Subtilisin-catalyzed resolution of N-acyl arylsulfinamides

Article abstract of DOI:10.1021/ja045397b

We report the first biocatalytic route to sulfinamides (R-S(O)-NH ₂), whose sulfur stereocenter makes them important chiral auxiliaries for the asymmetric synthesis of amines. Subtilisin E did not catalyze hydrolysis of N-acetyl or N-butanoyl arylsulfinamides, but did catalyze a highly enantioselective (E > 150 favoring the (R)-enantiomer) hydrolysis of N-chloroacetyl and N-dihydrocinnamoyl arylsulfinamides. Gramscale resolutions using subtilisin E overexpressed in Bacillus subtilis yielded, after recrystallization, three synthetically useful auxiliaries: (R)-p- toluenesulfinamide (42% yield, 95% ee), (R)-p-chlorobenzenesulfinamide (30% yield, 97% ee), and (R)-2,4,6-trimethylbenzenesulfinamide (30% yield, 99% ee). Molecular modeling suggests that the N-chloroacetyl and N-dihydrocinnamoyl groups mimic a phenylalanine moiety and thus bind the sulfinamide to the active site. Molecular modeling further suggests that enantioselectivity stems from a favorable hydrophobic interaction between the aryl group of the fast-reacting (R)-arylsulfinamide and the S₁′ leaving group pocket in subtilisin E.

Full text of DOI:10.1021/ja045397b

Published on Web 01/26/2005
Subtilisin-Catalyzed Resolution of N-Acyl Arylsulfinamides
Christopher K. Savile,† Vladimir P. Magloire, and Romas J. Kazlauskas*,†
Contribution from the Department of Chemistry, McGill UniVersity, 801 Sherbrooke Street West,
Montre´al, Que´bec, Canada H3A 2K6
Received July 30, 2004; E-mail: rjk@umn.edu
Abstract: We report the first biocatalytic route to sulfinamides (R-S(O)-NH2), whose sulfur stereocenter
makes them important chiral auxiliaries for the asymmetric synthesis of amines. Subtilisin E did not catalyze
hydrolysis of N-acetyl or N-butanoyl arylsulfinamides, but did catalyze a highly enantioselective (E > 150
favoring the (R)-enantiomer) hydrolysis of N-chloroacetyl and N-dihydrocinnamoyl arylsulfinamides. Gram-
scale resolutions using subtilisin E overexpressed in Bacillus subtilis yielded, after recrystallization, three
synthetically useful auxiliaries: (R)-p-toluenesulfinamide (42% yield, 95% ee), (R)-p-chlorobenzenesulfin-
amide (30% yield, 97% ee), and (R)-2,4,6-trimethylbenzenesulfinamide (30% yield, 99% ee). Molecular
modeling suggests that the N-chloroacetyl and N-dihydrocinnamoyl groups mimic a phenylalanine moiety
and thus bind the sulfinamide to the active site. Molecular modeling further suggests that enantioselectivity
stems from a favorable hydrophobic interaction between the aryl group of the fast-reacting (R)-arylsulfinamide
and the S1′ leaving group pocket in subtilisin E.
Scheme 1. Synthesis of Enantiopure Amines from Sulfinamides
Introduction
The chiral sulfinyl group (R-S(O)-) is an important
functional group for asymmetric synthesis because it effectively
transfers chirality to a wide range of centers.1-3 This efficacy
stems from the steric and stereoelectronic differences between
the substituents: a lone pair, an oxygen, and an alkyl or aryl
group. As well, the sulfinyl group is configurationally stable.2
Sulfinamides (R-S(O)-NH2) are useful sulfinyl chiral
auxiliaries for synthesis of amines (Scheme 1).4,5 When
condensed with an aldehyde or ketone to give the sulfinimine,
the N-sulfinyl group directs nucleophilic addition across the Cd
N bond. This yields the N-alkyl sulfinamide, which upon
hydrolysis of the S-N link yields an amine. Enantioselective
syntheses using sulfinimines include preparations of amines,6
R- and â-amino acids,5 amino alcohols,7 aziridines,8 and amino
phosphonic acids.9
philic displacement of menthol at chiral sulfur leads to enan-
tiopure aryl sulfinyl compounds4 including p-toluenesulfin-
amide.12 However, preparation of Andersen’s reagent relies on
the crystallization of diastereomers epimeric at sulfur, and only
p-toluenesulfinate and close analogues efficiently crystallize as
the menthyl derivative.2,4,13 Thus, this route is limited to
p-toluenesulfinamide and close analogues. Asymmetric oxida-
tion using 1 equiv of (+)- or (-)-N-(phenylsulfonyl)(3,3-
dichlorocamphoryl)oxaziridine also yields enantiopure aryl
sulfinyl moieties.14
The best route to enantiopure aryl sulfinyl moieties is via
menthyl-p-toluenesulfinate (Andersen’s reagent).2,10,11 Nucleo-
† Current address: Department of Biochemistry, Molecular Biology &
Biophysics and The Biotechnology Institute, University of Minnesota, 1479
Gortner Avenue, Saint Paul, MN 55108.
(1) Carren˜a, M. C. Chem. ReV. 1995, 95, 1717-1760 and references therein.
(2) Anderson, K. K. In The Chemistry of Sulphones and Sulphoxides; Patai,
S., Rappoport, Z., Stirling, C. J. M., Eds.; John Wiley & Sons: New York,
1988; pp 55-94.
(3) Walker, A. J. Tetrahedron: Asymmetry 1992, 3, 961-998.
(4) Nudelman, A. In The Chemistry of Sulphinic Acids, Esters and their
DeriVatiVes; Patai, S., Ed.; John Wiley & Sons: New York, 1990; pp 35-
85.
The two other routes to enantiopure sulfinamides use either
an enantioselective oxidation of tert-butyl disulfide6,15 or a
double displacement using a chiral auxiliary derived from
indanol.16 The enantioselective oxidation developed by Ellman
(11) Drabowicz, J.; Kielbasinski, P.; Mikolajczyk, M. In The Chemistry of
Sulphones and Sulphoxides; Patai, S., Rappoport, Z., Stirling, C. J. M.,
Eds.; John Wiley & Sons: New York, 1988; pp 233-378.
(12) Davis, F. A.; Xhang, Y.; Andemichael, Y.; Fang, T.; Fanelli, D. L.; Zhang,
H. J. Org. Chem. 1999, 64, 1403-1406.
(5) Davis, F. A.; Zhou, P.; Chen, B.-C. Chem. Soc. ReV. 1998, 27, 13-18.
(6) Liu, G.; Cogan, D. A.; Ellman, J. A. J. Am. Chem. Soc. 1997, 119, 9913-
9914.
(13) Herbranson, H. F.; Cusano, C. M. J. Am. Chem. Soc. 1961, 83, 2124-
(7) Kochi, T.; Tang, T. P.; Ellman, J. A. J. Am. Chem. Soc. 2002, 124, 6518-
2128.
6519.
(14) Davis, F. A.; Reddy, R. T.; Reddy, R. E. J. Org. Chem. 1992, 57, 6387-
6389.
(8) Davis, F. A.; Zhou, P.; Reddy, G. V. J. Org. Chem. 1994, 59, 3243-3245.
(9) Lefebvre, I. M.; Evans, S. A. J. Org. Chem. 1997, 62, 7532-7533.
(10) (a) Phillips, H. J. Chem. Soc. 1925, 127, 2552-2587. (b) Hulce, M.;
Mallamo, J. P.; Frye, L. L.; Kogan, T. P.; Posner, G. A. Org. Synth. 1986,
64, 196-206.
(15) Cogan, D. A.; Liu, G.; Kim, K.; Backes, B. J.; Ellman, J. A. J. Am. Chem.
Soc. 1998, 120, 8011-8019.
(16) Han, Z.; Krishnamurthy, D.; Grover, P.; Fang, Q. K.; Senanayake, C. H.
J. Am. Chem. Soc. 2002, 124, 7880-7881.
9
2104
J. AM. CHEM. SOC. 2005, 127, 2104-2113
10.1021/ja045397b CCC: $30.25 © 2005 American Chemical Society

Subtilisin-Catalyzed Resolution of N-Acyl Arylsulfinamides
A R T I C L E S
Scheme 2. Subtilisin-Catalyzed Kinetic Resolution of Sulfinamides
Figure 1. Empirical rules that predict the enantiopreference of subtilisins
toward secondary alcohols and sulfinamides. (a) In organic solvent,
subtilisins favor the secondary alcohol enantiomer with the shape shown,
where L is a large substituent, such as phenyl, and M is a medium
substituent, such as methyl. (b) In water, subtilisins favor the sulfinamide
enantiomer with the shape shown, where Ar is an aryl substituent. We
suggest that a favorable hydrophobic interaction between the aryl substituent
and S1′ pocket and good solvation of the polar sulfoxide oxygen in water
explain the enantiopreference with sulfinamides.
and co-workers relies on the stability of the oxidation product,
thiosulfinate (t-Bu-S(O)-S-t-Bu), to racemization and is thus
limited to tert-butylsulfinamide. The double displacement route
developed by Senanayake and co-workers uses N-sulfonyl-1,2,3-
oxathiazolidine-2-oxide intermediates. This route yields a variety
of enantiopure sulfinamides, but includes complicated steps
requiring low temperatures and moisture- and air-sensitive
reagents.
methylsilyl)amide followed by desilylation gave the correspond-
ing sulfinamides in 21-55% overall yield.20
We prepared tert-butylsulfinamide 8 from tert-butyl disulfide
in three steps. Oxidation of tert-butyl disulfide with 3-chloro-
peroxybenzoic acid followed by addition of sulfuryl chloride
gave tert-butylsulfinyl chloride,22 which reacted with NH4OH
to give 8 in 21% overall yield.20
To prepare N-acylsulfinamides a-m, we treated the appropri-
ate sulfinamide with 2 equiv of n-butyllithium in THF at -78
°C followed by rapid addition of the necessary symmetrical
carboxylic acid anhydride (15-90% yield).20
Initial Screening. Initial screening of the N-butanoyl deriva-
tive 1b with 50 hydrolases revealed no active hydrolases, but
screening the more reactive N-chloroacetyl derivative 1c identi-
fied four moderately to highly active proteases, one lipase and
one esterase (Table 1).23 To determine enantioselectivities, we
carried out small-scale reactions and measured the enantiomeric
purity of the starting materials and products by HPLC using a
column with a chiral stationary phase.
Proteinase from Bacillus subtilis var. biotecus A showed the
highest true enantioselectivity (Etrue > 150). At 50% conversion,
the product, (R)-1, had 93% ee while the unreacted starting
material, (S)-1c, had 92% ee corresponding to an apparent
enantioselectivity (Eapp) of 90. However, this substrate also
underwent ca. 1-3% spontaneous chemical hydrolysis. After
correction for this chemical hydrolysis,24 the true enantioselec-
tivity (Etrue) was >150.
A related protease, subtilisin Carlsberg, also catalyzed hy-
drolysis of 1c with low apparent enantioselectivity (Eapp).
However, the amount of product sulfinamide was much lower
than the amount of 1c that disappeared. Further analysis revealed
that in addition to the expected C-N bond hydrolysis, this
protease also catalyzed S-N bond hydrolysis.25 Details of this
unprecedented reaction will be reported separately.
The two other proteases, protease from Aspergillus oryzae
and protease/acylase from Aspergillus melletus, as well as a
lipase, Candida antarctica lipase A, and bovine cholesterol
esterase, also catalyzed the hydrolysis of 1c, but showed only
low to moderate enantioselectivity (Etrue ) 6-29). All hydro-
lases, except lipase A from Candida antarctica, favored the (R)-
enantiomer.
Many groups have reported biocatalytic routes to sulfinyl
stereocenters, but not to sulfinamides. A direct biocatalytic route
to sulfinyl compounds is oxidation. For example, enantioselec-
tive oxidation of unsymmetrical sulfides yields sulfoxides.11,17
These direct oxidation routes may not be suitable routes to
sulfinamides because oxidation of unsubstituted sulfenamides
(RSNH2) can involve nitrene intermediates.18 Another biocata-
lytic route to enantiopure sulfinyl groups is lipase-catalyzed
hydrolysis of a pendant ester group.19 Enantioselectivity can
be high even when the stereocenter is remote from the reacting
carbonyl. Ellman and co-workers tested the hydrolase-catalyzed
acylation of the sulfinamide amino group, but did not observe
any reaction.15 In this paper, we explore the reverse reaction-
hydrolase-catalyzed hydrolysis of N-acylsulfinamides (Scheme
2) and find that subtilisin E shows high enantioselectivity toward
many arylsulfinamides. Further, molecular modeling suggests
that the high enantioselectivity of subtilisin toward sulfinamides
compared to the moderate enantioselectivity toward the isosteric
secondary alcohols likely stems from the polar nature of the
oxygen substituents in sulfinamides (Figure 1).
Results
Synthesis of Substrates. We prepared racemic sulfinamides
1-8 by one of three established routes (Scheme 3). The simplest
route, from the corresponding sulfinic acid, yielded p-toluene-
sulfinamide 1 and benzenesulfinamide 2. Treating the sulfinic
acid with oxalyl chloride followed by ammonolysis yielded the
sulfinamides in 56-59% yield.20
To prepare sulfinamides for which no sulfinic acid precursors
were available (p-chlorobenzenesulfinamide 3, p-methoxyben-
zenesulfinamide 4, 2,4,6-trimethylbenzenesulfinamide 5, 1-naph-
thylenesulfinamide 6, and 2,4,6-triisopropylbenzenesulfinamide
7), we used the sulfonyl chlorides. Reduction of the sulfonyl
chloride with P(OMe)3 in ethanol gave the respective sulfinate
ethyl esters.21 Displacement of ethanol with lithium bis(tri-
(17) (a) Kagan, H. B. In Catalytic Asymmetric Synthesis; Ojima, I., Ed.; VCH:
New York, 1993; pp 203-226. (b) Holland, H. L. Chem. ReV. 1988, 88,
473-485. (c) ten Brink, H. B.; Holland, H. L.; Schoemaker, H. E.; van
Lingen, H.; Wever, R. Tetrahedron: Asymmetry 1999, 10, 4563-4572.
(18) (a) Capozzi, G.; Modena, G.; Pasquato, L. In The Chemistry of Sulphenic
Acids and Their DeriVatiVes; Patai, S., Ed.; Wiley: Chichester, U.K., 1990;
pp 403-516. (b) Atkinson, R. S.; Judkins, B. D. J. Chem. Soc., Perkin
Trans. 1 1981, 2615-2619.
(22) (a) Netscher, T.; Prinzbach, H. Synthesis 1987, 683-686. (b) Gontcharov,
A. V.; Liu, H.; Sharpless, K. B. Org. Lett. 1999, 1, 783-786.
(23) Janes, L. E.; Lo¨wendahl, C. P.; Kazlauskas, R. J. Chem.sEur. J. 1998, 4,
2324-2331.
(19) (a) Allenmark, S. G.; Andersson, A. C. Tetrahedron: Asymmetry 1993, 4,
2371-2376. (b) Burgess, K.; Henderson, I. Tetrahedron Lett. 1989, 30,
3633-3636. (c) Burgess, K.; Henderson, I.; Ho, K.-K. J. Org. Chem. 1992,
57, 1290-1295. (d) Serreqi, A. N.; Kazlauskas, R. J. Can. J. Chem. 1995,
73, 1357-1367.
(24) A computer program, Selectivity-KRESH, was used for all calculations.
For details, see: Faber, K.; Ho¨nig, H.; Kleewein, A. In PreparatiVe
Biotransformations; Roberts, S. M., Ed.; John Wiley & Sons: New York,
1995; pp 0.079-0.084.
(20) Backes, B. J.; Dragoli, D. R.; Ellman, J. A. J. Org. Chem. 1999, 64, 5472-
(25) 1H NMR and HRMS-EI identified p-toluenesulfinic acid as the primary
product in this enzymatic reaction: Mugford, P. F.; Magloire, V. P.;
Kazlauskas, R. J. Manuscript in preparation.
5478.
(21) Klunder, J. M.; Sharpless, K. B. J. Org. Chem. 1987, 52, 2598-2602.
9
J. AM. CHEM. SOC. VOL. 127, NO. 7, 2005 2105

A R T I C L E S
Savile et al.
Scheme 3. Synthesis of Sulfinamides 1-8 and N-Acylsulfinamides
Table 1. Screening of Hydrolases for Enantioselective Hydrolysis of 1c
enantio-
preferenceh
b
d,e
e,g
hydrolase
wta
%capp
ees (%)c,d
eep (%)c,d
Eapp
x(sp)f (%)
Etrue
B. subtilis var. biotecus A protease
subtilisin E
subtilisin Carlsberg
16
n.d.
34
50
92
56
82
88
30
11
27
93
97
64
68
75
65
67
90
3
2
n.d.
3
10
3
10
>150
>150
n.d.j
17
R
R
R
R
R
S
37i
56
115
n.d.j
14
9
5
7
A. oryzae protease
90
56
A. melletus protease/acylase
C. antarctica lipase A
bovine cholesterol esterase
250
137
80
29k
14
29
6
11
29k
R
a Weight of enzyme in milligrams. b Conversion %: amount of sulfinamide formed in 6 h, except where noted. c Enantiomeric excess %. Enantiomeric
excess of substrate and product was determined by HPLC analysis on Daicel Chiralcel OD or AD columns at 238 or 222 nm. d The eep, ees, and Eapp
(apparent enantioselectivity) are with no correction for chemical hydrolysis; n.d. ) not determined. e Enantiomeric ratio: the enantiomeric ratio E measures
the relative rate of hydrolysis of the fast enantiomer as compared to that of the slow enantiomer as defined by Sih (Chen, C. S.; Fujimoto, Y.; Girdaukas,
g
G.; Sih, C. J. J. Am. Chem. Soc. 1982, 104, 7294-7299). f The x(sp) refers to % spontaneous chemical hydrolysis. Etrue is the enantioselectivity corrected
for chemical hydrolysis.24 h The absolute configuration was determined by comparison to authentic samples prepared by the method of Davis and co-
workers.12 i Reaction for 3 h. j This reaction was complicated by competing sulfinyl hydrolysis. Accurate determination of Eapp and Etrue was more complicated
and will be reported elsewhere. k Reaction for 24 h.
Although this initial screen identified B. subtilis var. biotecus
A as the best commercial enzyme for hydrolysis of 1c, it is,
unfortunately, expensive (U.S. $725/gram), and its amino acid
sequence is not available from the supplier. We suspected that
subtilisin E might be a similar or even the same enzyme.
Subtilisin E is an alkaline serine protease produced by B. subtilis
that has been cloned and overexpressed,26 and its structure has
been elucidated by X-ray crystallography.27 Confirming our
fold lower conversion (28% capp after 24 h), and a slower
spontaneous chemical hydrolysis (4% after 24 h). Glycine
derivative 1f showed at least 3-fold lower enantioselectivity (Etrue
) 54) than did 1c, and 11-fold lower conversion (26% capp after
24 h).
Since subtilisin favors hydrolysis of peptides with aromatic
or large nonpolar residues at the P1 position,29 we prepared
several phenylalanine derivatives or analogues. The phenyl-
alanine derivative 1g demonstrated at least 3-fold lower selectiv-
ity (Dtrue ) 46) and 6-fold lower conversion (46% capp after 24
h) than 1c. However, the N-dihydrocinnamoyl derivative 1h
showed very high enantioselectivity (Etrue > 150) similar to that
of 1c, but with 6-fold lower conversion (47% capp after 24 h).30
Surprisingly, the closely related N-phenoxyacetyl derivative 1i
did not react with enzyme.
hunch, subtilisin E showed similar enantioselectivity (Etrue
>
150) to B. subtilis var. biotecus A in the hydrolysis of 1c (Table
1) as well as four other substrates (see below and Supporting
Information). We focused further experiments on subtilisin E
because it showed high enantioselectivity, it was inexpensive
to produce by fermentation, and the X-ray crystal structure
permitted a molecular-level interpretation of results.
Optimization of N-Acyl Group for Enantioselectivity and
Reactivity. The N-acetyl derivative 1a and N-butanoyl deriva-
tive 1b showed no reaction with subtilisin E,28 but several acyl
groups with electron-withdrawing functional groups similar to
those of 1c showed enantioselective hydrolysis (Table 2). The
N-bromoacetyl derivative 1d reacted similarly to 1c. It showed
high enantioselectivity (Etrue > 150) and conversion (53% capp
in 3 h), but also underwent a similar spontaneous chemical
hydrolysis (4% after 3 h). The N-methoxyacetyl derivative 1e
showed at least 6-fold lower enantioselectivity (Etrue ) 26), 10-
Three nonpolar substrates structurally related to leucine (1j,
1k, and 1l) showed at least 6-, 8-, and 21-fold lower enanti-
oselectivity (Etrue ) 7-24) and 24-, 40-, and 60-fold lower
conversion (5-13% capp after 24 h), respectively, than did 1c.
A similar compound, 1m, demonstrated at least 3-fold lower
enantioselectivity (Etrue ) 52) and 12-fold lower conversion
(24% capp after 24 h) than did 1c.
(29) (a) Estell, D. A.; Graycar, T. P.; Miller, J. V.; Powers, D. B.; Burnier, J.
P.; Ng, P. G.; Wells, J. A. Science 1986, 233, 659-663. (b) Wells, J. A.;
Powers, D. B.; Bott, R. R.; Graycar, T. P.; Estell, D. A. Proc. Natl. Acad.
Sci. U.S.A. 1987, 84, 1219-1223.
(30) Additional screening with the N-dihydrocinnamoyl derivative 1h showed
chymotrypsin from bovine pancreas catalyzed its hydrolysis with good
(26) (a) Park, S.-S.; Wong, S.-L.; Wang, L.-F.; Doi, R. H. J. Bacteriol. 1989,
171, 2657-2665. (b) Zhao, H.; Arnold, F. H. Proc. Natl. Acad. Sci. U.S.A.
1997, 94, 7997-8000.
enantioselectivity (E
) 45), favoring the (R)-enantiomer, and high
(27) Jain, S. C.; Shinde, U.; Li, Y.; Inouye, M.; Berman, H. M. J. Mol. Biol.
1998, 284, 137-144.
conversion (capp ) 52%) after 6 h. This enzyme was missed in our initial
activity screen because it reacted only slowly with the N-chloroacetyl
derivative 1c.
(28) Based on a limit of detection of 0.1%.
9
2106 J. AM. CHEM. SOC. VOL. 127, NO. 7, 2005

Subtilisin-Catalyzed Resolution of N-Acyl Arylsulfinamides
A R T I C L E S
Table 2. Reactivity and Enantioselectivity of Subtilisin E toward
Substrates with Differing N-Acyl Group (1a-m)
Table 3. Enantioselectivity of Subtilisin E toward Substrates with
Differing Sulfinamide Aryl Group
a Conversion %: conversion refers to the amount of sulfinamide formed.
All reactions as N-chloroacetyl were 3 h and as N-dihydrocinnamoyl were
24 h, except where noted; n.r. ) no reaction. b Enantiomeric excess %.
Enantiomeric excess of substrate and product was determined by HPLC
analysis on Daicel Chiralcel OD or AD columns at 238 or 222 nm; n.d. )
not determined, n.a. ) not applicable. c Enantiomeric ratio: the enantiomeric
ratio E measures the relative rate of hydrolysis of the fast enantiomer as
compared to the slow enantiomer as defined by Sih (Chen, C. S.; Fujimoto,
Y.; Girdaukas, G.; Sih, C. J. J. Am. Chem. Soc. 1982, 104, 7294-7299).
d Similar E results were obtained with B. subtilis var. biotecus A (see
Supporting Information Table S1). e Corrected for ca. 1% chemical hy-
drolysis. f Corrected for ca. 3% chemical hydrolysis. g Reaction for 24 h.
a Conversion %: amount of sulfinamide formed in 24 h, except where
noted. b Conversion % per hour (assumes linear rate throughout course of
reaction); n.a. ) not applicable. c Enantiomeric excess %. Enantiomeric
excess of substrate and product was determined by HPLC analysis on Daicel
Chiralcel OD or AD columns at 238 or 222 nm; n.d. ) not determined.
d Enantiomeric ratio: the enantiomeric ratio E measures the relative rate of
hydrolysis of the fast enantiomer as compared to that of the slow enantiomer
as defined by Sih (Chen, C. S.; Fujimoto, Y.; Girdaukas, G.; Sih, C. J. J.
Am. Chem. Soc. 1982, 104, 7294-7299). e Reaction for 3 h. f Corrected
for ca. 2% chemical hydrolysis. g Corrected for ca. 4% chemical hydrolysis.
h Diastereomeric ratio.
1a.28 The conversion was 6-fold lower for 1h, as compared that
of 1c, but it did not suffer spontaneous chemical hydrolysis.
Although the reactivity and enantioselectivity of 1d were
comparable to that of 1c, its synthesis was less efficient (max
yield 20% for 1d vs 50% for 1c).
Sulfinamide Substrate Range and Enantioselectivity. To
determine the substrate range of subtilisin E, we tested seven
additional arylsulfinamides with enzyme (Table 3). Subtilisin
E catalyzed the hydrolysis of 2c and para-substituted benzene-
The N-dihydrocinnamoyl derivative 1h and N-chloroacetyl
derivative 1c were the best acyl groups for p-toluenesulfinamide
since they showed the highest enantioselectivity and reactivity.
Thus, compounds 1c and 1h, which contain these groups, reacted
at least 100 times faster than the simplest N-acylsulfinamide,
sulfinamides, 3c and 4c, with high enantioselectivity (Etrue >
150), but the enantioselectivity decreased with the more
substituted sulfinamide, 5c (Etrue ) 98), and the more hindered
sulfinamides, 6c (Etrue ) 13) and 7c (Etrue ) 1.2). Compound
8c did not react with subtilisin E.
9
J. AM. CHEM. SOC. VOL. 127, NO. 7, 2005 2107

A R T I C L E S
Savile et al.
Scheme 4. Gram-Scale Kinetic Resolution of Sulfinamides 1, 3, and 5
Subtilisin E showed moderate to high conversion with 2c,
3c, 4c, and 5c (capp ) 14, 35, 21, and 41%, respectively, after
3 h), but the conversion decreased with the more hindered
arylsulfinamides, 6c and 7c (capp ) 19 and 30%, respectively,
after 24 h). Competing chemical hydrolysis (1-3%) occurred
with all N-chloroacetyl derivatives; however, reducing the
reaction temperature to 10 °C reduced spontaneous hydrolysis
to <1% after 24 h. B. subtilis var. biotecus A gave similar
conversions and enantioselectivity (see Supporting Information
Table S1).
The N-dihydrocinnamoyl sulfinamides showed similar enan-
tioselectivity to the N-chloroacetyl sulfinamides, but lower
conversion and no spontaneous chemical hydrolysis. N-Dihy-
drocinnamoyl derivatives 2h, 3h, and 4h showed high enanti-
oselectivity (Etrue > 150) and low to moderate conversion, capp
) 14% (8-fold lower than 2c), 9% (30-fold lower than 3h),
and 40% (4-fold lower than 4c), respectively, after 24 h.
Compound 5h reacted very slowly (0.5% capp after 24 h), and
6h and 8h did not react. This lack of reaction may be due to
either poor solubility31 or poor fit in the enzyme active site.
We resolved the more hindered substrates using the more
reactive N-chloroacetyl group, but used the N-dihydrocinnamoyl
group for all other sulfinamides to avoid spontaneous chemical
hydrolysis.
Comparing the HPLC traces of the product sulfinamides to
samples of known absolute configuration, prepared using either
Davis and co-workers’ method12 for toluenesulfinamide or
Senanayake and co-workers’ method16 for all other sulfinamides
(see Supporting Information), established the absolute configu-
rations. Subtilisin E favored the (R)-enantiomer in all cases.
Gram-Scale Resolutions. To demonstrate the synthetic
usefulness of this reaction, we resolved sulfinamides 1, 3, and
5 on a multigram scale with subtilisin E (Scheme 4). We chose
these sulfinamides for preparative reactions because sulfinamide
1 is widely used as a chiral auxiliary,5 the aryl group of 3 has
different electronic properties than 1 and might be a useful
alternative, and 2,4,6-trimethylbenzenesulfinamide 5 is more
hindered than 1 and gives fewer byproducts during nucleophilic
addition.32
Resolution of 1h (3.59 g) with subtilisin E gave (R)-1 (840
mg, 42% yield; the maximum yield is 50% in a resolution) with
95% ee and (S)-1h (1.53 g, 43% yield) with 95% ee at ca. 50%
conversion (3 days). If desired, both product and starting
material can be recrystallized to >99% ee. The N-dihydrocin-
namoyl group could be removed with hydrazine hydrate.33
Treating (S)-1h (144 mg) with hydrazine hydrate gave (S)-1
(70 mg, 90% yield) with 95% ee.
We resolved 3h (4.00 g) to give (R)-3 (683 mg, 30% yield)
with 97% ee and (S)-3h (2.42 g, 60% yield) with 64% ee at ca.
40% conversion (3 days). For unknown reasons, this resolution
stopped at 40% conversion, but recrystallization of unreacted
starting material gave (S)-3h (1.09 g, 27%) with 99% ee.
Treating (S)-3h (155 mg) with hydrazine hydrate gave (S)-1
(81 mg, 92%) with 99% ee.
We used the N-chloroacetyl group to resolve 2,4,6-tri-
methylbenzenesulfinamide 5 (1.90 g) because the N-dihydro-
cinnamoyl derivative reacted very slowly. The resolution yielded
(R)-5 (520 mg, 39% yield) with 90% ee at ca. 48% conversion
(6 h). Recrystallization of (R)-5 gave 395 mg (30%) with 99%
ee. The unreacted starting material, (S)-5c, was isolated with
45% yield (860 mg, 82% ee). The N-chloroacetyl group cleaved
in ethanolic KOH. Hydrolysis, followed by recrystallization,
gave (S)-5 (295 mg, 22% yield) with 99% ee. Recrystallization
was necessary because 5% of starting material spontaneously
hydrolyzed during reaction. However, chemical hydrolysis could
be reduced to <1% at 10 °C, while the enzymatic reaction was
only 3-fold slower (capp ) 16% at 10 °C vs 41% at 37 °C after
3 h).
These enzymatic resolutions are simple, convenient, and avoid
costly auxiliaries.12,16 The current synthetic route to enantiopure
1 relies on the starting material menthyl-p-toluenesulfinate (U.S.
$15-40/g), and enantiopure 3 and 5 rely on the amino alcohol,
cis-1-amino-2-indanol (U.S. $30/g) and give low yields of the
sterically hindered 5.34 This enzymatic resolution yields enan-
tiopure N-acylsulfinamides as intermediates. These may be
useful for diastereoselective enolate alkylation reactions as a
route to enantiopure R-substituted carboxylic acids.20
Molecular Basis for Higher Reactivity of 1c and 1h. To
understand why (R)-1h and (R)-1c reacted with subtilisin E
while (R)-1a did not, we modeled the first tetrahedral intermedi-
(31) The solubility of both 5h and 6h in 10% MeCN/buffer was <5 mM. Organic
cosolvents (1,4-dioxane, EtOH, DMSO) or additives (10 mM guanidium
chloride) did not significantly improve solubility. Performing the reaction
experiment at 50 °C or using a biphasic reaction system with tert-
butylmethyl ether also did not increase conversion. Using MeCN or 1,4-
dioxane at higher concentrations (20-90%) improved substrate solubility,
but did not increase conversion.
(33) Keith, D. D.; Tortora, J. A.; Yang, R. J. Org. Chem. 1978, 43, 3711-
3713.
(34) In our hands, the sodium bis(trimethylsilyl)amide displacement16 gave low
yields of (R)-2-4 and (R)-6 (18-52%) and no yield for (R)-5. When LiNH2/
NH3 was used in place of sodium bis(trimethylsilyl)amide, (R)-5 formed
in low yield (3%) and enantiomeric excess (68% ee).
(32) Han, Z.; Krishnamurthy, D.; Pflum, D.; Grover, P.; Wald, S. A.;
Senanayake, C. H. Org. Lett. 2002, 4, 4025-4028.
9
2108 J. AM. CHEM. SOC. VOL. 127, NO. 7, 2005

Subtilisin-Catalyzed Resolution of N-Acyl Arylsulfinamides
A R T I C L E S
Table 4. Minimized Structures for the Tetrahedral Intermediate of Subtilisin-E-Catalyzed Hydrolysis of 1h, 1c, 1a, and 7c
H bond Nꢀ
−Oγ distance (Å)
2
a
conformation
group in S1
′
(N−H−O
γ angle, deg)b
comments
(R)-1h
(Figure 2)
(R)-1h
tolyl
2.96 (155)
3.08 (152)
3.03 (154)
steric contact with Tyr 217, His 64,
and Met 222 in S1′ pocket
severe steric clash with
Gly219 and Asn155
unhindered in S1′ pocket
c
OS
(S)-1h
(Figure 2)
(S)-1h
(R)-1c
(not shown)
(R)-1a
(not shown)
(R)-7c
(Figure S2)
(S)-7c
(Figure S2)
OS
tolylc
tolyl
2.99 (157)
2.98 (153)
steric clash with catalytic His 64
possible hydrogen bond with
Thr 220 (ClR-Oγ1 distance ) 3.81 Å)
no contact with S1 residues
tolyl
none
none
2.97 (153)
2.94 (157)
2.93 (156)
binds above S1′ pocket
binds above S1′ pocket
a Narrow, hydrophobic pocket where leaving group alcohol, amide, or sulfinamide binds. b Unless otherwise noted, hydrogen bonds (His64 Nꢀ2-N or O,
Ser221 NH-O-, and Asn155 NHδ2-O-) were present in all structures (N-N or O distance 2.7-3.1 Å, N-H-O or N-H-N angle 120-175°). c Group
is forced out of S1′ because of steric clash with active site residues.
ate for hydrolysis of these substrates (see Experimental Section
for modeling details). The modeling identified interactions that
bind the N-acyl group of (R)-1h and (R)-1c, but not (R)-1a, to
the S1 pocket.
combined with the inductive effects of the R-chlorine atom,
which increases the electrophilicity of the carbonyl carbon, may
improve the binding and reactivity of (R)-1c.
Although the tetrahedral intermediate for (R)-1a also fits in
the subtilisin active site and makes all five key hydrogen bonds,
the small N-acyl group lacks contact with S1 pocket residues
(not shown). This lack of favorable interaction with S1 residues
to improve binding and an electron-withdrawing group to
increase reactivity may account for its lack of reaction.42
Molecular Basis for Enantioselectivity of Subtilisin E with
1h. Empirical rules based on substituent size can often predict
the fast-reacting enantiomer in subtilisin-catalyzed reactions of
secondary alcohols (Figure 1a).43 Sulfinamides mimic secondary
alcohols, where the aryl group is the large substituent and the
oxygen is the medium substituent, because of their similar shape.
However, the solvation of the medium group of a secondary
alcohol (e.g., methyl group) and the sulfoxide oxygen differs.
The incremental Gibbs free energy of transfer36,37 from n-octanol
to water for a methyl group is +0.56 kcal/mol, while that for
sulfoxide oxygen is -3.0 kcal/mol.44 We hypothesized that
subtilisin E favors the (R)-enantiomer because the sulfoxide
oxygen favors a water-exposed orientation, while the aryl group
binds in the S1′ leaving group pocket because of a favorable
hydrophobic interaction (Figure 1b).
The (R)-1h tetrahedral intermediate fit well in the active site,
and all five catalytically relevant hydrogen bonds were normal
length (<3.1 Å) (Table 4). The benzyl moiety of the dihydro-
cinnamoyl group bound in the S1 pocket, as expected based on
its similarity to phenylalanine.29,35 This benzyl moiety contacted
hydrophobic portions of five out of eight residues lining this
pocket (Leu126, Gly127, Gly128, Ala152, and Gly154). Using
the incremental Gibbs free energy of transfer from n-octanol to
water,36,37 we estimate the energy for this benzyl-S1 pocket
hydrophobic interaction to be 2.7-5.5 kcal/mol.38 Thus, the
hydrophobic interaction between the benzyl moiety and S1
pocket likely improves binding of (R)-1h, as compared to (R)-
1a.
The (R)-1c tetrahedral intermediate also fit well in the active
site, and all five hydrogen bonds were normal length (<3.1 Å)
(not shown). The R-chloro atom of (R)-1c was bound in the S1
pocket near the hydroxyl group (Oγ1-H) of Thr220 (Oγ1-Cl
distance ) 3.81 Å). This Oγ1-Cl distance was closer than the
van der Waals contact (3.91 Å),39 but longer than a hydrogen
bond between a hydroxyl group and chlorine (2.91-3.52 Å).40
A typical hydrogen bond between substrate and protein lowers
the energy by 0.5-1.5 kcal/mol.41 This Oγ1-Cl interaction
To test this hypothesis, we modeled the first tetrahedral
intermediate for hydrolysis of 1h. The protein modeling
molecular mechanics force field (AMBER) did not include
parameters for the sulfinamide group, so we estimated these
parameters using geometric information for methyl sulfinamide45
(35) Lee, T.; Jones, J. B. J. Am. Chem. Soc. 1996, 118, 502-508.
(36) (a) Fujita, T.; Iwasa, J.; Hansch, C. J. Am. Chem. Soc. 1964, 86, 5175-
5180. (b) Leo, A.; Hansch, C.; Elkins, D. Chem. ReV. 1971, 71, 525-616.
(c) Hansch, C.; Coats, E. J. Pharm. Sci. 1970, 59, 731-743. (d) Hansch,
C.; Leo, A. Substituent Constants for Correlation Analysis in Chemistry
and Biology; John Wiley and Sons: New York, 1979; pp 48-51.
(37) Fersht, A. Structure and Mechanism in Protein Science; W. H. Freeman
and Company: New York, 1998; pp 324-348.
(39) Estimated from the van der Waals radii of chlorine (1.75 Å) and hydrogen
(1.20 Å) and the O-H bond length (0.96 Å) from: Bondi, A. J. Phys.
Chem. 1964, 68, 441-451.
(40) Jeffrey, G. A.; Saenger, W. Hydrogen Bonding in Biological Structures;
Springer-Verlag: Berlin, 1991; pp 29-31.
(41) Removal of a hydrogen bond donor or acceptor in an enzyme active site
weakens binding by 0.5-1.5 kcal/mol.37 Although the hydrogen bond
inventory is zero (i.e., the number of hydrogen bonds is the same for both
the free enzyme and the enzyme-substrate complex), hydrogen bonds
between an enzyme and substrate increase the entropy due to the release
of bound water molecules.
(42) This reasoning predicts lower KM values for 1h and 1c than for 1a, but
low solubility of these substrates (<5 mM) prevented us from testing this
prediction.
(38) A hydrophobic interaction is the tendency of nonpolar compounds to transfer
from an aqueous phase to an organic phase. The energy of a hydrophobic
interaction is from the regaining of entropy by water after it is removed
from a hydrophobic group and can be estimated using the incremental Gibbs
free energy of transfer.37 The incremental Gibbs free energy of transfer of
a benzyl group from n-octanol to water is 2.7 kcal/mol using ∆Gtrans
)
2.303RTπ, where π is the hydrophobicity constant of the benzyl group (π
) 2.01)36d relative to hydrogen. However, binding of the benzyl group in
the hydrophobic S1 cavity removes two water-hydrophobic interfaces (i.e.,
substrate and protein), so the value might be as high as 5.5 kcal/mol.37
Since the S1 pocket lies on the surface of the protein and the benzyl group
is not completely buried, we estimate the energy for this hydrophobic
interaction to be 2.7-5.5 kcal/mol.36a,b,37
(43) (a) Fitzpatrick, P. A.; Klibanov, A. M. J. Am. Chem. Soc. 1991, 113, 3166-
3171. (b) Kazlauskas, R. J.; Weissfloch, A. N. E. J. Mol. Catal. 1997, 3,
65-72.
9
J. AM. CHEM. SOC. VOL. 127, NO. 7, 2005 2109

A R T I C L E S
Savile et al.
from ab initio calculations and from force field parameters for
a sulfonamide-based inhibitor46 also derived from ab initio
calculations (see Supporting Information for details). Using these
estimates, we expect only a qualitative rationalization of the
enantioselectivity of subtilisin E with N-acylsulfinamides.
Modeling 1h with subtilisin E gave one productive conforma-
tion for each enantiomer (Table 4). The two other plausible
conformations encountered severe steric clash with the protein.47
The productive complex of (R)-1h had its p-tolyl group bound
in the S1′ pocket and the sulfoxide oxygen exposed to solvent
water (Figure 2). All hydrogen bond angles were >120°, and
all five hydrogen bond lengths were <3.1 Å (Table 4). The
p-tolyl group appears to just fit in the S1′ pocket: Met222 in
the bottom of the S1′ pocket, Tyr217 at the back of the S1′
pocket, and catalytic His64 bumped the p-tolyl group.47 This
tight fit suggests a favorable hydrophobic interaction between
the tolyl group and the S1′ residues.
The productive complex of the slower-reacting (S)-1h had
its sulfoxide oxygen in the S1′ pocket and the tolyl group
exposed to solvent water (Figure 2). All hydrogen bond angles
were >120°, and all five hydrogen bond lengths were <3.1 Å
(Table 4). The sulfoxide oxygen fits well in the S1′ pocket, and
the protein does not hinder the p-tolyl group.47 Unlike the
favored enantiomer, the smaller oxygen does not make steric
contact with S1′ residues. Although the slow-reacting (S)-
enantiomer avoids steric hindrances, it also lacks favorable
hydrophobic interactions between the tolyl group and the S1′
residues. Using octanol-water partitioning data, we estimate
that the hydrophobic interactions favor binding of (R)-1h by
∼6.4 kcal/mol over the (S)-enantiomer in water.44 The (R)-
enantiomer places the hydrophobic aryl group in the S1′ pocket
and the hydrophilic sulfoxide oxygen in the solvent (water),
while the (S)-enantiomer does the opposite.
Consistent with this explanation, the enantioselectivity of the
subtilisin-E-catalyzed hydrolysis of 1h decreased from E > 150
in 9:1 water/acetonitrile to E ) 12 in 1:9 water/acetonitrile.
Similarly, the enantioselectivity with 1c decreased from E >
150 in 9:1 water/acetonitrile to E ) 41 in 1:9 water/acetonitrile.
The high concentration of acetonitrile favors the p-tolyl group
in the solvent; thus, the enantiomer preference shifts toward
the (S)-enantiomer, which orients with the oxygen in the S1′
pocket and the tolyl group in solvent. This result suggests that
the enantioselectivity for acylation of sulfinamides in organic
solvent would also be low, but we did not detect any acylation
of p-toluenesulfinamide, consistent with Ellman’s earlier report
of no reaction.15
Figure 2. Catalytically productive tetrahedral intermediates for the sub-
tilisin-E-catalyzed hydrolysis of fast-reacting (R)-1h (I) and slow-reacting
(S)-1h (II), as identified by molecular modeling. The important active site
and substrate atoms (sticks) are colored as follows: gray (carbon), red
(oxygen), blue (nitrogen), and orange (sulfur). Surrounding atoms (space
fill) of subtilisin are shown in blue. For clarity, all hydrogen atoms and
water molecules are hidden. Both I and II maintain all catalytically essential
hydrogen bonds, and the benzyl moiety of the dihydrocinnamoyl group binds
in the S1 pocket, as expected based on its similarity to phenylalanine.29 In
the fast-reacting enantiomer, (R)-1h (I), the p-tolyl group binds in the
hydrophobic S1′ pocket and sulfoxide oxygen is exposed to solvent water.
In the slow-reacting enantiomer, (S)-1h (II), the sulfoxide oxygen binds in
the hydrophobic S1′ pocket and the p-tolyl group is exposed to solvent water.
The nonproductive conformations (not shown) encountered severe steric
clash with the active site residues.
The decreasing enantioselectivity with the larger sulfinamides,
5c (Etrue ) 98) and 6c (Etrue ) 13), and the loss of enantiose-
lectivity with the very large sulfinamide, 7c (Etrue ) 1.2), are
also consistent with this explanation for enantioselectivity.
Larger aryl substituents increase the steric hindrance in the S1′
pocket, which overwhelms the energy gained through a hydro-
phobic interaction between the pocket and the aryl group and,
therefore, reduces the enantioselectivity. In other words, when
either the aryl group or sulfoxide oxygen can fit in the S1′ pocket
(1-4), a hydrophobic interaction favors the aryl group and the
enantioselectivity is high. However, as the aryl group becomes
larger, increased steric hindrance with S1′ residues oppose the
hydrophobic interaction and lower the enantioselectivity (5-
7). Modeling with 7c suggests that this sulfinamide leaving
group is too large for the S1′ pocket and binds above it (see
Supporting Information Figure S2).
(44) The incremental Gibbs free energy of transfer from n-octanol to water for
the tolyl group was estimated to be ca. +3.4 kcal/mol, and the sulfoxide
oxygen was ca. -3.0 kcal/mol using ∆Gtrans ) 2.303RTπ, where π is the
hydrophobicity constant of the group relative to hydrogen.36a,b,37 Switching
from (S)-1h to (R)-1h places the tolyl group into a hydrophobic pocket
(3.4 kcal/mol for the tolyl group) and removes the sulfoxide group from
the hydrophobic pocket (3.0 kcal/mol) for a total of 6.4 kcal/mol in water.
The hydrophobicity constant of the tolyl group (π ) 2.52) was estimated
by adding π values of a phenyl and a methyl group.36d The hydrophobicity
constant of the sulfoxide oxygen (π ) -2.19) was estimated by subtracting
π values of a methyl sulfide group from a methyl sulfoxide group.36d The
energy difference may be higher if one includes the hydrophobic surface
of the S1′ pocket.
(45) Bharatam, P. V.; Kaur, A.; Kaur, D. J. Phys. Org. Chem. 2002, 15, 197-
203.
(46) Rossi, K. A.; Merz, K. M., Jr.; Smith, G. M.; Baldwin, J. J. J. Med. Chem.
1995, 38, 2061-2069.
(47) See Supporting Information for complete molecular modeling details.
9
2110 J. AM. CHEM. SOC. VOL. 127, NO. 7, 2005

Subtilisin-Catalyzed Resolution of N-Acyl Arylsulfinamides
A R T I C L E S
Discussion
enantioselectivity with the structurally related sulfinamides. For
secondary alcohols and isosteric primary amines, the substituents
are usually both hydrophobic, so the hydrophobic binding
contribution differences are smaller than those for arylsulfin-
amides, where the substituents are the polar sulfoxide oxygen
and the hydrophobic aryl group. This polarity difference between
the two substituents appears to dominate subtilisin E’s enan-
tiorecognition of arylsulfinamides, resulting in high enantiose-
lectivity.
In this paper, we identified subtilisin E as the most suitable
hydrolase of those examined for preparation of enantiopure
arylsulfinamides. The reactivity and enantioselectivity of sub-
tilisin E toward N-acyl arylsulfinamides depend on the N-acyl
group. Simple N-acyl compounds, such as N-acetyl and N-
butanoyl, did not react with subtilisin E. Molecular modeling
suggests that the reactive acyl groups may mimic a phenylala-
nine moiety. Other research groups have also modified unre-
active substrates to convert them into good substrates. For
example, the fungus BeauVeria bassiane did not hydroxylate
cyclopentanone, but did hydroxylate the N-benzoylspirooxazo-
lidine derivative with high yield and diastereoselectivity.48 In a
second example, changing from the 2-pyridylacetyl to the
4-pyridylacetyl increased the rate 8-fold and the enantioselec-
tivity 3-fold for a penicillin G acylase-catalyzed hydrolysis of
1-phenethyl esters.49 In a third example, the enantioselectivity
of Pseudomonas cepacia lipase-catalyzed acylation of 2-[(N,N-
dimethylcarbamoyl)methyl)]-3-cyclopenten-1-ol varied from E
) 4 to 156, depending on the acylating agent.50
There are three synthetic routes to enantiopure sulfinamides:
the method of Davis and co-workers12 for p-toluenesulfinamide,
the method of Ellman and co-workers6 for tert-butylsulfinamide,
and the method of Senanayake and co-workers16 for a variety
of alkyl- and arylsulfinamides. Using subtilisin E, we resolved
gram quantities of 1h, 3h, and 5c. These resolutions were simple,
convenient, and inexpensive. Our strategy does not provide us
with as wide a variety of enantiopure sulfinamides as reported
by Senanayake, but the selectivity and mildness of the biocata-
lytic route make it the preferred route when there is a choice.
The biocatalytic route is amenable to scale-up, is environmen-
tally acceptable, and is performed under mild conditions. As
well, the catalyst, subtilisin E, is inexpensive to produce and
could be recycled.
Molecular modeling of the first tetrahedral intermediate for
subtilisin-E-catalyzed hydrolysis of 1h suggests that the (R)-
enantiomer reacts faster because of preferential binding of the
nonpolar p-tolyl group in the hydrophobic S1′ pocket versus
the polar sulfoxide oxygen, which prefers to be exposed to
solvent water. Changing the solvent to 1:9 water/acetonitrile
decreased the enantioselectivity. A similar decrease in enanti-
oselectivity occurred for subtilisin-catalyzed reactions of amino
acid derivatives.51 In water, subtilisin Carlsberg showed high
enantioselectivity for the hydrolysis of a natural L-amino acid
ester, but in organic solvent, transesterification was 1-2 orders
of magnitude less enantioselective. Klibanov and co-workers
suggested that the L-amino acid, but not the D-amino acid, binds
to the hydrophobic S1 pocket. Water as the solvent favors this
interaction; thus, the enantioselectivity is higher in water. As
expected, the difference in enzyme enantioselectivity was greater
for amino acid derivatives with more hydrophobic side chains.
Subtilisins usually show only low to moderate enantioselec-
tivity with secondary alcohols and isosteric amines, but high
Experimental Section
1
General. H and 13C NMR spectra were obtained as dilute CDCl3
solutions at 300 and 75 MHz, respectively. Chemical shifts are
expressed in parts per million (δ) and are referenced to tetramethylsilane
or trace CHCl3 in CDCl3. Coupling constants are reported in hertz (Hz).
HPLC analyses were performed on a 4.6 × 250 mm Daicel Chiralcel
OD or Chiralpak AD column (Chiral Technologies, Exton, U.S.A.) and
monitored at 238 or 222 nm. Flash chromatography with silica gel
(230-400 mesh) or preparative TLC (20 × 20 cm, 1000 µm) was used
to purify all intermediates and substrates. Visualization of UV-inactive
materials on TLC was accomplished using phosphomolybdic acid or
ninhydrin followed by heating. All reagents, buffers, starting materials,
and anhydrous solvents were purchased from Sigma-Aldrich Canada
(Oakville, Canada) and used without purification. All air- and moisture-
sensitive reactions were performed under N2. The pBE3 Escherichia
coli-Bacillus subtilis shuttle vector,26b containing the subtilisin E gene,
was kindly provided by Dr. F. Arnold (Caltech), and the Bacillus subtilis
strain DB10452 was a gift from Dr. S. L. Wong (University of Calgary).
Synthesis of Substrates. Sulfinamides (1-8). We prepared racemic
sulfinamides 1-8 using literature procedures (Scheme 3).20-22 The
relevant analytical data are in the Supporting Information.
N-Acylsulfinamides. We prepared N-acylsulfinamides by treating
sulfinamides 1-8 with 2 equiv of n-BuLi in THF, followed by rapid
addition of the symmetrical anhydride of the appropriate carboxylic
acid.53 The relevant analytical data are given below or in the Supporting
Information.
N-Chloroacetyl-p-toluenesulfinamide (1c) was obtained as a white
solid (186 mg, 50%): mp 119-121 °C; 1H NMR δ 2.43 (s, 3H, PhCH3),
4.29 (s, 2H, C(O)CH2Cl), 7.31 (d, J ) 8.1, 2H), 7.63 (d, J ) 8.1, 2H);
13C NMR δ 21.8 (PhCH3), 42.5 (C(O)CH2Cl), 124.8, 130.6, 139.9,
143.5 (phenyl), 167.2 (CdO); HRMS calcd for C9H1035ClNO2S (M+)
231.0120, found 231.0123. The enantiomers were separated by HPLC
(Daicel Chiralcel OD column, 90:10 hexanes/EtOH, 0.5 mL/min, 238
nm; (R)-1c, tR ) 21.3 min; (S)-1c, tR ) 51.4 min).
N-Dihydrocinnamoyl-p-toluenesulfinamide (1h) was obtained as
a white solid (550 mg, 59%): mp 85-87 °C (lit.20 mp 94-96 °C); 1H
NMR δ 2.43 (s, 3H, PhCH3), 2.70 (m, 2H, C(O)CH2), 3.01 (t, J ) 7.8,
2H, CH2Ph), 7.17-7.29 (m, 7H, phenyl), 7.41 (d, J ) 8.1, 2H, phenyl),
7.81 (br s, 1H, NH); 13C NMR δ 21.9 (tolyl CH3), 31.2 (CH2Ph), 37.9
(C(O)CH2), 124.9, 126.6, 128.7, 128.8, 130.1, 140.1, 140.3, 142.6
(phenyl), 173.6 (CdO). The enantiomers were separated by HPLC
(Daicel Chiralcel OD column, 90:10 hexanes/EtOH, 0.5 mL/min, 238
nm; (R)-1h, tR ) 20.4 min; (S)-1h, tR ) 22.4 min).
Hydrolase Library. All screening was performed at pH 7.2. The
hydrolases were dissolved in BES buffer (5.0 mM, pH 7.2) at the
concentration listed in Table 1, centrifuged for 10 min at 2000 rpm,
and titrated to pH 7.2. The supernatant was used for screening.
Screening of Commercial Hydrolases with pH Indicators. The
assay solution was prepared by mixing 1c (1 mL of a 440 mM solution
in CH3CN) and p-nitrophenol (6.71 mL of a 1.0 mM solution in 5.0
mM BES, pH 7.2) with BES buffer (5.14 mL of a 5.0 mM solution,
pH 7.2). Hydrolase solutions (10 µL/well) were transferred to a 96-
(48) (a) Braunegg, G.; de Raadt, A.; Feichtenhofer, S.; Griengl, H.; Kopper, I.;
Lehman, A.; Weber, H.-J. Angew. Chem., Int. Ed. 1999, 38, 2763-2766.
(b) de Raadt, A.; Griengl, H.; Weber, H. Chem.sEur. J. 2001, 7, 27-31.
(49) Pohl, T.; Waldmann, H. Tetrahedron Lett. 1995, 36, 2963-2966.
(50) Ema, T.; Maeno, S.; Takaya, Y.; Sakai, T.; Utaka, M. J. Org. Chem. 1996,
61, 8610-8616.
(51) (a) Zaks, A.; Klibanov, A. M. J. Am. Chem. Soc. 1986, 108, 2767-2768.
(b) Margolin, A. L.; Tai, D.-F.; Klibanov, A. M. J. Am. Chem. Soc. 1987,
109, 7885-7887. (c) Sakurai, T.; Margolin, A. L.; Russell, A. J.; Klibanov,
A. M. J. Am. Chem. Soc. 1988, 110, 7236-7237.
(52) Kawamura, F.; Doi, R. H. J. Bacteriol. 1984, 160, 442-444.
(53) Chen, F. M. F.; Kuroda, K.; Benoiton, N. L. Synthesis 1978, 12, 928-
930.
9
J. AM. CHEM. SOC. VOL. 127, NO. 7, 2005 2111

A R T I C L E S
Savile et al.
well microtiter plate followed by assay solution (90 µL/well). The final
concentration in each well was 3.1 mM substrate, 4.65 mM BES, 0.46
mM p-nitrophenol in a total volume of 100 µL 7% acetonitrile in buffer.
The plate was shaken for 5 s on the microplate reader, and the
absorbance decrease was monitored at 404 nm for 1 h. The assay was
performed in quadruplicate at 25 and 37 °C.23 The low pKa of the
N-acylsulfinamide NH group (ca. pKa 6) reduced the sensitivity of the
assay by protonating the pH indicator, p-nitrophenoxide, but a decrease
in the absorbance could still be observed at pH 7.2.
RT until the reaction was complete by TLC (3 h). The reaction mixture
was diluted with CH2Cl2 (15 mL) and washed with 1 N HCl (2 × 10
mL), saturated NaHCO3 (2 × 10 mL), and saturated NaCl (10 mL).
The organic layer was dried over Na2SO4 and concentrated in vacuo
to give (S)-1 (70 mg, 90%, 95% ee).
3h. Substrate (4.00 g, 13.0 mmol) in MeCN (100 mL) was added to
enzyme solution (900 mL, ca. 15 U/mL with suc-AAPF-pNA) and
incubated at 37 °C until ca. 40% conversion (3 days), as determined
by HPLC. The mixture was filtered through Celite, extracted with
CH2Cl2 (3 × 200 mL), dried over Na2SO4, and concentrated in vacuo.
Separation of substrate and product on silica (1:1 EtOAc/hexanes to
EtOAc) gave (R)-3 (683 mg, 30%, 97% ee) and (S)-3h (2.42 g, 60%,
64% ee). Recrystallization from EtOAc/hexanes gave (S)-3h (1.09 g,
27%, 99% ee), as determined by HPLC. Remaining starting material
(S)-3h (155 mg, 0.5 mmol) was treated with hydrazine hydrate (1 mL)
at 0 °C followed by stirring at RT until the reaction was complete by
TLC (3 h). The reaction mixture was diluted with CH2Cl2 (15 mL)
and washed with 1 N HCl (2 × 10 mL), saturated NaHCO3 (2 × 10
mL), and saturated NaCl (10 mL). The organic layer was dried over
Na2SO4 and concentrated in vacuo to give (S)-3 (81 mg, 92%, 99%
ee).
Small-Scale Reactions with Commercial Hydrolases to Determine
Enantioselectivity. Reactions of N-chloroacetylsulfinamides with com-
mercial enzymes were carried out at 25 °C. For example, proteinase
from B. subtilis var. biotecus A (16 mg) and substrate (5 mM in
CH3CN) in a total volume of 10 mL of buffer (20 mM BES, pH 7.2)
were stirred together in reaction vials for 3-6 h. Upon completion,
the aqueous phase was extracted with CH2Cl2 (2 × 10 mL). The
combined organic layers were dried over anhydrous Na2SO4 and
concentrated in vacuo. The residue was dissolved in 1 mL of EtOH,
filtered through a nylon filter (0.45 µm), and analyzed by HPLC (Daicel
OD column, 90:10 hexanes/EtOH, 1.0 mL/min, 238 nm).
Cultivation and Isolation of Subtilisin E. Protease-deficient B.
subtilis DB104 was transformed with vector pBE3, as previously
described.26b,54 Protease-deficient B. subtilis DB104 secretes proteases
into the culture medium and accumulates them to a high level.
Endogenous protease activity is less than 3%.52 Cells were grown in
Schaeffer’s sporulation medium (2XSG) supplemented with kanamycin
(50 µg/mL). Enzyme activity was monitored by adding 90 µL of assay
solution [0.2 mM succinyl-AAPF-p-nitroanilide (suc-AAPF-pNA)/100
mM Tris pH 8.0/10 mM CaCl2] to 10 µL of supernatant and following
the reaction at 410 nm at 37 °C for 15 min. When maximal activity
was reached (48 h), the cells were centrifuged, and the supernatant
was filter sterilized and lyophilized. The residue was dissolved in 10
or 50 mM Tris buffer (pH 8.0, 10 mM CaCl2), and the initial activity
was assessed as above. Typical activity of the supernatant or prepared
solution was 10-20 U/mL, with suc-AAPF-pNA.
5c. Substrate (1.90 g, 7.30 mmol) in CH3CN (30 mL) was added to
enzyme solution (270 mL, ca. 20 U/mL with suc-AAPF-pNA) and
incubated at 37 °C until ca. 49% conversion (6 h), as determined by
HPLC. The mixture was filtered through Celite, extracted with
CH2Cl2 (3 × 200 mL), dried over Na2SO4, and concentrated in vacuo
to give the crude mixture of substrate and product. Separation of
substrate and product on silica (2:1 EtOAc/hexanes to EtOAc) gave
(R)-5 (520 mg, 39%, 90% ee) and (S)-5c (860 mg, 45%, 82% ee).
Recrystallization from EtOAc/hexanes gave (R)-5 (395 mg, 30%, 99%
ee), as determined by HPLC. Remaining starting material was
hydrolyzed in 1 N KOH (1:1 H2O/EtOH) for 1 h at RT. The reaction
mixture was extracted with CH2Cl2 (3 × 15 mL), dried over Na2SO4,
and concentrated in vacuo to give (S)-5 (295 mg, 22%, 99% ee) after
recrystallization.
Modeling of Tetrahedral Intermediates Bound to Subtilisin E.
All modeling was performed using Insight II 2000.1/Discover (Accelrys,
San Diego, CA) on an SGI Octane UNIX workstation using the
AMBER force field.55 We used a nonbonded cutoff distance of 8 Å
and a distance-dependent dielectric of 1.0, and we scaled the 1-4 van
der Waals interactions by 50%. Protein structures in Figures 2 and S1
(Supporting Information) were created using PyMOL (Delano Scientific,
San Carlos, CA). The X-ray crystal structure of subtilisin E (entry
1SCJ)27 from the Protein Data Bank is a Ser221Cys subtilisin
E-propeptide complex. Using the Builder module of Insight II, we
replaced the Cys221 with a serine and removed the propeptide region.
The hydrogen atoms were added to correspond to pH 7.0. Histidines
were uncharged, aspartates and glutamates negatively charged, and
arginines and lysines positively charged. The catalytic histidine (His64)
was protonated. The positions of the water hydrogens and then the
enzyme hydrogens were optimized using a consecutive series of short
(1 ps) molecular dynamic runs and energy minimizations.56 This
optimization was repeated until there was <2 kcal/mol in the energy
of the minimized structures. Thereafter, an iterative series of geometry
optimizations were performed on the water hydrogens, enzyme
hydrogens, and full water molecules. Finally, the whole system was
geometry optimized.
Hydrolysis of N-Acylsulfinamides with Subtilisin E. Reactions
of N-acylsulfinamides with subtilisin E were carried out at 37 °C. Each
mixture consisted of subtilisin E (solution in 10 mM Tris buffer (pH
8.0, 1 mM CaCl2), 178 µL), CaCl2 (1 M, 2 µL), and substrate (50 mM
in CH3CN, 20 µL). The reaction was allowed to continue for 3 h with
N-bromoacetyl and N-chloroacetyl sulfinamides and for 24 h with all
other substrates. The reaction was diluted with dH2O (400 µL) and
terminated with the addition of CH2Cl2 (500 µL). The organic layer
was removed, and the aqueous layer was twice extracted with CH2Cl2
(500 µL). The phases were separated by centrifugation, and the organic
layers were collected. The combined organics were evaporated under
a stream of N2, and the residue was diluted with EtOH (150 µL) for
analysis by HPLC (Brownlee Si-10 silica column followed by Daicel
OD column, 90:10 hexanes/EtOH, 0.6 mL/min, 238 or 222 nm).
Although the chiral column separated enantiomers, it did not separate
the substrate and product. This two-column system separated the
substrate and product in the silica column and the enantiomers of each
in the chiral column.
Gram-Scale Resolution of N-Acylsulfinamides. 1h. Substrate (3.59
g, 12.5 mmol) in CH3CN (50 mL) was added to enzyme solution (450
mL, ca. 20 U/mL with suc-AAPF-pNA) and incubated at 37 °C until
ca. 50% conversion (3 days), as determined by HPLC. The mixture
was filtered through Celite, extracted with CH2Cl2 (3 × 200 mL), dried
over Na2SO4, and concentrated in vacuo. Separation of substrate and
product on silica (1:1 EtOAc/hexanes to EtOAc) gave (R)-1 (840 mg,
42%, 95% ee) and (S)-1h (1.53 g, 43%, 95% ee), as determined by
HPLC. Remaining starting material (S)-1h (144 mg, 0.5 mmol) was
treated with hydrazine hydrate (1 mL) at 0 °C followed by stirring at
The tetrahedral intermediates were built manually and covalently
linked to Ser221. Since the parameters for the sulfinamide group were
not included in the AMBER force field, they were assigned in analogy
to existing parameters.45-47 The geometric properties of the sulfinamide
(55) (a) Weiner, S. J.; Kollman, P. A.; Case, D. A.; Singh, U. C.; Ghio, C.;
Alagona, G.; Profeta, S.; Weiner, P. J. Am. Chem. Soc. 1984, 106, 765-
784. (b) Weiner, S. J.; Kollman, P. A.; Nguyen, D. T.; Case, D. A. J.
Comput. Chem. 1986, 7, 230-252.
(54) Harwood, C. R.; Cutting, S. M. Molecular Biological Methods for Bacillus;
John Wiley & Sons: Chichester, U.K., 1990; pp 33-35 and 391-402.
(56) Raza, S.; Fransson, L.; Hult, K. Protein Sci. 2001, 10, 329-338.
9
2112 J. AM. CHEM. SOC. VOL. 127, NO. 7, 2005

Subtilisin-Catalyzed Resolution of N-Acyl Arylsulfinamides
A R T I C L E S
moiety of minimized 1h were compared to an available X-ray crystal
structure57 and adjusted as necessary. Nonstandard partial charges were
calculated using a formal charge of -1 for the substrate oxyanion.
Energy minimization proceeded in three stages: first, minimization of
substrate with only the protein constrained (25 kcal mol-1 Å-2); second,
minimization with only the protein backbone constrained (25 kcal mol-1
Å-2); and for the final stage, the minimization was continued without
constraints until the root mean square value was less than 0.0005 kcal
mol-1 Å-1. A catalytically productive complex required all five
hydrogen bonds within the catalytic machinery. We set generous limits
for a hydrogen bond: a donor to acceptor atom distance of less than
3.1 Å with a nearly linear arrangement (>120° angle) of donor atom,
hydrogen, and acceptor atom. Structures lacking any of the five
catalytically relevant hydrogen bonds or encountering severe steric clash
with enzyme were deemed nonproductive.
Natural Sciences and Engineering Research Council (Canada)
for a postgraduate fellowship. We thank Dr. Frances Arnold
(California Institute of Technology, USA) for the subtilisin E
plasmid, Dr. S. L. Wong (University of Calgary, Canada) for
the B. subtilis DB104 cells, Dr. Fred Schendel and Rick
Dillingham (University of Minnesota Biotechnology Institute,
USA) for the large-scale fermentation and purification of
subtilisin E, Dr. Ronghua Shu (McGill University) for the initial
substrate synthesis and screening experiments, Linda Fransson
(Royal Institute of Technology (KTH), Sweden), and the
Minnesota Supercomputing Institute (University of Minnesota,
USA) for assistance with the modeling software.
Supporting Information Available: Synthesis and charac-
terization data for compounds 1-8, enantioselectivity of Bacillus
subtilis var. biotecus A with 2c-5c, complete molecular
modeling details, and the modeling figures for 7c. This material
is available free of charge via the Internet at http://pubs.acs.org.
Acknowledgment. We thank the donors of the Petroleum
Research Fund administered by the American Chemical Society,
and McGill University for financial support. C.K.S. thanks the
(57) Robinson, P. D. Acta Crystallogr. 1991, C47, 594-596.
JA045397B
9
J. AM. CHEM. SOC. VOL. 127, NO. 7, 2005 2113