29
59.0 31
52.1 13
53.7 61
45.9 11
49.7 67
52.3 11
48.0 14
53.9 33
from glycine than with acyl donors derived from
L-phenylalanine
and -alanine. In addition, the time necessary to reach approxi-
L
a
b
c
Standard procedure given in Table 1 unless otherwise stated.
Determined by GC after derivatization.
Determined by HPLC.
Table 3
Screening of acyl donors for the kinetic resolution of 3-amino-1-phenylbutane 1 at
21 °C with Subtilisin Novoa
d
1 mg of enzyme per 1.25 mmol of substrate.
Entry Enzyme
1
1’
2
3
4
4’
5
5’
6
Acyl
donor
Time
(min)
1 eeb
(%)
2 ee or dec
(%)
C
(%)
E
phenylalanine derivative 3d than with 3a or 3f, even though no
correlation could be made with E (entries 100, 13, 15). Among the
most efficient acyl donors, Properase 1600L reacted more slowly
with 3f compared to the other enzymes, although this parameter
should be discussed cautiously. It is easier to comment on enantio-
or diastereoselectivity than on the approximate rate of reactions
since the true concentration of most enzymes is unknown, and
therefore the comparison of the times necessary to reach half con-
version might be hazardous.
3a
3a
3b
3c
3d
3d
3e
3e
3f
60
180
50
150
20
40
30
50
25
30
90
50
240
52.2
97.0
62.9
46.7
77.3
>99.5
89.6
>99.5
86.8
>99.5
73.2
64.5
62.0
83.3
72.9
77.8
76.3
88.5
69.7
72.1
42.6
92.5
81.3
59.7
62.2
84.7
38.5 19
57.1 26
44.7 15
38.0 11
46.6 39
58.8 32
55.4 18
70.0 13
48.4 73
55.1 71
Subtilisin
Novo
6’
7
8
3f
3g
3h
3q
55.0
50.9
42.3 22
8
8
The non-acylated glycine derivative 3h (entries 8, 80, 17, 26, 260,
35) did not give good enantioselectivities except with Properase
1600L.
These results are in good agreement with literature data con-
cerning subtilisin specificity. The approximate order of preference
9
a
b
c
Standard procedure given in Table 1.
Determined by GC after derivatization.
Determined by HPLC.

2826
A.-L. Bottalla et al. / Tetrahedron: Asymmetry 20 (2009) 2823–2834
Table 4
O
O
H
N
Binding sites relative populations, minimum free energy, and mean free energy, all
ligands and all subtilisin structures taken together
(S)
R
R1
O
CF3
R2
O
R
CF3
R2
R2
R2
Binding site
% of docked
conformations
D
mol)
Gmin (kcal/
DGmean (kcal/
mol)
R1
H
H
H
3i COn-C7H14Ph
3n COn-C8H16Ph
3o COn-C9H18Ph
3p COn-C11H23
3r COn-C9H18Ph
3s COn-C11H23
3t COCH3
CH2Ph
CH2Ph
CH2Ph
CH2Ph
H
H
CH3
CH3
3j
3k
3l
n-C2H5
n-C3H7
n-C10H21
n-C6H13
1
SER221/HIS64/
ASP32
LEU235
LYS 43/VAL44/
PRO86
GLN206
GLY47/MET50
SER249
THR164/LYS170
THR244/
43
ꢀ5.87
ꢀ2.22
2
3
21
14
ꢀ5.22
ꢀ5.02
ꢀ2.16
ꢀ2.28
CH2Ph
3m
4
5
6
7
8
7
4
3
2
2
ꢀ4.73
ꢀ3.85
ꢀ3.39
ꢀ3.58
ꢀ3.61
ꢀ1.80
ꢀ1.71
ꢀ1.64
ꢀ1.42
ꢀ2.12
3u COn-C11H23
Figure 2. Structures of docked potential acyl donor candidates.
ARG247
9
ASP60/LYS94
1
1
1
ꢀ2.60
ꢀ3.03
ꢀ3.53
ꢀ1.28
ꢀ1.52
ꢀ1.47
mately 55% conversion decreased while E increased (see entries 10,
10 ARG186
11 GLN10
40, 60).25
3. Computational data
site triad, that is, SER221/HIS64/ASP32 accounts for 43% of the con-
formations. The second most populated site (21%) is close to
LEU235. The third one (14%) is located around the LYS43/VAL44/
PRO86 triad. The other eight sites account for 1–7% of the docked
conformations. For each site a mean binding free energy and the
minimum binding free energy are given.
As aforementioned, automated docking was used to explore the
binding affinities of different acyl donors for Subtilisin Novo that
was selected as prototype for serine proteases. Blind docking sim-
ulations were performed with the AUTODOCK 4.0 program26 with five
different structures of Subtilisin Novo issued from the Protein Data
Bank (PDB). It has been shown recently that the docking approach
could be used to predict protease specificities.18
The three most populated sites correspond to the lowest mini-
mum free energy, and thus to the strongest interactions.
Focusing our attention on the active site enabled a more accu-
rate comparison of the various acyl donors. For each ligand in the
series, the percentage of conformations docked in the active site,
their average binding free energy, and the binding energy of the
most stable docked conformation are plotted in Figures 4–6,
respectively. Three groups can be distinguished.
For ligands with a short chain length (3b, 3c, 3e, 3h, 3j, 3k, 3t),
the competition between the active site and other binding sites is
not in favor of the former [nconf (%) does not exceed 41%, Fig. 4].
Thus, within this first group of ligands, N-acetyl phenylalanine 3e
seems to be the best ligand. The percentage of conformations
docked at the active site shows that peptide mimetics have more
affinity for the active site than simple ester derivatives. Ligands
3j and 3k, which are the poorest ligands, are not peptide mimetics.
However, if one looks at the second group of ligands 3f, 3a, 3d,
3l, 3u, 3p, 3s, excluding 3m, the preference for peptide mimetics
can nearly be offset by the influence of chain length. Octanoyl gly-
For sake of comparison, the docking of ligands like trifluoroethyl
N-dodecanoyl glycinate 3s, a peptide mimetic with a long hydro-
phobic chain, and trifluoroethyl butanoate 3j, trifluoroethyl pent-
anoate 3k, or trifluoroethyl dodecanoate 3l, which are not
peptide mimetics, was also examined. In addition, potential pep-
tide mimetics candidates bearing alkanoyl chains in C8, C9, C10,
and C11 with a terminal phenyl group, that is, 3i, 3n, 3o, 3g, 3q,
3r, were investigated with the purpose of evaluating the interde-
pendence of binding subsites. The structures of all these additional
ligands are given in Figure 2.
Figure 3 shows views of the enzyme structure represented by
its Connelly surface. The active site can easily be located by the
location of SER221 and the secondary binding sites are colored in
pink.
As summarized in Table 4, on average, when all ligands and all
enzyme structures were taken together, 99% of the docked confor-
mations were located in 11 binding sites. Docking near the active
Figure 3. Connelly surface views of the two faces of Subtilisin Novo structure showing the active site (SER221), the S1–S4 channel (blue), and secondary binding sites (pink).

A.-L. Bottalla et al. / Tetrahedron: Asymmetry 20 (2009) 2823–2834
2827
Figure 4. Population of the active site versus the main chain length (numbered
from the scissile bond without taking into account terminal phenyl groups) for each
ligand. Ligands containing terminal phenyl groups are pointed by a circle.
Figure 6. Minimum binding free energy for each ligand docked in the active site
according to its structural features.
Since calculations of binding affinities led us to the conclusion
that ligands containing 12 or 13 carbons in the acyl chain bound
to nitrogen (3q, 3o, 3n, 3r) were better recognized by the active
site, for the sake of synthesis facility, the glycine derivative 3q
was selected and synthesized. The results of kinetic resolution by
the selected enzymes are given in Table 2 (entries 9, 18, 27, 36)
and in Table 3 (entry 9) for the resolution of 1 by Subtilisin Novo.
The highest E-value (73) was observed with Properase 1600L (Ta-
ble 2, entry 27). The lowest E-values (21–23) were observed for
alkaline protease, Protex 40L (Table 2, entries 9 and 18), and for
subtilisin (Table 3, entry 9). An intermediate E-value (33) was reg-
istered with Protex 6L (Table 2, entry 36). The reaction times nec-
essary to reach approximately 50% conversion were in all cases
superior to those needed with 3a.
The most important data extracted from experimental results
(including ligand 3q) are summarized in the diagrams given in Fig-
ures 7 and 8.
Figure 7 shows that alkaline protease and subtilisin exhibit
approximately the same trend (columns 1 and 5), although the for-
mer is in all cases more enantioselective. Properase 1600L shows a
rather small variation in the enantioselectivity with the selected
ligands.
cine and alanine derivatives 3a and 3f, respectively, show a lower
affinity than octanoyl phenylalanine 3d. But the affinity of ester 3l
is even greater. This confirms again that molecular recognition by
subtilisin is not limited to S1 site.15,16 The binding affinity for the
PHE, ALA, and GLY donors varies with increasing the chain length,
but the variation is not as important as that observed with the
introduction of a phenyl group at the terminal carbon of the alkyl
chain in the third group of ligands (3g, 3i, 3n, 3o, 3q, 3r). The high-
est affinity for the active site was calculated for ligands of this third
group, that is, ligands having 11–13 atom chain length and bearing
a phenyl substituent at their terminal carbon atom.
This statistical analysis was confirmed by the trend shown in
Figures 5 and 6, for the mean and minimum binding free energies,
respectively, for each ligand according to its family.
The mean and mininum free energies seem not to be greatly af-
fected by the chain length in the GLY series provided an alkyl chain
is attached to the N-acyl group (3b, 3a, 3s). Conversely, the chain
length significantly increases the enzyme specificity in the ALA
and PHE series.
The series of 15 atoms chain length ligands shows a variation of
affinity in the order GLY < ALA < PHE (Figs. 5 and 6: 3p, 3u, 3s). The
affinity for ligand 3m, that is not a peptide mimic, shows the
importance of the benzyl group at the P1 site.
The most stable docked conformations are formed with ligands
derived from phenylalanine bearing terminal phenyl groups 3n, 3o.
It should be noted that adding one carbon to the alkylchain
bearing a terminal phenyl group (3q compared to 3g) led to a sig-
nificant increase in enantioselectivity that could be up to three
times higher depending on the enzyme (Fig. 8).
Thus, blind docking was helpful in predicting subtilisin specific-
ity, that is, the most promising ligands. However, the multi-step-
Figure 7. Compared enantioselectivity of ligands 3a, 3d, 3f, and 3q for the selected
enzymes and subtilisin.
Figure 5. Mean binding free energy for each ligand docked in the active site
according to its structural features.

2828
A.-L. Bottalla et al. / Tetrahedron: Asymmetry 20 (2009) 2823–2834
Figure 8. Compared enantioselectivity of ligands 3g, and 3q for the selected
enzymes and subtilisin.
catalyzed reaction requires the ligands to dock in the active site
with a preorganized structure27 close to that of the tetrahedral
intermediate. In order to study more accurately the structures of
the ligands docked near the catalytic triad, we performed new sim-
ulations, where the ligands were allowed to dock only near the ac-
tive site and not all over the enzyme as previously.
The structures for which the distance between the carbon of the
ester group and SER221 oxygen atom (r1) was shorter than 3.5 Å
were reckoned as preorganized conformations (Fig. 9). The number
of these preorganized conformations for each ligand is given in Fig-
ure 10. Compared to other similar studies,28 far less preorganized
conformations were observed. Moreover, these conformations
were not necessarily the ones with the strongest binding free en-
ergy (except for ligand 3g). Nevertheless, we have attempted to
analyze these results.
The number of preorganized conformations increases with the
chain length for peptide mimetics ligands. The nature of the pep-
tide shows a higher frequency of preorganized conformations in
the order PHE < ALA < GLY which points to the influence of hydro-
phobic interactions in the specificity pocket. Among the ligands
with a phenyl group at the end of the acyl chain, the ones with a
chain of 9 carbons (3n and 3q) are favored. The evolution of the
free energy for these preorganized conformations according to
the structural features of the ligands shows the same tendency
as observed in the blind docking (see Fig. 5).
Figure 10. Percentage of preorganized docked conformations in the active site for
each ligand according to its structural features.
As the tetrahedral intermediate is stabilized by hydrogen bonds,
it was interesting to check if these ligand-receptor interactions oc-
curred in the preorganized conformations (Fig. 9). Therefore, three
distances between the atoms susceptible to form hydrogen bonds
were monitored: first, the distance between the ester carbonyl
oxygen atom and the proton of the NH group of the ASN155 resi-
due in the oxyanion hole (r2), second, the distance between the
oxygen atom of SER125 carbonyl group and the glycine NH proton
in the ligand (r3), and third, the distance between the oxygen atom
of the CF3CH2O group and the proton of SER221 hydroxyl proton
(r4). In the docked structures for which r1 is less than or equal to
3.5 Å, hr3i is relatively long (4.4 Å in average over all ligands), while
hr2i and hr4i are shorter (3.2 Å and 3.5 Å, respectively). Moreover, in
about half of these structures r2 and r4 are shorter than 3.5 Å at the
same time, suggesting that water molecules are not absolutely nec-
essary to mediate proton transfers.
The relatively small number of preorganized conformations for
each ligand makes the statistical approach difficult and therefore, it
was not possible to evaluate the effect of the chain length or of the
nature of the peptide mimic on parameters r2, r3, and r4. It was pos-
sible, however, to observe that within the ligands bearing a termi-
nal phenyl group in the GLY series (3g, 3q, 3r), ligand 3q showed
the shortest distances hr2i and hr4i, thus prefiguring its greater
affinity. However, docking simulations in the active site did not
help us to discriminate the ligands more accurately.
4. Conclusion
The screening of a large panel of commercially available prote-
ases for the resolution of 3-amino-1-phenylbutane resulted in the
selection of four serine proteases which were used to determine
the best acyl donor from a series of trifluoroethyl esters. The most
efficient donors happened to be trifluoroethyl esters derived from
N-acyl a-amino acids bearing an aliphatic acyl chain 8 or 9 carbon
long. Blind automated docking enabled a rapid selection of the
most promising ligands, by looking at their binding affinity for
Subtilisin Novo.
A noteworthy increase in the enantiomeric ratio was observed
with alkaline protease by replacing the glycine moiety by alanine
(3a/3f; E factor up to 99).
5. Experimental
Figure 9. Preorganized docked conformation of 3q in the active site. Stabilizing
hydrogen bonding interactions occurs between the ligand amide NH and SER125,
the ester carbonyl oxygen atom and ASN155 in the oxyanion hole.
The1Hand13CNMRspectrawererecordedat200 or300 MHzand
50 or 75 MHz, respectively. Chemical shifts are reported in ppm, and

A.-L. Bottalla et al. / Tetrahedron: Asymmetry 20 (2009) 2823–2834
2829
coupling constants (J) are given in hertz. For the MS analysis, a QStar
Elite (Applied Biosystems SCIEX) mass spectrometer fitted with an
atmospheric pressure ionization (API) source was used. The samples
were ionized in positive electrospray mode, ion source voltage (ISV):
5500 V, orifice voltage (OR): 40–150 V, air pressure: 10–24 psi.
Enantiomeric excesses (ee) were determined by chiral GC and chiral
HPLC, respectively, for amines and amides. The chiral HPLC analyses
were performed on Chiralcel OD-H (250 ꢂ 4.6 mm, cellulose tris-
(3,5-dimethylphenylcarbamate)), Chiralpak AD, or AD-H
(250 ꢂ 4.6 mm, amylose tris-(3,5-dimethylphenylcarbamate)),
Sepapak-2-HR (250 ꢂ 4.6 mm, cellulose tris(3-chloro-4-methyl-
phenylcarbamate)), and Chiralpak IC (250 ꢂ 4.6 mm, cellulose
tris(3,5-dichlorophenylcarbamate)) with UV detection. The reten-
tion times tR are given in minutes for each enantiomer, the retention
factors k = (tR ꢀ t0)/t0 (t0 is the retention time for an unretained peak
determined by injection of tri-tertiobutyl-benzene), the enantiose-
10.30 g) and 2,2,2-trifluoroethanol (1 equiv, 50 mmol, 3.60 mL)
were added. The mixture was refluxed overnight. After evaporation
of the solvent and dilution in diethyl ether, the dicyclohexylurea
was removed by filtration. The solvent was removed and the resi-
due was purified by recrystallization from a chloroform/hexane
mixture. Compound 3b (5.9 g , 29 mmol, 59%) was isolated as a
white powder. 1H NMR (200 MHz, CDCl3): 2.05 (s, 3H), 4.13 (d,
3
2H, J = 5.5), 4.52 (q, 2H, JHF = 8.4), 6.14 (br s, 1NH). 13C NMR
2
(75 MHz, CDCl3): 23.1 (CH3), 41.3 (CH2), 61.0 (q, CH2, JCF = 37),
1
123.2 (q, CF3, JCF = 277), 169.1 (C), 171.1 (C). Mp 72–73 °C. HRMS
([M+H]+, ESI): m/z calcd for C6H9NO3F3: 200.0529. Found:
200.0528.
5.2.2. N-Benzoyl glycine trifluoroethyl ester 3c
To a 0 °C solution of N-benzoyl glycine (50 mmol, 8.95 g) in
CHCl3 (50 mL), dicyclohexyl carbodiimide (DCC) (1 equiv,
50 mmol, 10.30 g) and 2,2,2-trifluoroethanol (1 equiv, 50 mmol,
3.60 mL) were added. The mixture was refluxed overnight. After
evaporation of the solvent and dilution in diethyl ether, the dicy-
clohexylurea was removed by filtration. The solvent was removed
and the residue was purified by recrystallization from a chloro-
form/hexane mixture. Compound 3c (7.30 g, 28 mmol, 56%) was
isolated as a white powder. 1H NMR (200 MHz, CDCl3): 4.37 (d,
lectivity factor
a = k2/k1 and the resolution Rs are given to character-
ize the chiral separations. The enantiomeric excesses of the primary
amine were determined after derivatization in trifluoroacetamide
with 1.5 equiv of N-methyl-bis-trifluoroacetamide, by gas chroma-
tography (GC) analysis on a chromatograph fitted with a Lipodex D
column with flame ionization detector (FID), using the derivatized
racemic compounds as references.
3
N-Acetyl glycine, N-benzoyl glycine, N-acetyl
L
-phenylalanine,
2H, J = 5.4 Hz), 4.59 (q, 2H, JHF = 8.3), 6.72 (br s, 1NH), 7.44 (m,
and 3-amino-1-phenyl butane 1 are commercially available and
used without further purification. N-Octanoyl glycine trifluoro-
ethyl ester 3a, (S)-3-amino-1-phenyl butane, and N-(2-oxo-2-(4-
phenylbutan-2-ylamino)ethyl)octanamide 2a were prepared
according to the literature procedures.1
3H), 7.84 (m, 2H). 13C NMR (75 MHz, CDCl3): 42.2 (CH2), 62.0 (q,
2
1
CH2, JCF = 37), 123.5 (q, CF3, JCF = 277), 127.8 (2 ꢂ CH), 129.5
(2 ꢂ CH), 132.8 (CH), 134.2 (C), 168.5 (C), 169.5 (C). Mp 123–
124 °C. HRMS ([M+H]+, ESI): m/z calcd for C11H11NO3F3:
262.0686. Found: 262.0685.
All enzymes were commercially available, they were used as
received.
5.2.3. N-Octanoyl L-phenylalanine trifluoroethyl ester 3d
The assay methods used by Genencor, Novozymes, and Valley Re-
search for the determination of enzymatic activities are based on the
proteolytic hydrolysis of casein or hemoglobin. A specific method
was developed for each protease, so the enzymatic activities are ex-
pressed in different units. The specific assay methods are available
from these companies upon request. From Genencor: Protex 26L:
2120 SAPU/g, Protex 50FP: 524000 HUT/g, Protex 51FP: 472501
HU/g, Protex 14L: 160 U/g, Protex 15L: 1088 SAPU/g, Protex 7L:
1776 AZO/g, Protex 30L: 3117 GSU/g, Protex 89L: 3274 GSU/mL, Pro-
tex 6L: 611000 DSU/g, Protex 40XL: 57.1 MPU/g, Protex 40L: 47000
GSU/L, Properase 1600L: 1738 PU/g. From Novozymes: Subtilisin
Novo: 25 AU(H)/g. From Valley Research: Validase AFP 1000L:
1000 APU/g, Validase FP500: 500000 HUT/g, Alkaline Protease:
1660 DAPU/g. From Sigma: Alpha chymotrypsin: 40,000 U/g.
O
Bn
n-C7H15-COCl
OH
L-Phenylalanine
O
C7H15
N
H
91%
A
O
Bn
.
BF3 OEt2
CF3CH2OH
O
CF3
C7H15
N
H
3d
O
71%
5.2.3.1. N-Octanoyl
L
-phenylalanine A.
To a solution of L-
phenylalanine (50 mmol, 8.25 g) in a 15% NaOH (50 mL) at 0 °C
was added octanoyl chloride (1.5 equiv, 13.3 mL). The reaction mix-
ture was stirred overnight at rt. After adding 20% HCl to pH 2, the or-
ganic products were extracted with dichloromethane (2 ꢂ 50 mL).
The solvent was evaporated and the residue was purified by recrys-
tallization from a chloroform/hexane mixture. Compound A (13.2 g,
45 mmol, 91%) was isolated as a white powder. 1H NMR (200 MHz,
CDCl3): 0.91 (t, 3H, J = 6.0), 1.33 (m, 8H), 1.48 (quint, 2H, J = 7.5),
2.17 (t, 2H, J = 7.5), 3.12 (dd, 1H, J = 14.1 and 5.6), 3.25 (dd, 1H,
J = 14.1 and 6.3), 4.90 (q, 1H, J = 6.0), 6.11 (d, 1H, J = 7.4, NH), 7.26
(m, 5H), 9.73 (br s, 1H, COOH). 13C NMR (75 MHz, CDCl3): 14.2
(CH3), 22.7 (CH2), 25.7 (CH2), 29.0 (CH2), 29.2 (CH2), 31.7 (CH2),
36.5 (CH2), 37.4 (CH2), 53.2 (CH), 127.2 (2 ꢂ CH), 128.3 (2 ꢂ CH),
129.4 (CH), 136.1 (C), 174.2 (C), 174.8 (C). Mp 122–123 °C. HRMS
([M+H]+, ESI): m/z calcd for C17H26NO3: 292.1907. Found:
5.1. Immobilization of proteases
At first, 250 mg of octyl
a,b-D-glucopyranoside and 250 mg of
BrijÒ 56 (polyethylene glycol hexadecyl ether) were diluted in
100 mL of buffer at 60 °C (phosphate 0.1 M pH 7; phosphate
0.1 M pH 8; or citrate 0.1 M pH 4 depending on the enzyme), these
solutions were then cooled to rt.
The protease liquid extracts (250 mg) or protease solid extracts
(Protex 50FP and 51FP, 125 mg) were added to 12 mL of these solu-
tions, according to the optimal pH of each enzyme (phosphate
0.1 M pH 7: Alkaline protease, Protex 30L, 40L, 6L, 7L, 89L; phos-
phate 0.1 M pH 8: Protex 14L, 1600L, 40XL, 51FP; citrate 0.1 M
pH 4: Protex 15L, 26L, 50FP, Validase FP 500, AFP 1000L). The mix-
ture was rapidly frozen in liquid N2 and lyophilized for 20 h.
292.1906. ½a 2D5
¼ þ75:0 (c 1.06, CHCl3).
ꢃ
5.2.3.2. N-Octanoyl
3d.
L
-phenylalanine
trifluoroethyl
ester
After dilution of N-octanoyl -phenylalanine A (1 equiv,
5 mmol, 1.46 g) in trifluoroethanol (300 equiv, 1.5 mol, 115 mL),
150
L
l
L of BF3ꢄOEt2 was added at 0 °C. The reaction mixture was re-
5.2. Acyl donor synthesis
fluxed for 8 days. After distillation of trifluoroethanol, the mixture
was diluted in dichloromethane (50 mL) and washed successively
with 1 M HCl, 1 M NaOH, and brine. After evaporation of the sol-
vent, 1.32 g of pure 3d was obtained (71% yield). 1H NMR
5.2.1. N-Acetyl glycine trifluoroethyl ester 3b
To a 0 °C solution of N-acetyl glycine (50 mmol, 5.85 g) in CHCl3
(50 mL), dicyclohexyl carbodiimide (DCC) (1 equiv, 50 mmol,

2830
A.-L. Bottalla et al. / Tetrahedron: Asymmetry 20 (2009) 2823–2834
(200 MHz, CDCl3): 0.87 (t, 3H, J = 6.9), 1.25 (m, 8H), 1.57 (quint, 2H,
J = 7.3), 2.17 (t, 2H, J = 7.3), 3.10 (dd, 1H, J = 14.1 and 6.4), 3.20 (dd,
1H, J = 14.1 and 5.9), 4.49 (AB pattern of ABX3, 2H, JAB = 12.5,
5.2.6. N-(8-Phenyl)octanoyl glycine trifluoroethyl ester 3g
Ph-nC7H14-COCl
O
JAX = JBX = 8.3,
Dm = 22), 4.96 (q, 1H, J = 6.1), 5.77 (d, 1NH, J = 7.8),
Glycine
OH
Ph
7.09 (m, 2H), 7.31 (m, 3H). 13C NMR (75 MHz, CDCl3): 14.5 (CH3),
23.0 (CH2), 25.9 (CH2), 29.4 (CH2), 29.5 (CH2), 32.0 (CH2), 36.8
N
94%
H
C
O
O
2
(CH2), 38.0 (CH2), 53.1 (CH), 61.1 (CH2, q, JCF = 37), 124.9 (CF3, q,
O
CF3
Ph
DCC/CF3CH2OH
32%
1JCF = 278), 127.8 (2 ꢂ CH), 129.2 (2 ꢂ CH), 129.5 (CH), 135.6 (C),
170.8 (C), 173.2 (C). Mp 69–70 °C. HRMS ([M+H]+, ESI): m/z calcd
for C19H27NO3F3: 374.1938. Found: 374.1939. Ee >99.5% (chiral
HPLC chiralpak AD, n-hexane/isopropanol 90:10, flow
rate = 1 mL/min, UV 220 nm, tR (R) = 6.1 min, tR (S) = 7.5 min,
N
H
3g
O
= 1.46, Rs = 1.4), ½a D25
¼ þ25:5 (c 1.00,
ꢃ
5.2.6.1. 2-(8-Phenyloctanamido)acetic acid C.
To a solution
k(R) = 1.00, k(S) = 1.46,
CHCl3).
a
of glycine (432 mg, 4.2 mmol) in a mixture of water (400
lL) and
acetone (2 mL) was added triethylamine (940 L, 4.2 mmol). The
l
solution was cooled down to 0 °C and 8-phenyloctanoyl chloride
(1.0 g, 4.2 mmol) was added dropwise. The solution was allowed
to warm up to room temperature. After 18 h at room temperature,
the solvent was removed. The mixture was acidified to pH 2 with
37% aqueous HCl. The aqueous phase was extracted with CHCl3
(3 ꢂ 5 mL). The combined organic phases were dried over anhy-
drous MgSO4, filtered, and concentrated in vacuo to get C (1.1 g,
3.9 mmol, 94%) as a white solid which was used in the next step
without further purification. Mp 70–72 °C. 1H NMR (300 MHz,
CDCl3): 9.23 (br s, 1H, COOH), 7.32–7.27 (m, 2H), 7.20–7.17 (m,
3H), 6.37 (t, 1H, J = 5.0), 4.05 (d, 2H, J = 5.0), 2.60 (t, 2H, J = 7.6),
2.26 (t, 2H, J = 7.5), 1.66–1.59 (m, 4H), 1.38–1.30 (m, 6H). 13C
NMR (75 MHz, CDCl3): 174.2 (C), 173.3 (C), 142.7 (C), 128.3
(2 ꢂ CH), 128.2 (2 ꢂ CH), 125.5 (CH), 45.8 (CH2), 36.2 (CH2), 34.0
(CH2), 31.4 (CH2), 29.1 (CH2 ꢂ 3), 25.5 (CH2). HRMS ([M+H]+,
ESI): m/z calcd for C16H24NO3: 278.1751. Found: 278.1748.
5.2.4. N-Acetyl
L-phenylalanine trifluoroethyl ester 3e
The procedure is the same as that for 3d, using N-acetyl
L
-phen-
ylalanine (10 mmol, 2.07 g). Compound 3e (1.74 g, 6 mmol, 60%)
was obtained as a white powder. 1H NMR (200 MHz, CDCl3): 1.98
(s, 3H), 3.14 (AB pattern of ABX, 2H, JAB = 14.0,
Dm = 14), 4.49 (AB
pattern of ABX3, 2H), 4.97 (q, 1H, J = 7.9), 5.85 (d, 1H, J = 7.6, NH),
7.09–7.31 (m, 5H). 13C NMR (75 MHz, CDCl3): 23.4 (CH3), 37.9
2
1
(CH2), 53.3 (CH), 61.1 (CH2, q, JCF = 38), 121.4 (CF3, q, JCF = 246),
127.8 (2 ꢂ CH), 129.2 (2 ꢂ CH), 129.5 (CH), 135.6 (C), 170.1 (C),
170.7 (C). Mp 130–131 °C. HRMS ([M+H]+, ESI): m/z calcd for
C13H15NO3F3: 290.0999. Found: 290.0997. Ee >99.5% (chiral HPLC,
chiralpak AD, n-hexane/isopropanol 90:10, flow rate = 1 mL/min,
UV 220 nm, tR (R) = 13.2 min, tR (S) = 15.3 min, k(R) = 3.33,
k(S) = 4.02,
a
= 1.21, Rs = 1.7), ½a D25
¼ þ44:7 (c 1.01, CHCl3).
ꢃ
5.2.5. N-Octanoyl
L-alanine trifluoroethyl ester 3f
O
5.2.6.2. 2,2,2-Trifluoroethyl
3g. To a solution of acid C (1.06 g, 3.8 mmol) in CHCl3 at
0 °C 2,2,2-trifluoroethanol (273 L, 3.8 mmol) and DCC (784 mg,
2-(8-phenyloctanamido)acetate
n-C7H15-COCl
OH
L-Alanine
O
C7H15
N
H
l
92%
B
O
3.8 mmol) were added. The solution was heated for 18 h at 65 °C.
The solvent was removed under reduced pressure. The residue
was suspended in Et2O and the precipitate was filtered. The solvent
was removed in vacuo and the solid was recrystallized from hex-
ane to afford pure 3g (439 mg, 1.2 mmol, 32%) as a white solid.
Mp 68–70 °C. 1H NMR (300 MHz, CDCl3): 7.32–7.26 (m, 2H), 7.21
.
BF3 OEt2
CF3CH2OH
O
CF3
C7H15
N
H
3f
O
74%
5.2.5.1. N-Octanoyl
L
-alanine B.
The procedure is the same
-alanine (100 mmol, 8.9 g). Compound B
3
as that for A, using
L
(m, 3H), 5.99 (br s, 1H), 4.54 (q, 2H, JHF = 8.3), 4.16 (d, 2H,
(19.9 g, 92.4 mmol, 92%) was obtained as a white powder. 1H
NMR (200 MHz, CDCl3): 0.87 (m, 3H), 1.27 (m, 8H), 1.45 (d, 3H,
J = 7.0), 1.63 (m, 2H), 2.21 (t, 2H, J = 7.0), 4.58 (pseudo quint, 1H,
J = 6.7), 6.23 (d, 1H, J = 5.8, NH), 7.45 (s, 1H, COOH). 13C NMR
(75 MHz, CDCl3): 14.1 (CH3), 18.2 (CH3), 22.6 (CH2), 25.7 (CH2),
29.0 (CH2), 29.2 (CH2), 31.7 (CH2), 36.4 (CH2), 48.3 (CH), 174.4
(C), 175.8 (C). Mp 81–82 °C. HRMS ([M+H]+, ESI): m/z calcd for
J = 5.5), 2.61 (t, 2H, J = 7.4), 2.26 (t, 2H, J = 7.4), 1.71–1.58 (m, 4H),
1.39–1.33 (m, 6H). 13C NMR (75 MHz, CDCl3) 174.2 (C), 169.6 (C),
143.6 (C), 129.2 (2 ꢂ CH), 129.0 (2 ꢂ CH), 126.3 (CH), 122.5 (CF3,
1
2
q, JCF = 278), 61.1 (CH2, q, JCF = 37), 41.6 (CH2), 36.7 (CH2), 32.2
(CH2), 31.7 (CH2), 29.9 (3 ꢂ CH2), 26.2 (CH2). 19F NMR (282 MHz,
CDCl3) ꢀ74.2 (CF3). HRMS ([M+H]+, ESI): m/z calcd for
C18H25NO3F3: 360.1781. Found: 360.1778.
C11H22NO3: 216.1594. Found: 216.1595.
½
a 2D5
ꢃ
¼ þ5:8 (c 1.00,
CHCl3).
5.2.7. N-Phenyl glycine trifluoroethyl ester 3h
5.2.5.2. N-Octanoyl
L
-alanine trifluoroethyl ester 3f.
The
CF3CH2OH
Et3N
procedure is the same as that for 3d, using B (20 mmol, 4.3 g).
Compound 3f (4.4 g, 14 mmol, 74%) was obtained as a white pow-
der. 1H NMR (200 MHz, CDCl3): 0.84 (t, 3H, J = 6.4), 1.27 (m, 8H),
1.45 (d, 3H, J = 7.2), 1.62 (quint, 2H, J = 7.2), 2.21 (t, 2H, J = 7.2),
4.41 (m, 1H), 4.62 (m, 2H), 5.99 (d, NH, J = 6.6). 13C NMR
(50 MHz, CDCl3): 14.1 (CH3), 18.1 (CH3), 22.7 (CH2), 25.6 (CH2),
29.0 (CH2), 29.2 (CH2), 31.7 (CH2), 36.4 (CH2), 47.8 (CH), 60.5
Cl
O
CF3
Br
Br
62%
CF3
O
O
E
Ph
O
Aniline
71%
N
H
O
3h
2
1
(CH2, q, JCF = 37), 123.2 (CF3, q, JCF = 279), 171.8 (C), 172.9 (C).
Mp 79–81 °C. HRMS ([M+H]+, ESI): m/z calcd for C13H23NO3F3:
298.1625. Found: 298.1626. Ee >99.5% (chiral HPLC, chiralcel OD-
H, n-hexane/ethanol 98:2, flow rate = 1 mL/min, UV 220 nm, tR
5.2.7.1. 2,2,2-Trifluoroethyl bromoacetate E.
Triethylamine
(1.1 equiv, 55 mmol, 7.4 mL) was added at 0 °C to a solution of tri-
fluoroethanol (1 equiv, 50 mmol, 3.6 mL) in diethyl ether (80 mL).
Bromoacetyl chloride (1 equiv, 50 mmol, 4.1 mL) was added drop-
(R) = 14.2 min, tR (S) = 17.3 min, k(R) = 3.66, k(S) = 4.67,
a = 1.28,
Rs = 1.8), ½a 2D5
¼ ꢀ7:5 (c 1.03, CHCl3).
ꢃ

A.-L. Bottalla et al. / Tetrahedron: Asymmetry 20 (2009) 2823–2834
2831
wise at 0 °C, then the mixture was stirred for 2 h at rt. After careful
evaporation of diethyl ether, 2,2,2-trifluoroethyl bromoacetate E
was distilled (3 mbar, 80 °C) (6.9 g, 31 mmol, 62%). 1H NMR
methyl-bis-trifluoroacetamide. The amide ee was determined by
high performance liquid chromatography after dilution in the elut-
ing solvent. In the case of 2h, the acyl donor and amide retention
times interfered, so a purification by flash-chromatography was
necessary before injection.
3
(200 MHz, CDCl3): 3.92 (s, 2H), 4.56 (q, 2H, JH–F = 8.3). 13C NMR
2
(50 MHz, CDCl3): 24.4 (CH2), 62.1 (CH2, q, JCF = 37), 119.9 (CF3, q,
1J = 275), 166.0 (C).
5.4. General procedure for the synthesis of racemic amides 2a–
2h and 2q for chiral HPLC analysis
5.2.7.2. N-Phenyl Glycine trifluoroethyl ester 3h.
To a solu-
tion of aniline (2 equiv, 10 mmol, 911 L) in acetonitrile (20 mL)
l
was added E at rt. The reaction mixture was refluxed for 15 h. After
cooling and filtration, the reaction mixture was concentrated un-
der vacuum, diluted in diethyl ether, and then washed with water.
After purification by flash-chromatography (CH2Cl2 100%), 3h
(825 mg,3.5 mmol, 71%) was isolated. 1H NMR (200 MHz, CDCl3):
To a mixture of the carboxylic acid corresponding to each acyl
donor (1 mmol) and 1 (1 mmol) in chloroform (5 mL) was added
dicyclohexyl carbodiimide (1 mmol). The reaction mixture was
heated at 65 °C (2a,1b 2b and 2c) or stirred at room temperature
(2d, 2e and 2f) for 15 h. After concentration, the product was puri-
fied by crystallization from hexane or by chromatography on gel-
silica (yield 43–75%). Racemic amides 2g, 2h, and 2q were obtained
by reaction of the corresponding trifluoroesters with 1 catalyzed
by the non-stereoselective enzyme Protex 50FP.
3
4.05 (d, 2H, J = 5.9), 4.24 (m, 1H, NH), 4.59 (q, 2H, JH–F = 8.4),
6.65 (d, 2H, J = 7.8) , 6.80 (t, 1H, J = 7.3), 7.23 (t, 2H, J = 8). 13C
2
NMR (75 MHz, CDCl3): 45.8 (CH2), 61.4 (CH2, q, JCF = 37), 113.5
1
(2xCH), 119.2 (CH), 122.4 (CF3, q, JCF = 276), 129.8 (2 ꢂ CH),
147.0 (C), 170.4 (C). Mp 65–66 °C. HRMS ([M+H]+, ESI): m/z calcd
for C10H11NO2F3: 234.0736. Found: 234.0737.
5.4.1. 2-Ethanamido-N-(4-phenylbutan-2-yl)ethanamide 2b
From 234 mg (2 mmol) of N-acetyl glycine, the racemic amide
2b (320 mg, 1.3 mmol, 65%) was obtained after recrystallization
(hexane/chloroform). 1H NMR (200 MHz, CDCl3): 1.17 (d, 3H,
J = 6.6), 1.78 (m, 2H), 2.04 (s, 3H), 2.62 (m, 2H), 3.84 (m, 2H),
4.06 (m, 1H), 6.00 (d, 1H, J = 7.8, NH), 6.40 (broad m, 1H, NH),
7.17–7.28 (m, 5H). 13C NMR (75 MHz, CDCl3): 20.9 (CH3), 23.0
(CH3), 32.5 (CH2), 38.4 (CH2), 43.6 (CH2), 45.5 (CH), 126.0 (CH),
128.4 (2 ꢂ CH), 128.5 (2 ꢂ CH), 141.7 (C), 168.4 (C), 170.9 (C). Mp
106–107 °C. HRMS ([M+H]+, ESI): m/z calcd for C14H21N2O2:
249.1598. Found: 249.1602. HPLC conditions: Chiralcel OD-H, n-
hexane/ethanol 95:5, flow rate = 1 mL/min, UV 254 nm, tR
5.2.8. N-(9-Phenyl)nonanoyl glycine trifluoroethyl ester 3q
O
Ph-nC8H16-COCl
OH
Glycine
Ph
N
H
68%
D
O
DCC
O
CF3
O
CF3CH2OH
N
H
Ph
3r
89%
O
5.2.8.1. N-(9-Phenyl)nonanoyl glycine D.
The procedure is
the same as that for C, using 1 equiv of N-(9-phenyl)nonanoyl chlo-
ride (3.97 mmol, 1 g) and 1.2 equiv of glycine (4.76 mmol, 357 mg).
Compound D (780 mg, 2.6 mmol, 68%) was obtained as a white
powder. 1H NMR (200 MHz, CDCl3): 1.34 (m, 8H), 1.59 (m, 4H),
2.24 (t, 2H, J = 7.5), 2.58 (t, 2H, J = 8.0), 4.01 (d, 2H, J = 5.0), 6.55
(m, 1H, NH), 7.17–7.26 (m, 5H), 9.32 (broad s, 1H, COOH). 13C
NMR (75 MHz, CDCl3): 25.9 (CH2), 29.6 (CH2), 29.6 (CH2), 29.7
(CH2), 29.7 (CH2), 31.9 (CH2), 36.4 (CH2), 36.7 (CH2), 41.9 (CH2),
126.0 (CH), 128.6 (2xCH), 128.8 (2xCH), 143.3 (C), 173.0 (C),
175.0 (C). Mp 101–102 °C. HRMS ([M+H]+, ESI): m/z calcd for
C17H26NO3: 292.1907. Found: 292.1904.
(S) = 9 min, tR (R) = 13.2 min, k(S) = 1.95, k(R) = 3.33,
a = 1.71,
Rs = 2.1. An authentic sample of (S)-amide (ee >99.5), prepared
from the kinetic resolution of (S)-3-amino-1-phenyl butane, was
used to determine the ½a D25
¼ þ3:9 (c 1.08, CHCl3).
ꢃ
5.4.2. N-(2-Oxo-2-(4-phenylbutan-2-ylamino)ethyl)benzamide
2c
From 358 mg (2 mmol) of hippuric acid, and after purification
by flash-chromatography (diethyl ether/methanol (90:10)), the
racemic amide 2c (328 mg, 1.1 mmol, 53%) was isolated. 1H NMR
(200 MHz, CDCl3): 1.26 (d, 3H, J = 6.6), 1.80 (m, 2H), 2.63 (m, 2H),
4.05 (sept, 1H, J = 6.8), 4.16 (d, 2H, J = 5.0), 7.00 (br d, 1H, J = 6.6,
NH), 7.10–7.30 (m, 5H), 7.35–7.50 (m, 3H), 7.55 (m, 1H, NH),
7.85 (m, 2H). 13C NMR (50 MHz, CDCl3): 21.7 (CH3), 33.3 (CH2),
39.2 (CH2), 44.8 (CH2), 46.3 (CH), 126.7 (CH), 127.9 (CH), 128.9
(CH), 129.1 (CH), 129.2 (CH), 129.4 (CH), 132.6 (CH), 134.3 (C),
142.4 (C), 168.5 (C), 168.9 (C). Mp 115–117 °C. HRMS ([M+H]+,
ESI): m/z calcd for C19H23N2O2 : 311.1754. Found: 311.1756. HPLC
conditions: Chiralpak AD, n-hexane/isopropanol 95:5, flow
rate = 1 mL/min, UV 254 nm, tR (S) = 13.8 min, tR (R) = 14.9 min,
5.2.8.2. N-(9-Phenyl)nonanoyl glycine trifluoroethyl ester
3q.
The procedure is the same as that for N-octanoyl L-phen-
ylalanine trifluoroethyl ester, using D (2.2 mmol, 640 mg). Com-
pound 3q (727 mg, 1.9 mmol, 89%) was obtained as a white
powder. 1H NMR (200 MHz, CDCl3): 1.31 (m, 8H), 1.62 (m, 4H),
2.25 (t, 2H, J = 7.7), 2.59 (t, 2H, J = 8.0), 4.15 (d, 2H, J = 5.5), 4.53
(q, 2H, J = 8.3), 5.98 (m, NH), 7.15–7.27 (m, 5H). 13C NMR
(75 MHz, CDCl3): 26.5 (CH2), 30.2 (CH2), 30.3 (CH2), 30.4 (CH2),
32.5 (CH2), 36.9 (CH2), 37.3 (CH2), 41.9 (CH2), 62.0 (CH2, q,
1
2JCF = 37), 121.9 (CF3, q, JCF = 275), 126.6 (CH), 129.2 (2 ꢂ CH),
k(S) = 3.52, k(R) = 3.89, a = 1.10, Rs = 1.4. An authentic sample of
(S)-amide (ee >99.5), prepared from the kinetic resolution of (S)-
129.4 (2 ꢂ CH), 143.9 (C), 169.8 (C), 174.7 (C). Mp 83–85 °C. HRMS
([M+H]+, ESI): m/z calcd for C19H27NO3F3: 373.1938. Found:
373.1939.
3-amino-1-phenyl butane, was used to determine the
½
a 2D5
ꢃ
¼ þ1:3 (c 0.83, CHCl3).
5.3. General procedure for the kinetic resolution of 1
5.4.3. N-(1-Oxo-3-phenyl-1-(4-phenylbutan-2-ylamino)propan-
2-yl)octanamide 2d
To a solution of the acyl donor (0.125 mmol) in 3-methyl-3-
From 582 mg (2 mmol) of N-octanoyl L-phenylalanine (A), and
pentanol (250
l
L), 6 mg of coated protease and 19 mg
after flash-chromatography (ethyl ether 100%), a mixture of diaste-
reomeric (S,R)- and (S,S)-amides (438 mg, 1.04 mmol, 52%) was ob-
tained. 1H NMR of the mixture (50/50 ratio) (200 MHz, CDCl3): 0.85
(two superimposed t, 3H, J = 6.5), 0.95 (d, 1.5H, J = 6.6), 1.07 (d,
1.5H, J = 6.6), 1.30 (br s, 8H), 1.43–1.66 (m, 4H), 2.16 (t, 2H,
J = 7.2), 2.44 (pseudo t, 1H, J = 8.0), 2.54 (pseudo t, 1H, J = 8.0),
2.87–3.13 (m, 2H), 3.83–3.98 (m, 1H), 4.55–4.71 (two superim-
(0.125 mmol) of 3-amino-1-phenyl butane (1) were added. The
resulting mixture was stirred at 21 °C. The kinetic resolution was
monitored by analyzing aliquots at different time intervals. The
amide ee was determined by HPLC after dilution in the eluting sol-
vent. The amine ee was determined by GC after dilution in diethyl
ether and derivatization in trifluoroacetamide using 1.5 equiv of N-

2832
A.-L. Bottalla et al. / Tetrahedron: Asymmetry 20 (2009) 2823–2834
posed q, 1H, J = 7.9), 5.82 (d, 0.5H, J = 8.3, NH), 6.05 (d, 1H, J = 8.4,
NH), 6.35 (d, 1H, J = 8.9, NH), 6.39 (superimposed d, 1H, J = 8.9,
NH), 7.07–7.31 (m, 10H). Mp 126–127 °C. HRMS ([M+H]+, ESI):
m/z calcd for C27H39N2O2: 423.3006. Found: 423.3006. HPLC condi-
tions: Chiralpak IC, n-hexane/isopropanol 97:3, flow rate = 1 mL/
min, UV 220 nm, tR (S,R) = 11.0 min, tR (S,S) = 13.4 min,
0.88 (t, 3H, J = 6.5), 1.17 (d, 3H, J = 6.6), 1.29 (br s, 8H), 1.38 (d, 3H,
J = 7.0), 1.72 (m, 4H), 2.15 (pseudo t, 2H, J = 7.5), 2.65 (pseudo t, 2H,
J = 8.0), 4.01 (m, 1H), 4.47 (quint, 1H, J = 7.0), 6.21 (d, 1H, J = 7.6,
NH), 6.32 (d, 1H, J = 8.3, NH), 7.28–7.41 (m, 5H). 13C NMR
(50 MHz, CDCl3): 14.8 (CH3), 19.4 (CH3), 21.6 (CH3), 23.4 (CH2),
26.4 (CH2), 29.8 (CH2), 30.0 (CH2), 32.4 (CH2), 33.2 (CH2), 37.4
(CH2), 39.8 (CH2), 46.0 (CH), 49.6 (CH), 126.7 (CH), 129.2
(2 ꢂ CH), 129.2 (2 ꢂ CH), 142.4 (C), 172.4 (C), 173.9 (C).
k(S,R) = 2.61, k(S,S) = 3.39,
a = 1.30, Rs = 1.6. An authentic sample
of (S,S)-amide (ee >99.5), prepared from the kinetic resolution of
(S)-3-amino-1-phenyl butane, was used to determine the
½
a 2D5
ꢃ
¼ ꢀ2:3 (c 0.83 in CHCl3). (S,S)-2d: 1H NMR (200 MHz, CDCl3):
5.4.6. N-(2-Oxo-2-(4-phenylbutan-2-ylamino)ethyl)-8-
phenyloctanamide 2g
0.85 (t, 3H, J = 6.5), 1.05 (d, 3H, J = 6.6), 1.30 (m, 8H), 1.55 (m, 4H),
2.15 (t, 2H, J = 7.2), 2.44 (pseudo t, 2H, J = 8.0), 3.05 (m, 2H), 3.89
(sept, 1H, J = 6.6), 4.62 (q, 1H, J = 7.7), 5.87 (d, 1H, J = 8.6, NH),
6.34 (d, 1H, J = 7.8, NH), 7.07–7.31 (m, 10H). 13C NMR (CDCl3):
14.8 (CH3), 21.4 (CH3), 23.4 (CH2), 26.4 (CH2), 29.7 (CH2), 29.9
(CH2), 31.7 (CH2), 32.4 (CH2), 37.4 (CH2), 38.9 (CH2), 39.4 (CH2),
46.1 (CH), 55.6 (CH), 126.6 (CH), 127.7 (CH), 129.1 (4 ꢂ CH),
129.4 (2 ꢂ CH), 130.1 (2 ꢂ CH), 137.6 (C), 142.3 (C), 170.9 (C),
173.9 (C).
Synthesized by kinetic resolution, starting from 0,125 mmol
(45 mg) of 3g. After purification by flash-chromatography (dichlo-
romethane/methanol (95:05)), racemic 2g (27 mg, 0.066 mmol,
53%) was obtained. 1H NMR (200 MHz, CDCl3): 1.16 (d, 3H,
J = 6.6), 1.30 (m, 8H), 1.56 (m, 2H), 1.71 (q, 2H, J = 6.9), 2.17 (t,
2H, J = 7.3), 2.54 (m, 4H), 3.86 (d, 2H, J = 5.0), 3.98 (m, 1H), 6.33
(m, 2H, NH), 7.14–7.30 (m, 10H). 13C NMR (75 MHz, CDCl3): 21.4
(CH3), 26.0 (CH2), 29.5 (CH2), 29.6 (CH2), 31.8 (CH2), 32.9 (CH2),
36.3 (CH2), 36.8 (CH2), 38.8 (CH2), 43.9 (CH2), 45.8 (CH), 125.9
(CH), 126.3 (CH), 128.6 (2 ꢂ CH), 128.7 (2 ꢂ CH), 128.8 (2 ꢂ CH),
128.9 (2 ꢂ CH), 142.1 (C), 143.2 (C), 168.6 (C), 174.1 (C). HRMS
([M+H]+, ESI): m/z calcd for C26H37N2O2: 409.2850. Found:
409.2851. HPLC conditions: Chiralpak AD-H, n-hexane/ethanol
80:20, flow rate = 1 mL/min, UV 254 nm, tR (S) = 15.2 min, tR
5.4.4. 2-Ethanamido-3-phenyl-N-(4-phenylbutan-2-
yl)propanamide 2e
From 307 mg (1.5 mmol) of N-acetyl
L-phenylalanine, and after
flash-chromatography (dichloromethane/methanol (95:05)),
a
mixture of diastereomeric (S,R)- and (S,S)-amides (379 mg,
1.1 mmol, 75%) was obtained. 1H NMR of the mixture (60/40 ratio)
(200 MHz, CDCl3): 0.92 (d, 1.8H, J = 6.6), 1.07 (d, 1.2H, J = 6.7),
1.48–1.68 (m, 2H), 1.95 (s, 3H), 2.40 (m, 0.8H), 2.55 (m, 1.2H),
2.80–3.14 (m, 2H), 3.75–3.96 (m, 1H), 4.50–4.64 (m, 1H), 5.64 (d,
0.6H, J = 8.2, NH), 5.90 (d, 0.4H, J = 8.2, NH), 6.46 (d, 0.6H, J = 8.1,
NH), 6.52 (superimposed d, 0.4H, J = 9.5, NH), 7.07–7.32 (m,
10H). Mp 175–176 °C. HRMS ([M+H]+, ESI): m/z calcd for
C21H27N2O2 : 339.2067. Found: 339.2066. HPLC conditions: Sepa-
pak-2HR, n-hexane/isopropanol 90:10, flow rate = 1 mL/min, UV
220 nm, tR (S,S) = 10.7 min, tR (S,R) = 12.7 min, k(S,S) = 2.51,
(R) = 18.7 min, k(S) = 3.98, k(R) = 5.13,
a = 1.29, Rs = 1.8. An authen-
tic sample of (S)-amide (ee >99.5), prepared from the kinetic reso-
lution of (S)-3-amino-1-phenyl butane, was used to determine the
½
a 2D5
ꢃ
¼ þ3:1 (c 1.03, CHCl3).
5.4.7. 2-(Phenylamino)-N-(4-phenylbutan-2-yl)ethanamide 2h
This racemic amide was synthesized by enzymatic reaction. To
a solution of N-phenyl glycine trifluoroethyl ester 3h (58 mg,
0.25 mmol) in 500 lL of 3-methyl-3-pentanol, 12 mg of coated
Protex 50FP and 37 mg (0.25 mmol) of 1 were added. The resulting
mixture was stirred for 24 h at 21 °C. After evaporation and purifi-
cation by flash-chromatography (ether 100%), racemic 2h (53 mg,
0.19 mmol, 76%) was obtained. 1H NMR (200 MHz, CDCl3): 1.12
(d, 3H, J = 6.6), 1.70 (m, 2H), 2.57 (t, 2H, J = 8.0), 3.60 (br s, 1H),
3.65 (s, 2H), 4.1 (m, 1H), 6.49 (d, 1H, J = 8.0, NH), 6.60 (m, 2H),
6.82 (m, 1H), 7.25 (m, 7H). 13C NMR (75 MHz, CDCl3): 21.9 (CH3),
33.3 (CH2), 39.2 (CH2), 45.8 (CH), 49.8 (CH2), 114.1 (2xCH), 120.0
(CH), 126.6 (CH), 129.1 (2 ꢂ CH), 129.2 (2 ꢂ CH), 130.2 (2 ꢂ CH),
142.6 (C), 147.9 (C), 170.5 (C). Mp 112–113 °C. HRMS ([M+H]+,
ESI): m/z calcd for C18H23N2O: 283.1805. Found: 283.1803. HPLC
conditions: Chiralcel OD-H, n-hexane/ethanol 80:20, flow
rate = 1 mL/min, UV 254 nm, tR (S) = 11.7 min, tR (R) = 14.3 min,
k(S,R) = 3.16,
a = 1.26, Rs = 1.3. An authentic sample of (S,S)-amide
(ee >99.5), prepared from the kinetic resolution of (S)-3-amino-1-
phenyl butane, was used to determine the ½a D25
¼ ꢀ3:4 (c 1.01,
ꢃ
CHCl3). (S,S)-2e: 1H NMR (200 MHz, CDCl3): 1.06 (d, 3H, J = 6.6),
1.53 (q, 2H, J = 6.5), 1.96 (s, 3H), 2.38 (m, 2H), 2.96 (dd, 1H,
J = 13.5 and 8.6), 3.10 (dd, 1H, J = 13.5 and 6.1), 3.88 (quint, 1H,
J = 6.5), 4.58 (pseudo q, 1H, J = 7.0), 5.60 (d, 1H, J = 8.4, NH), 6.35
(d, 1H, J = 7.5, NH), 7.07–7.32 (m, 10H). 13C NMR (75 MHz, CDCl3):
20.6 (CH3), 23.1 (CH3), 32.0 (CH2), 38.0 (CH2), 38.7 (CH2), 45.2 (CH),
54.9 (CH), 125.8 (CH), 126.9 (CH), 128.1 (2 ꢂ CH), 128.2 (2 ꢂ CH),
128.6 (2 ꢂ CH), 129.1 (2 ꢂ CH), 136.8 (C), 141.6 (C), 170.0 (C),
170.1 (C).
k(S) = 2.84, k(R) = 3.69,
a = 1.30, Rs = 1.7. An authentic sample of
5.4.5. N-(1-Oxo-1-(4-phenylbutan-2-ylamino)propan-2-
yl)octanamide 2f
(S)-amide (ee >99.5), prepared from the kinetic resolution of (S)-
3-amino-1-phenyl butane, was used to determine the
From 215 mg (1 mmol) of B, and after a flash-chromatography
(dichloromethane/methanol (95:05)), a mixture of diastereomeric
(S,R)- and (S,S)-amides (149 mg, 0.43 mmol, 43%) was obtained.
1H NMR of the mixture (50/50 ratio) (300 MHz, CDCl3): 0.88 (t,
3H, J = 6.5), 1.17 (d, 3H, J = 6.6), 1.29 (br s, 8H), 1.38 (d, 3H, ,
J = 7.0), 1.55–1.75 (m, 4H), 2.15 (pseudo t, 2H, J = 7.5), 2.60 (pseudo
t, 2H, J = 7.9), 3.99 (pseudo sept, 1H, J = 7.0), 4.44 (quint, 1H, J = 7.0),
6.09 (two superimposed d, 1H, J = 7.8, NH), 6.23 (two superim-
posed d, 1H, J = 7.6, NH), 7.28–7.41 (m, 5H). HRMS ([M+H]+, ESI):
m/z calcd for C21H35N2O2 : 347.2693. Found: 347.2692. HPLC con-
ditions: Chiralcel OD-H, n-hexane/ethanol 98:2, flow rate = 1 mL/
min, UV 254 nm, tR (S,S) = 10.8 min, tR (S,R) = 12.7 min,
½
a 2D5
ꢃ
¼ þ3:1 (c 1.21, CHCl3).
5.4.8. N-(2-Oxo-2-(4-phenylbutan-2-ylamino)ethyl)-9-
phenylnonanamide 2q
Synthesized by kinetic resolution, starting from 0.25 mmol
(94 mg) of 3q. After evaporation and purification by flash-chroma-
tography (dichloromethane/methanol (95:05)), racemic 2q (78 mg,
0.18 mmol, 74%) was obtained as a colorless oil. 1H NMR (200 MHz,
CDCl3): 1.15 (d, 3H, J = 6.5), 1.28 (m, 8H), 1.59 (m, 4H), 1.77 (q, 2H,
J = 7.4), 2.17 (t, 2H, J = 7.5), 2.55 (m, 4H), 3.87 (d, 2H, J = 5.1), 4.04
(m, 1H), 6.63 (m, 2H, NH), 7.18–7.26 (m, 10H). 13C NMR
(75 MHz, CDCl3): 21.7 (CH3), 26.4 (CH2), 30.0 (2 ꢂ CH2), 32.2
(CH2), 32.4 (CH2), 33.2 (CH2), 36.7 (2 ꢂ CH2), 37.1 (CH2), 39.1
(CH2), 44.3 (CH2), 46.3 (CH), 126.3 (CH), 126.7 (CH), 127.0 (CH),
129.0 (CH), 129.1 (2 ꢂ CH), 129.2 (3 ꢂ CH), 129.3 (CH), 142.4 (C),
143.6 (C), 169.1 (C°, 174.7 (C). HRMS ([M+H]+, ESI): m/z calcd for
k(S,R) = 2.54, k(S,S) = 3.16,
a = 1.24, Rs = 1.3. An authentic sample
of (S, S)-amide (ee >99.5), prepared from the kinetic resolution of
(S)-3-amino-1-phenyl butane, was used to determine the
½
a 2D5
ꢃ
¼ ꢀ45:4 (c 1.00 in CHCl3). (S,S)-2f: 1H NMR (300 MHz, CDCl3):

A.-L. Bottalla et al. / Tetrahedron: Asymmetry 20 (2009) 2823–2834
2833
C27H39N2O2: 423.3006. Found: 423.3006. HPLC conditions: Chir-
alpak AD, n-hexane/ethanol 80:20, flow rate = 1 mL/min, UV
254 nm, tR (S) = 14.7 min, tR (R) = 19.2 min, k(S) = 3.82, k(R) = 5.30,
set to 2,500,000 and other parameters left to their default values
(nb: using a number of evaluations 20 times larger did not signif-
icantly improve the results). The 200 lowest energy structures cor-
responding to each 21 ligand/enzyme couples were stored for
analysis.
As this targeted approach was not able to highlight the
docked conformations with a proper preorganized structure, we
considered it was not worthwhile using several enzyme
structures.
a
= 1.39, Rs = 2.6. An authentic sample of (S)-amide (ee >99.5), pre-
pared from the kinetic resolution of (S)-3-amino-1-phenyl butane,
was used to determine the ½a D25
ꢃ
¼ þ2:1 (c 0.99, CHCl3).
5.5. Blind docking method
For the sake of saving time, the receptor was maintained rigid
during the docking, only the degrees of freedom of the ligand were
considered, that is, translation of its center of gravity; orientation;
and torsion around each rotatable bond (it is generally assumed
that bond lengths and the angles formed by adjacent bonds do
not change). This is the reason why a sampling of five different
structures was used to roughly evaluate the incidence of enzyme
conformational changes on its affinity for the ligands. Five protein
structures were selected from the PDB, that is, 1SNB, 1SIB, 2SBT,
2SNI, and 1SBT (1SIB and 1SBN structures are identical; only the
resolution is different).
Acknowledgments
Gifts of Alkaline protease, Validases FP500 and AFP1000L from
Valley Research, of Subtilisin Novo from Novozymes, and gifts of
Protex 26L, 50FP, 51FP, 14L, 15L, 7L, 30L, 89L, 6L, 40XL, 40L, Prop-
erase 1600L from Genencor, are gratefully acknowledged. This
work was supported by the Agence Nationale de la Recherche
(ANR 06-Blan-0137).
References
All structures have been cleaned from any foreign molecule
(solvent, inhibitor, ions, etc.). Then, hydrogen atoms and other
1. (a) Nechab, M.; Azzi, N.; Vanthuyne, N.; Bertrand, M. P.; Gastaldi, S.; Gil, G. J.
Org. Chem. 2007, 72, 6918; (b) Nechab, M.; El Blidi, L.; Vanthuyne, N.; Gastaldi,
S.; Bertrand, M. P.; Gil, G. Org. Biomol. Chem. 2008, 6, 3917.
2. Gastaldi, S.; Escoubet, S.; Vanthuyne, N.; Gil, G.; Bertrand, M. P. Org. Lett. 2007,
9, 837.
3. The spatial arrangement of the catalytic triad in the protease active site is the
mirror image of that in lipases. As a consequence, their enantiopreference is
opposite to that of lipases. See: (a) Fitzpatrick, P. A.; Klibanov, A. M. J. Am. Chem.
Soc. 1991, 113, 3166; (b) Kazlauskas, R. J.; Weissfloch, A. N. E. J. Mol. Catal. B:
Enzym. 1997, 3, 65; (c) Savile, C. K.; Kazlauskas, R. J. J. Am. Chem. Soc. 2005, 127,
2104.
missing atoms were added using the ADT graphical interface of
26
AUTODOCK
,
or the MOLPROBITY program.29 A short energy minimiza-
tion of the structures was further performed with the TINKER pro-
gram,30 and AMBER99 force field.31 The latter was selected
32
because it is very close to the force field used in AUTODOCK
.
It can
be noted that at this stage 1SIB and 1SBN structures became
slightly different.
The program ADT was used to assign partial charges to all the
atoms, according to Gasteiger method,[33] it enabled us to create
a unified atom model for the enzyme. The Autogrid program,
which is part of the AUTODOCK package produced a potential energy
mapping for interactions of the ligands with the protein embedded
in three-dimensional grid that contained 60 ꢂ 60 ꢂ 60 grid points
spaced by 0.703 Å.
The ligand structures have been created from AGUI, the graph-
ical interface of the AMPAC program.34 Their energy was minimized
with AMPAC using the semi-empirical AM1 method. They were then
processed with ADT, to assign Gasteiger partial charges and produce
a united atom model for each of them.
4. For other examples of protease-mediated amine acylation, see: (a) Kitagushi,
H.; Fitzpatrick, P. A.; Huber, J. E.; Klibanov, A. M. J. Am. Chem. Soc. 1989, 111,
3094; (b) Gutman, A. L.; Meyer, E.; Kalerin, E.; Polyack, F.; Sterling, J. Biotechnol.
Bioeng. 1992, 40, 760; (c) Orsat, B.; Alper, P. B.; Moree, W.; Mak, C.-P.; Wong, C.-
H. J. Am. Chem. Soc. 1996, 118, 712; (d) Wang, Y. F.; Yakolevsky, K.; Margolin, A.
L. Tetrahedron Lett. 1996, 37, 5317; (e) Wang, Y. F.; Yakolevsky, K.; Zhang, B.;
Margolin, A. L. J. Org. Chem. 1997, 62, 3488; (f) Takayama, S.; Moree, W. J.;
Wong, C.-H. Tetrahedron Lett. 1996, 37, 6287.
5. Routaboul, L.; Vanthuyne, N.; Gastaldi, S.; Gil, G.; Bertrand, M. P. J. Org. Chem.
2008, 73, 364.
6. El Blidi, L.; Nechab, M.; Vanthuyne, N.; Gastaldi, S.; Gil, G.; Bertrand, M. P. J. Org.
Chem. 2009, 74, 2901–2903.
7. For structural comparisons of subtilisins and some alkaline proteases, see: (a)
McPhalen, C. A.; James, M. N. G. Biochemistry 1988, 27, 6582; (b) Schubert
Wright, C. S.; Alden, R. A.; Kraut, J. K. Nature 1969, 221, 235; (c) Miyatake, H. H.;
Hata, Y.; Fujii, T.; Hamada, K.; Morihara, K.; Katsube, Y. J. Biochem. 1995, 118,
474; (d) van der Laan, J. M.; Teplyakov, A. V.; Kelders, H.; Kalk, K. H.; Misset, O.;
Mulleners, L. J. S. M.; Dijkstra, B. W. Protein Eng. 1992, 5, 405; For a mini review,
see also: (e) Gupta, R.; Beg, Q. K.; Lorenz, P. Appl. Microbiol. Biotechnol. 2002, 59,
15. and references cited therein.
All the ligands were docked on each of the five enzyme struc-
tures using the Lamarckian genetic Algorithm implemented in
AUTODOCK with the number of evaluations set to 2,500,000 and other
parameters left to their default values. The 100 lowest energy
structures corresponding to each 5 ꢂ 21 ligand/enzyme couples
8. Rao, M. B.; Tanksale, A. M.; Ghatge, M. S.; Deshpande, V. V. Microbiol. Mol. Biol.
Rev. 1998, 62, 597.
9. For a general review, see: Bordusa, F. Chem. Rev. 2002, 102, 4817.
10. For the use of coated subtilisin in alcohol DKR, see: (a) Borén, L.; Martín-
Matute, B.; Xu, Y.; Córdova, A.; Bäckvall, J.-E. Chem. Eur. J. 2006, 12, 225; (b)
Kim, M.-J.; Chung, Y. I.; Choi, Y. K.; Li, H. K.; Kim, D.; Park, J. J. Am. Chem. Soc.
2003, 125, 11494; See also: (c) Martinez, S. G.; Alvira, E.; Vergara Cordero, L.;
Ferrer, A.; Montanés-Clemente, I.; Barletta, G. Biotechnol. Prog. 2002, 18, 1462.
11. E-Values were calculated according to: Chen, C.-S.; Fujimoto, Y.; Girdaukas, G.;
Sih, C. J. Am. Chem. Soc. 1982, 104, 7294.
12. No enantioselectivity was observed with these enzymes, even under different
experimental conditions.
13. Bornscheuer, U. T.; Kazlauskas, R. J. In Hydrolases in Organic Synthesis, Regio-
and Stereoselective Biotransformations, 2nd ed.; Wiley-VCH: Weinheim, 2006;
Chapter 6 and references cited therein.
were stored for analysis. Each docked conformation was assigned
32
a binding energy by AUTODOCK
.
To make the statistics of Table 4
and Figures 4-6, the ligands position was plotted by the coordi-
nates of the carbonyl at the scissile bond in the spherical coordi-
nate system of the enzyme. Each residue was represented by the
coordinates of its geometric center. The different ligands confor-
mations were gathered into clusters on this spherical coordinates
map. We have thus performed statistics about these clusters and
named them according to the closest residues of the enzyme.
14. Trifluoroethanol is known as a good leaving group for enzymatic acylations, for
a general review on acyl donors, see: Hanefeld, U. Org. Biomol. Chem. 2003, 1,
2405. and relevant references cited therein.
15. Schechter, I.; Berger, A. Biochem. Biophys. Res. Commun. 1967, 27, 157.
16. Graham, L. D.; Haggett, K. D.; Jennings, P. A.; Le Brocque, D. S.; Whittaker, R. G.;
Schober, P. A. Biochemistry 1993, 32, 6250.
5.6. Targeted docking method
The enzyme 2SNI was used for docking in the active site. This
structure has been cleaned from any foreign molecule (solvent,
inhibitor, ions, etc.) and hydrogen atoms were added with the MOL-
PROBITY program,30 and their positions optimized. The grid used in
AUTODOCK was centered on the active site and contained
60 ꢂ 60 ꢂ 60 grid points spaced by 0.375 Å. All the ligands were
docked on this enzyme structure using the Lamarckian genetic
algorithm implemented in AUTODOCK with the number of evaluations
17. The use of N-protected amino acid carbamoylesters as acyl donors was also
shown to accelerate the rate the
a-CT-catalyzed amidation of chiral amines.
See: Miyazawa, T.; Yabuuchi, N.; Yanagihara, R.; Yamada, T. Biotechnol. Lett.
1999, 21, 725.
18. For another application of the concept of substrate mimetics, see: Günther, R.;
Elsner, C.; Schmidt, S.; Hofmann, H.-J.; Bordusa, F. Org. Biomol. Chem. 2004, 2,
1442.

2834
A.-L. Bottalla et al. / Tetrahedron: Asymmetry 20 (2009) 2823–2834
19. For each acyl donor conversion was monitored by GC and HPLC. No significant
incidence of conversion rate on the calculated E-value was noticed.
20. With donor 3a, 5% conversion was detected in 2 h (14% in 5 h) for the non
catalyzed reaction performed in 3-methyl-3-pentanol at 21 °C at 0.5 M
concentration. With donor 3d, under the same experimental conditions 2%
conversion was registered in 2 h (4% in 5 h). With donor 3f, the conversion was
2% in 2 h (15% in 5 h).
25. Attempts to measure KM with achiral amines like NH3 or BuNH2 failed due to
the importance of non-catalyzed reaction.
26. Morris, G. M.; Goodsell, D. S.; Halliday, R. S.; Huey, R.; Hart, W. E.; Belew, R. K.;
Olson, A. J. J. Comput. Chem. 1998, 19, 1639.
27. For theoretical approach to reorganization and preorganization in enzymes
active sites, see: Smith, A. T. J.; Müller, R.; Toscano, M. D.; Kast, P.; Hellinga, H.
W.; Hilvert, D.; Houk, K. J. Am. Chem. Soc. 2008, 130, 15361.
28. Veld, M. A. J.; Fransson, L.; Palmans, A. R. A.; Meijer, E. W.; Hult, K.
ChemBioChem 2009, 10, 1330.
21. (a) Wells, J. A.; Cunningham, B. C.; Graycar, T. P.; Estell, D. A. Proc. Natl. Acad. Sci.
U.S.A. 1987, 84, 5167; (b) Estell, D. A.; Graycar, T. P.; Miller, J. V.; Powers, D. B.;
Wells, J. A.; Burnier, J. P.; Ng, P. G. Science 1986, 233, 659.
29. Davis, I. W.; Leaver-Fay, A.; Chen, V. B.; Block, J. N.; Kapral, G. J.; Wang, X.;
Murray, L. W.; Snoeyink, W. B. A., III; Richardson, J. S.; Richardson, D. C. Nucleic
Acids Res. 2007, 35, W375.
0
22. An optimal combined volume of 160 ÅA3 for the S1 and P1 side chains was
deduced from the reactivity of subtilisin mutants, see: Perona, J. J.; Craik, C. S.
Protein Sci. 1995, 4, 337.
30. See: the website at http://dasher.wustl.edu/tinker.
23. For a complementary study, see: Grøn, H.; Medal, M.; Breddam, K. Biochemistry
1992, 31, 6011; Bech, L. M.; Sørensen, S. B.; Breddam, K. Eur. J. Biochem. 1992,
209, 869.
24. Only the natural (S)-enantiomer substrate was recognized by the enzyme. The
resolution performed over 90 min with 2 equiv of racemic acyl donor, instead
of 1 equiv (cf. Table 2, entry 8), led to the following results: amine ee (50.1%);
2f de (97.0%), C (34.1%), E (102). The ee of the remaining donor was 22.8%.
31. (a) Wang, J.; Cieplak, P.; Kollman, P. A. J. Comput. Chem. 2000, 21, 1049; (b)
Moyna, G.; Williams, H. J.; Nachman, R. J.; Scott, A. I. Biopolymers 1999, 49, 403.
32. Huey, R.; Morris, G. M.; Olson, A. J.; Goodsell, D. S. J. Comput. Chem. 2007, 28,
1145.
33. Gasteiger, J.; Marsili, M. Tetrahedron 1980, 36, 3219.
34. AMPAC 8, 1992–2004 Semichem, Inc. PO Box 1649, Shawnee, ks 66222.