Extreme synergistic mutational effects in the directed evolution of a baeyer-villiger monooxygenase as catalyst for asymmetric sulfoxidation

Article abstract of DOI:10.1021/ja5098034

Structure-based directed evolution utilizing iterative saturation mutagenesis (ISM) has been applied to phenyl acetone monooxygenase (PAMO), a thermally robust Baeyer-Villiger monooxygenase, in the quest to access a mutant which displays reversed enantioselectivity in the asymmetric sulfoxidation of prochiral thioethers. Whereas WT PAMO leads to 90% ee in the sulfoxidation of p-methylbenzyl methyl thioether with preference for the (S)-sulfoxide, the evolved mutant I67Q/P440F/A442N/L443I is 95% (R)-selective in the reaction of this and other thioethers. Partial deconvolution of the (R)-selective mutant with generation of the respective four single mutants shows that all of them are (S)-selective, which points to pronounced synergism (cooperative nonadditivity) when they interact in concert. Complete deconvolution with formation of all combinatorial forms of the respective double and triple mutants allows the designed construction of a fitness landscape featuring all 24 upward pathways leading from WT to the (R)-selective quadruple mutant. In all 24 trajectories strong cooperative mutational effects were found as well, which indicates that such mutational changes in enzymes constitute nonlinear systems. A theoretical analysis based on induced fit docking explains many of the observed effects on a molecular level.

Full text of DOI:10.1021/ja5098034

Subscriber access provided by - Access paid by the | UCSF Library
Article
Extreme Synergistic Mutational Effects in the Directed Evolution of a
Baeyer-Villiger Monooxygenase as Catalyst for Asymmetric Sulfoxidation
Zhi-Gang Zhang, Richard Lonsdale, Joaquin Sanchis, and Manfred T. Reetz
J. Am. Chem. Soc., Just Accepted Manuscript • DOI: 10.1021/ja5098034 • Publication Date (Web): 14 Nov 2014
Downloaded from http://pubs.acs.org on November 18, 2014
Just Accepted
“Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted
online prior to technical editing, formatting for publication and author proofing. The American Chemical
Society provides “Just Accepted” as a free service to the research community to expedite the
dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts
appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been
fully peer reviewed, but should not be considered the official version of record. They are accessible to all
readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered
to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published
in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just
Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor
changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers
and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors
or consequences arising from the use of information contained in these “Just Accepted” manuscripts.
Journal of the American Chemical Society is published by the American Chemical
Society. 1155 Sixteenth Street N.W., Washington, DC 20036
Published by American Chemical Society. Copyright © American Chemical Society.
However, no copyright claim is made to original U.S. Government works, or works
produced by employees of any Commonwealth realm Crown government in the course
of their duties.

Page 1 of 28
Journal of the American Chemical Society
1
2
3
4
5
6
7
8
9
Extreme Synergistic Mutational Effects in the Directed Evolu-
tion of a Baeyer-Villiger Monooxygenase as Catalyst for
Asymmetric Sulfoxidation
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Zhi-Gang Zhanga,b, Richard Lonsdalea,b, Joaquin Sanchisc and Manfred T. Reetza,b*
a Department of Chemistry, Philipps-Universität Marburg, Hans-Meerwein Str., 35032 Marburg, Germany
b Max-Planck-Institut für Kohlenforschung, Kaiser-Wilhelm-Platz 1, 45470 Mülheim an der Ruhr, Germany
c Monash Institute for Pharmaceutical Sciences, Monash University, 381 Royal Parade, Parkville, Australia
KEYWORDS Baeyer-Villiger monooxygenases; asymmetric sulfoxidation; directed evolution; iterative saturation mu-
tagenesis; cooperative catalytic effects
ABSTRACT: Structure-based directed evolution utilizing iterative saturation mutagenesis (ISM) has been applied to phe-
nyl acetone monooxygenase (PAMO), a thermally robust Baeyer-Villiger monooxygenase, in the quest to access a mutant
which displays reversed enantioselectivity in the asymmetric sulfoxidation of prochiral thioethers. Whereas WT PAMO
leads to 90% ee in the sulfoxidation of p-methylbenzyl methyl thioether with preference for the (S)-sulfoxide, the evolved
mutant I67Q/P440F/A442N/L443I is 95% (R)-selective in the reaction of this and in other thioethers. Partial deconvolu-
tion of the (R)-selective mutant with generation of the respective four single mutants shows that all of them are (S)-
selective, which points to pronounced synergism (cooperative non-additivity) when they interact in concert. Complete
deconvolution with formation of all combinatorial forms of the respective double and triple mutants allows the designed
construction of a fitness landscape featuring all 24 upward pathways leading from WT to the (R)-selective quadruple
mutant. In all 24 trajectories strong cooperative mutational effects were found as well, which indicates that such muta-
tional changes in enzymes constitute non-linear systems. A theoretical analysis based on induced fit docking explains
many of the observed effects on a molecular level.
INTRODUCTION
oethers.8 Utilizing the crystal structure of PAMO,9 Mi-
hovilovic, Fraaije and coworkers applied rational design in
the successful attempt to enhance (R)-selectivity signifi-
cantly in these cases, variant M446G leading to 92-95% ee
for several thioethers.10 However, reversal of stereoselec-
tivity has not been achieved to date.
Here we report a directed evolution11 study, which pro-
vides both (R)- and (S)-selective PAMO variants for sev-
eral structurally different thioethers. We also made the
surprising discovery that upon deconvoluting the best
(R)-selective mutant characterized by four point muta-
tions, all of the respective single mutants are (S)-selective.
This is the most extreme case of cooperative non-
additivity recorded thus far in protein engineering,12 an
unusual phenomenon that inspired us to construct a
complete fitness landscape featuring all 4! = 24 pathways
leading from WT PAMO to the best (R)-selective mutant,
which likewise revealed drastic non-additivity in all of the
upward evolutionary trajectories. In an attempt to unveil
the source of enantioselectivity, induced fit docking cal-
culations were applied.
The catalytic asymmetric sulfoxidation of prochiral thi-
oethers constitutes an important transformation in organ-
ic chemistry, mediated by chiral synthetic transition met-
al complexes,1 organocatalysts2 or enzymes of the type
Baeyer-Villiger monooxygenases (BVMOs).3,4 These fla-
vin-dependent enzymes react with molecular oxygen to
form an intermediate flavin-hydroperoxide (Fl-OOH)
which reacts electrophilically with either one of the two
enantiotopic lone electron pairs (or both) of the sulfur
functionality.4 In contrast, the Baeyer-Villiger reaction is
initiated by nucleophilic addition of deprotonated flavin-
hydroperoxide (Fl-OO-) to the carbonyl function of ke-
tones with formation of
a
short-lived Criegee-
intermediate.3,5 Stereodifferentiation between the two
lone electron pairs of prochiral thioethers can be achieved
by using wildtype (WT) cyclohexanone monooxygenase
(CHMO)3,4 or mutants for “difficult” substrates,6 but this
BVMO is only moderately thermostable, requiring immo-
bilization and other biotechnological techniques for prac-
tical applications.3j In contrast, phenyl acetone monooxy-
genase (PAMO) is an exceptionally robust BVMO.7 Unfor-
tunately, wildtype (WT) PAMO generally shows low de-
grees of enantioselectivity in the oxidation of most thi-
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 2 of 28
1
2
Scheme 1. Model Reaction Used in all Protein Engineering Experiments.
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Table 1. Best (S)- and (R)-selective PAMO Mutants (lysates) as Catalysts in the Asymmetric Sulfoxidation of Thi-
oether 1b, All Originating from Previous PAMO Studies Regarding Stereoselective BV Reactions.13 (Experimental
uncertainty in the %ee-values here and in all other stereoselectivity data is better than +/- 2%)
Entry
Variant
Q93N
% Conversion
% ee
96
92
95
99
95
98
94
97
93
10
Config.
% Sulfone
1
2
99
89
99
99
99
99
99
86
99
85
75
S
S
S
S
S
S
S
S
S
R
R
S
6
6
Q93N/P94D
L443F
3
12
16
11
4
5
P440A
P440V
P440D
P440E
6
7
3
5
8
9
10
11
12
P440W
P440T
6
13
11
P440Y
P440L
5
22
23
P440F
82
56
RESULTS AND DISCUSSION
In order to achieve high levels of reversed enantioselectiv-
ity, we turned to directed evolution11 using iterative satu-
ration mutagenesis (ISM).11f,14 Guided by the crystal struc-
ture of PAMO,9 and by previously identified “hot spots”
(CAST sites),13 we considered the loop residues P440,
A442 and L443 as potential randomization sites for satu-
ration mutagenesis, in addition to V54, I67, Q152 identi-
fied earlier as “hot spots” by Mihovilovic, Fraaije and
coworkers10 (Figure. 1).
Experimental platform and initial results using a
model substrate. The asymmetric sulfoxidation of thio-
ether 1b was chosen as the model reaction (Scheme 1),
with WT PAMO leading to (S)-2b with high enantioselec-
tivity (90% ee), similar to the sulfoxidation of substrate 1a
(99% ee in favor of (S)-2a).8 Prior to applying directed
evolution for improving (S)-selectivity and reversing en-
antioselectivity in favor of (R)-2b, we first tested 40 dif-
ferent PAMO mutants that have been previously evolved
for stereoselective BV reactions13 (Supplementary Infor-
mation). Most of these variants proved to be (S)-selective.
Some of the most (S)-selective variants, together with the
two slightly (R)-selective variants are shown in Table 1,
where lysates were used in all cases. Depending upon the
nature of the variant, and the conditions used, some de-
gree of overoxidation with formation of the respective
sulfone 3 was observed in most cases. This means that
kinetic resolution of the sulfoxides may be operating,
which could influence the apparent ee-value, a commonly
observed feature in BVMO-catalyzed sulfoxidation reac-
tions.6
ACS Paragon Plus Environment

Page 3 of 28
Journal of the American Chemical Society
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
Figure 1. Potential randomization sites (blue) in PAMO.
The flavin and NADP cofactors are displayed in orange
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
and yellow, respectively. Based on the 2YLX crystal
structure of PAMO.15
Table 2. Best Variants for Substrate 1b Identified by Saturation Mutagenesis (using lysates) at Amino Acid Posi-
tions 67, 152 and 442. Further Variants are detailed in the Supporting Information (Table S2)
Entry
1
Mutant
WT
% Conversion
% ee
91
Config.
% Sulfone
62
ND
ND
ND
63
99
99
99
55
S
S
7
8
16
11
5
12
4
3
3
1
2
A442S
83
49
56
61
3
A442D
S
4
A442Y
S
5
A442F/P440Y
A442T/P440Y
A442H/P440Y
A442Q/P440Y
I67T
R
R
R
R
S
6
57
55
36
44
61
7
8
9
10
11
I67G
24
18
S
I67A
52
70
70
82
79
72
76
70
65
69
60
97
97
97
97
97
S
1
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
I67S
51
S
2
3
7
1
I67L
99
99
99
99
99
99
99
99
99
62
50
49
57
S
I67T/P440Y
I67E/P440Y
I67G/P440Y
I67A/P440Y
I67F/P440Y
I67Q/P440Y
I67Y/P440Y
I67C/P440Y
Q152I
R
R
R
R
R
R
R
R
S
6
2
1
2
1
6
1
Q152V
S
1
Q152S
S
1
Q152M
S
1
Q152A
49
S
1
ND= not determined
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 4 of 28
One of several strategies would be to group these single which provided six variants showing (R)-selectivity in
1
amino acid positions into multi-residue saturation
sites,14 e.g., three 2-residue sites and then to apply ISM,
six upward pathways being possible. In order to keep
the screening effort as low as possible, we first per-
formed exploratory experiments by focusing saturation
mutagenesis separately at the following residues: V54,
I67, Q152 and A442. WT PAMO and the P440Y mutant
(showing 10% (R)-selectivity) were used as templates in
two separate series of experiments. Instead of applying
conventional NNK codon degeneracy encoding for all 20
canonical amino acids, the “smart-intelligent” library
construction strategy for avoiding amino acid bias re-
ported by Tang16a was used as an alternative to the 22c
trick applied previously to P450 monooxygenses.16b This
technique involves four pairs of primers with degenerate
codons of NDT (encoding 12 amino acids N, S, I, H, R, L,
Y, C, F, D, G and V), VMA (encoding 6 amino acids E, A,
Q, P, K, T), as well as ATG (encoding M) and TGG (en-
coding W) at the target sites, which were mixed in a
ratio of 12:6:1:1. Saturation mutagenesis at residue V54
failed to yield improved variants, in contrast to random-
ization at positions 67, 152 and 442 (Table 2). In each
case, only 60 variants had to be screened for complete
library coverage.
the range 70-75% ee (Supporting Information, Table S3).
The subsequent site to be targeted was residue I67, this
position also being a “hot spot” (site B). In principle, any
individual one of the six best mutants originating from
library A could be employed as a template for saturation
mutagenesis. Instead, we employed all six of them sim-
ultaneously, because this increases structural diversity,
while keeping screening at a minimum (20x3x6 = 360
transformants for 95% library coverage). In order to
eliminate amino acid bias, the method of Tang was used
once more. Upon screening only 90 transformants, sev-
eral improved variants were identified, the two best
ones being I67C/P440F/A442F/L443D (Variant ZGZ-1)
and I67Q/P440F/A442N/L443I (Variant ZGZ-2), show-
ing enantioselectivities of 92% ee and 95 % ee, respec-
tively, in favor of (R)-2. The resulting experimental ISM
scheme featuring pathway A → B which provides two
excellent variants is shown in Figure 2. The change in
free energy ∆∆Gǂ associated with reversing enantioselec-
tivity upon going from WT PAMO (90% ee S) to variant
ZGZ-2 (95% ee R) is considerable, amounting to 16.4
KJ/mol (7.3 KJ/mol + 9.1 KJ/mol – see Table 6).
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
In addition to improved (S)-selectivity (Table 2, entries
22-26), several variants with increased (R)-selectivity
were identified (Table 2, entries 5-8, 14-21), which indi-
cates that positions 67, 152 and 442 appear to be “hot
spots”. The best (R)-selective variant proved to be
I67T/P440Y (82% ee), originating from randomization at
position 67 using mutant P440Y as a template. It was
then used as a template for saturation mutagenesis at
site A442/L443, but this failed to deliver a notably im-
proved (R)-selective variant. Rather than applying the
previously developed strategy for escaping from local
minima by utilizing an inferior mutant for further ISM
experiments,14c we chose a different strategy.
The plan was to apply ISM, with WT PAMO serving as
the template, and P440/A442/L443 as the initial 3-
residue randomization site (site A). These residues be-
long to part of a loop that was known to be sensitive to
mutational changes in stereoselective BV reactions.7,10,13
NNK codon degeneracy encoding of all 20 canonical
amino acids would require considerable oversampling in
the screening step, in order to achieve 95% coverage (≈
98,000 transformants) according to traditional statis-
tics.14,17 In order to minimize such efforts, reduced ami-
no acid alphabets were chosen.14 Table 1 reveals that the
single mutants P440Y and P440L cause slight reversal of
enantioselectivity in favor of (R)-2, and P440F reduces
(S)-selectivity considerably. Thus, for position 440 only
these three amino acids were chosen as building blocks
(Y, L and F). At positions 442 and 443, NDT codon de-
generacy was used. Taking all three positions together,
the choice of the designed reduced amino acid alpha-
bets requires the screening of about 1700 transformants
for 95% library coverage. We screened 900 trans-
formants, corresponding to ≈ 80% library coverage,
Figure 2. ISM scheme in the directed evolution of
PAMO as a catalyst in the asymmetric sulfoxidation of
substrate 1b with preferential formation of (R)-2b.
In view of the high degree of stereoselectivity, visiting
further randomization sites (Figure 1) was not necessary.
However, it was of methodological interest to test the
alternative pathway B → A. The initial saturation muta-
genesis library was created at site B with WT plasmid
serving as template. Three mutants I67C, I67A and I67T
with reduced S-selectivity (and therefore going in the
right direction) were found after screening only 90 col-
onies for 95% library coverage. To explore the reverse
pathway B→A, site A was revisited with the same codon
degeneracy as before, mutants I67C, I67A and I67T
serving as templates. After screening 900 transformants,
one quadruple mutant I67A/P440Y/A442V/L443I (ZGZ-
3) with 94% ee was found (Figure 2). In theory, mutant
ZGZ-1 should be found in this library. Since only 900
transformants were screened corresponding to low li-
ACS Paragon Plus Environment

Page 5 of 28
Journal of the American Chemical Society
cient level of accuracy that would explain experimental
brary coverage (3x1700 = 5100 transformants for 95%
1
2
3
4
5
6
7
8
coverage), this result need not come as a surprise. Table
3 summarizes the final results of exploring the two ISM
pathways (data from reaction by purified enzymes).
differences of a single order of magnitude.
Table 5. Kinetic Parameters of Selected PAMO Mu-
tants as Catalysts in the Sulfoxidation of Thioether
1b.
Table 3. Best (R)-selective PAMO Mutants as Cata-
lysts in the Asymmetric Sulfoxidation of the Model
Compound 1b. Data using purified enzymes.
Km
(mM)
kcat
(s-1)
kcat/Km
(M-1s-1)
Entry Mutant Selectivity
9
Variant % Conversion
% ee
92
Config
% Sulfone
1
WT
S
S
S
S
S
S
R
R
0.58
0.43
1.26
0.22
0.61
0.97
0.39
0.92
0.30
0.14
0.69
371
1413
768
1351
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
ZGZ-1
ZGZ-2
ZGZ-3
96
97
99
R
R
R
14
6
2
3
4
5
6
7
8
Q152I
Q152V
Q152S
Q152M
Q152A
ZGZ-1
ZGZ-2
95
0.29
0.78
0.29
0.04
0.37
94
12
1175
1038
3649
1879
Thermostability tests. In order to check whether the
mutational changes influence thermal stability adverse-
ly, the T60 values were measured, the temperature at
50
which 50% of the residual activity in the reaction of
substrate 1b remains after a heat treatment for one
hour.17 Thermally robust WT PAMO has a T60 value of
Deconvolution experiments of best quadruple mu-
tant ZGZ-2. Studies that focus on uncovering (or pre-
dicting) the effect of a single point mutation on the
catalytic profile of an enzyme enrich our understanding
of biocatalytic mechanisms. The situation becomes
more complex when introducing two point mutations,
because the question of so-called additivity versus non-
additivity arises.12 When additivity pertains, for example
when considering enzyme activity, the influence of the
two respective separate point mutations add up mathe-
matically when they are combined in a double mutant.
This means that they do not interact with one another
on a molecular level. Fersht18a and later Wells18b were the
first to study such thermodynamic cycles in order to
address mechanistic questions, the two initial single
mutants being combined with formation of the respec-
tive double mutant. Initially, the majority of cases stud-
ied using various enzymes led to the conclusion that
additivity holds,18b but as time went on, a few exceptions
were noted.19 Accordingly, the enzyme activity of the
double mutant may prove to be either higher or lower
than expected on the basis of a mathematically additive
influence of the two single mutations.
50
57.6. Table 4 shows that with the exception of variant
ZGZ-1, no significant reduction in thermostability occurs
due to the respective mutational changes.
Table 4. T6050 Values (°C) of Selected Purified PAMO
Variants.
60
50
Entry
Variant
ꢀ
1
WT
57.6
56.2
57.4
56.6
56.4
57.8
53.2
56.4
2
3
4
5
6
7
8
Q152I
Q152V
Q152S
Q152M
Q152A
ZGZ-1
ZGZ-2
Determination of kinetic parameters. Kinetic pa-
rameters were measured for selected PAMO mutants as
catalysts in the sulfoxidation of substrate 1b. The usual
UV/Vis experimental setup measuring the consumption
of NADPH as a function of time was used throughout.
The results reveal that the kcat/Km values vary considera-
bly, in almost all cases catalytic efficiency being greater
than that of WT PAMO (Table 5). The lower Km value of
variant ZGZ-1 plays an important role, but thus far we
are not able to identify the effect at the molecular level.
Whereas induced fit docking computations proved to be
useful in understanding non-addditive mutational ef-
fects (see below), we decided not to perform such stud-
ies for ZGZ-1 because, to the best of our knowledge, it is
currently not possible to calculate Km values to a suffi-
In directed evolution, a double mutant constitutes the
simplest case in which additivity or non-additivity can
be studied. Deconvolution provides a means to assess
the individual catalytic contributions of both point mu-
tations. When more than two mutations are involved,
complete deconvolution requires the study of all indi-
vidual point mutations and of relevant combinatorial
sets of mutations.12,20 Influenced by new experimental
data,20 we recently postulated that non-additive effects
may be more common than traditionally assumed, and
that for this reason proteins behave as non-linear sys-
tems when multiple mutations are introduced.12 In such
cases a sound theoretical analysis of the effects provides
valuable mechanistic and structural information.20
ACS Paragon Plus Environment

Journal of the American Chemical Society
We performed deconvolution of the best quadruple
Page 6 of 28
1
2
3
4
5
6
7
8
mutant ZGZ-2 (I67Q/P440F/A442N/L443I), which
shows high (R)-selectivity in the sulfoxidation of sub-
strate 1b (Table 3). Conventional site directed mutagen-
esis using the QuikChange protocol of Stratagene was
applied,21 and the results are summarized in Table 6. It
should be noted that purified variant P440F leads to
high (S)-selectivity (97% ee), whereas the ee-value re-
sulting from the use of the lysate is only 56% (Table S1).
This can be explained by the difference in reaction con-
ditions and the fact that this particular variant leads to
significant overoxidation with formation of the sulfone,
which involves kinetic resolution of the chiral sulfoxide.
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
The deconvolution results are remarkable for several
reasons. Variant ZGZ-2 is highly (R)-selective (95% ee),
yet the four respective single mutants show opposite
enantioselectivity in favor of (S)-2b, some with very high
ee-values (P440F and L443I) as summarized in Table 6.
All of the possible double and triple mutants which are
combinatorially possible using the four point mutations
of variant ZGZ-2 were generated and tested in the model
reaction. Here again a number of unusual results were
observed which do not correspond to traditional addi-
tive mutational effects.12 For example, not only are the
four single mutants (S)-selective, but also the double
mutant A442N/L443I (54% ee) and the triple mutant
I67Q/A442N/L443I (38% ee) as well. Only upon addition
of the fourth point mutation P440F to the latter, does
drastic inversion to (R)-selectivity (variant ZGZ-2) oc-
cur. In contrast, the double combinations I67Q/P440F
and P440F/A442N lead to inverted (R)-selectivity, which
likewise signals cooperative non-additive mutational
effects.
Table 6. Results of Deconvoluting PAMO Quadruple Mutant ZGZ-2 as Catalysts (as purified enzymes) in the
Asymmetric Sulfoxidation of Thioether 1b. *
Code
Mutant
% Conversion
% ee
Config.
% Sulfone
-∆∆Gǂ (KJ/mol)
WT-PAMO
I67Q/P440F/A442N/L443I (ZGZ-2)
I67Q
69
97
70
96
54
96
90
26
41
90
95
69
97
69
98
73
24
81
S
R
S
S
S
S
R
S
S
R
S
S
R
R
S
R
0
6
7.3
9.1
abcd
a
10
59
4
4.2
10.4
4.2
11.4
4.6
1.2
b
P440F
c
A442N
d
L443I
23
5
ab
ac
I67Q/P440F
I67Q/A442N
19
12
14
32
6
ad
bc
bd
cd
abc
abd
acd
bcd
I67Q/L443I
5.6
4.8
4.6
3.0
7.3
4.3
2.0
4.4
P440F/A442N
P440F/L443I
95
94
80
98
88
15
75
73
54
90
70
38
71
A442N/L443I
I67Q/P440F/A442N
I67Q/P440F/L443I
I67Q/A442N/L443I
P440F/A442N/L443I
3
15
27
4
70
ACS Paragon Plus Environment

Page 7 of 28
Journal of the American Chemical Society
1
2
3
Had the four single mutants been generated separately
and not originating from deconvolution of an evolved
quadruple variant, no researcher would have combined
them in hope of reversing the sense of stereoselectivity in
favor of (R)-2b! Combining the mutations of improved
variants has been used in directed evolution numerous
times in hope of improving further a given catalytic prop-
erty (e.g., enantioselectivity),11,22 but failures have also
been noted22 (which unfortunately in other cases are not
always reported). Therefore, combining “positive” muta-
tions may or may not result in the targeted biocatalyst
improvement. Moreover, the present data suggests that
combining “deleterious” mutations may in fact provide
variants displaying highly improved catalytic profiles, but
predicting such combinations is currently a difficult task.
A step in this direction is to generate an understanding of
the effects that cause the observed strong non-additive
synergistic effects.
4
5
6
7
8
9
Figure 3. The 14 possible mutants in the evolutionary
protein space between WT and ZGZ-2. Lines indicate
only one trajectory of the 24 combinatorial possibilities
to go from WT to ZGZ-2 (Q=a, F=b, N=c, I=d).
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
The result of this analysis, based on experimental data, is
depicted in Figures 4 and 5. It can be seen that six path-
ways are energetically “favored” displaying smooth up-
ward climbs, while the majority (18 trajectories) are char-
acterized by local minima. Of course, due to the designed
construction of the fitness pathway landscape, all 24 tra-
jectories terminate at variant ZGZ-2.
Constructing a fitness pathway landscape. For a tuto-
rial on the construction and interpretation of fitness
pathway landscapes, see reference 23. The data derived
from complete deconvolution of the quadruple mutant
I67Q/P440F/A442N/L443I (ZGZ-2) (Table 6) leads to a
difficult to represent conventional fitness landscape, con-
sisting of a 5-dimensional surface in which the four point
mutations of variant ZGZ-2 are independent vectors and
∆∆Gǂ constitutes the dependent variable.
Therefore, we have designed the construction of an exper-
imental fitness pathway landscape characterized by all 4!
= 24 pathways (trajectories) leading from WT PAMO to
ZGZ-2. In a previous study we utilized this type of ap-
proach, an evolved enantioselective mutant of an epoxide
hydrolase mutant being the focus of interest. In that de-
convolution investigation, all 5! = 120 theoretically possi-
ble trajectories were mapped by stacking the respective
steps from top (WT) to bottom (best mutant).23 We have
since realized that it is graphically more appealing to
stack the steps of such trajectories upwards starting from
WT (bottom) and ending at the evolved highly enantiose-
lective mutant (top), which represents an upward climb.
In the present study we have chosen this graphical repre-
sentation. For simplicity, we denote the four mutations as
follows:
Figure 4. Fitness pathway landscape showing the 24
pathways leading from WT PAMO (bottom) to best (R)-
selective variant ZGZ-2, a typical trajectory lacking local
minima (green pathway) and one having local minima
(red) being featured.
I67Q = a; P440F = b; A442N = c; L443I = d.
We first generated the mutants that are relevant when
considering all combinations of mutations obtained suc-
cessively in the previous ISM experiments (Figure 3).
Following their assessment regarding enantioselectivity,
the free energy values were used in the construction of
the 24 upward pathways.
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 8 of 28
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 5. Fitness pathway landscape in the frontal view of Figure 4 of all 24 trajectories leading from WT PAMO to
variant ZGZ-2 characterized by four point mutations. Green notations indicate energetically favored pathways,
whereas red notations represent disfavored trajectories having local minima. Letters in red in the dendrogram denote
a local minimum after the introduction of this mutation.
Figures 4 and 5 illustrate that epistatic interactions occur
between individual point mutations and sets of point muta-
tions. Among the features of interest is the influence of mu-
tation d (L443I). Whereas it consistently impedes upward
climbs in early phases, it is always required in the final result
(ZGZ-2). In order to analyze the cumulative effects quantita-
tively in any given upward pathway, we applied a previously
developed procedure.23 In the present case, the experimental
data in Table 6 was used to calculate the free energy of inter-
manifold is induced. If mathematical additivity had per-
tained, then the respective double mutant would show about
the same (S)-selectivity as WT PAMO. The cooperative inter-
action between the two point mutations amounts to about 4
KJ/mol. Strong cooperativity also operates at all other points
along this pathway, in the final step Δꢁꢂ‡ꢃ amounting to about
10 KJ/mol. In the case of pathway d→a→b→c (Figure 7),
pronounced cooperative mutational effects are also visible
(as in all of the 24 trajectories).
action (Δꢁꢂ‡ꢃ) between any two mutations or two sets of mu-
tations, designated as i and j according to Eq. 1, where ΔΔꢁꢂ‡,
ΔΔꢁ‡, and ΔΔꢁꢄ‡ꢅꢆ are the experimentally determined differ-
ꢃ
ences in activation energies for the formation of both enanti-
omers using mutants i, j, and the binary combination of i and
j, respectively. This thermodynamic cycle is in analogy to the
scheme originally suggested by Fersht17a regarding the inter-
action of two point mutations.
Δꢁꢂ‡ꢃ ꢇ ΔΔꢁꢄ‡ꢅꢆ ꢈ ꢉΔΔꢁ‡ ꢊ ΔΔꢁ‡ꢋ
Eq. 1
ꢂ
ꢃ
For illustrative purposes, a green pathway a→b→c→d (Fig-
ure 6) and a red one d→a→b→c (Figure 7) are analyzed here
(see SI for other trajectories). Figure 6 shows that in the
initial phase of the energetically favored pathway the first
mutation causes a slight decrease in (S)-selectivity, but the
second mutation alone causes a shift in the wrong direction,
i.e., (S)-selectivity increases relative to WT. Functioning in
concert, reversal of enantioselectivity in favor of the (R)-
Figure 6. Thermodynamic cycle (Eq. 1) highlighting the
interaction of point mutations and sets of mutations
involved at every stage along the energetically favored
ACS Paragon Plus Environment

Page 9 of 28
Journal of the American Chemical Society
pathway a→b→c→d toward variant ZGZ-2 as a catalyst
in the sulfoxidation of substrate 1b.
Table 7. Mutants of PAMO Studied with Model Sub-
1
2
3
4
5
6
strate 1b by Induced Fit Docking. Predicted Selectivi-
ty is Based on the Orientation of the Substrate in the
Active Site. Observed Selectivity is Based on Experi-
mental ee-value (Table 6). Distance Between the
Substrate Sulfur and the Hydroperoxo Oxygen Fur-
thest from the FAD Cofactor (r).
7
8
9
Mutant
Predicted
selectivity
Observed
selectivity
r(O-S)
[Å]
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
WT
P440F
S
S
S
S
R
R
S
S
S
S
R
R
4.16
5.35
5.78
5.64
5.55
3.66
L443I
P440F/L443I
I67Q/P440F/L443I
ZGZ-2
Figure 7. Thermodynamic cycle (Eq. 1) highlighting the
interaction of point mutations and sets of mutations
involved at every stage along the energetically disfa-
vored pathway d→a→b→c toward variant ZGZ-2 as a
catalyst in the sulfoxidation of substrate 1b.
Uncovering the reasons for non-additive mutational
effects. In an attempt to rationalize the observed non-
additive effects favoring cooperativity (synergy), induced
fit docking calculations5,24 were performed with the model
substrate 1b, on the WT enzyme, along with the mutant
enzymes detailed in Table 7. The FAD-OO- species has
been suggested to be the oxidizing species for the Baeyer-
Villiger
reaction
catalyzed
by
cyclohexanone-
monooxygenase (CHMO). This reaction has been studied
previously using QM/MM calculations.5 As the precise
mechanism and oxidizing species responsible for sulfoxi-
dation of thioethers have yet to be elucidated, the FAD
cofactor was modeled in the hydroperoxo form (FAD-
OOH). The likely oxidizing species for thioethers are the
FAD-OOH or FAD-OO- moieties. A QM/MM mechanistic
study may shed light on this issue, but is beyond the
scope of the current work. Our primary goal here is to
explore the effect of the mutations on the binding orien-
tation of the substrate within the active site, which will in
turn aid in predicting the selectivity of sulfoxidation. We
assume that the closest of the two sulfur lone pairs to the
-OOH moiety will be the one that undergoes oxidation,
yielding either (R)- or (S)-2b (Figure 8).
Figure 8. Oxidation of 1b by PAMO. Oxidation at the
blue and green lone pairs will lead to formation of R-
and S-2b, respectively
We observe that the selected docking poses exhibit a
range of O-S distances between the substrate sulfur and
the flavin (between 3.66 and 5.78 Å - see Table 7). We also
note that the mutants corresponding to docking poses
with distances greater than
5
Å
(P440F, L443I,
P440F/L443I and I67Q/P440F/L443I) are those that are
found experimentally to result in significant amounts of
sulfone.
A range of binding poses was obtained for each mutant.
Poses were considered for analysis where a distance of
less than 6 Å was observed between the substrate sulfur
and the hydroperoxo oxygen furthest from the FAD cofac-
tor. Binding poses of 1b in the active site of the WT and
selected mutants are depicted in Figure 9. These poses are
those in which the highest induced fit docking score was
obtained. The observed binding poses of substrate 1b in
the active site for all of the simulated mutants are con-
sistent with the experimentally observed enantioselectivi-
ty (Table 7).
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 10 of 28
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 9 Induced fit docking results for substrate 1b in (a) WT PAMO; (b) P440F; (c) L443I; (d) P440F/L443I; (e)
I67Q/P440F/L443I; (f) I67Q/P440F/A442N/L443I mutants. The mutated residues are displayed in yellow. Non-polar
hydrogen atoms are omitted for clarity. ꢈꢌꢌꢍ‡ values are given in KJ/mol and r(O-S) in Å.
The single mutants P440F and L443I display higher con-
version and enantioselectivity for (S)-2b than the WT
enzyme. Both of these mutations result in a change in the
shape of the active site pocket. The 443 residue lines the
pocket, but whilst the 440 residue is on the same loop, it
does not line the pocket. The L443I mutation hence
changes the shape of the pocket directly, but the influ-
ence of the P440F mutation is exerted by reducing the
backbone rigidity, enabling the movement of the loop
containing L443 towards the pocket. The I67 sidechain
adopts a different conformation in these single mutants,
compared to the WT enzyme, despite being on a different
loop to the mutated residues. This may result in the
change in shape of the active site. Because of the change
in conformation of I67, both single mutants display a
different binding pose of 1b in the active site (Figures 9(b)
and (c)), compared to the WT (Figure 9(a)), although in
both cases these still favor formation of (S)-2b.
The quadruple mutant I67Q/P440F/A442N/L443I, is the
most (R)-selective mutant studied here. The additional
A442N mutation results in a change in conformation
about the –OOH moiety (Figure 9(f)). A hydrogen bond is
formed between the Gln sidechain carbonyl oxygen and
the hydroperoxo hydrogen atom. This results in a differ-
ent conformation of the R337 side chain, which forms a
hydrogen bond to the same hydroperoxo group in the
other mutants. This in turn allows 1b to get closer to the
hydroperoxo group in a position favoring oxidation of the
(R)-lone pair, and may explain the higher degree of (R)-
enantioselectivity, compared to the triple mutant
(I67Q/P440F/L443I).
It is clear from the experimental results, that the P440F
mutation is important for obtaining (R)-selectivity, as the
triple mutant that does not contain this mutation
(I67Q/A442N/L443I, Table 7) displays (S)-selectivity,
although to a lesser extent than the WT. This is very in-
teresting, considering that this mutation alone increases
(S)-selectivity. The docking results described here support
the importance of this mutation, and also show the role of
the additional mutations for achieving the high level of
reversed (R)-selectivity.
The double mutant P440F/L443I is also selective towards
formation of (S)-2b, however, whilst the conversion is
higher than in the case of the WT, the ee is smaller. The
binding pose of 1b in the double mutant (Figure 9(d)) is
similar to that of the two single mutants (Figures 9(b) and
(c)) and hence it is surprising that this double mutant
shows a decrease in selectivity.
Exploring substrate scope of the evolved PAMO mu-
tants. The best (R)- and (S)-selective PAMO variants
evolved for thioether 1b were tested as catalysts in the
asymmetric sulfoxidation of several other substrates 1a,
4a-4c, 7 and 10 (Scheme 2), using purified enzymes. Since
lysates and purified BVMO variants may lead to slightly
different results in terms of ee-values and undesired
overoxidation with formation of the respective sulfones,
The triple mutant I67Q/P440F/L443I mutation is found
experimentally to be selective towards formation of (R)-
2b. In the binding pose shown in Figure 9(e), the sulfur
atom of 1b is close to the Q67 NH2 group, which may
interact with the pro-(S) lone pair, thus leaving the pro-
(R) lone pair available for reaction with the FAD moiety.
ACS Paragon Plus Environment

Page 11 of 28
Journal of the American Chemical Society
oxidation of the model thioether 1b was included in this
series of experiments. As already noted, when overoxida-
1
2
3
4
5
6
7
8
tion occurs to a significant degree, kinetic resolution may
influence the apparent ee-value of sulfoxidation. Fortu-
nately, mutants having the highest stereoselectivity cause
only small overoxidation. Highly (R)-selective variants for
all of the thioethers were identified (> 95% ee) except for
1a (91% ee), while equally high (S)-selectivity proved to be
possible only for substrates 1b, 7 and 10 (Table 8).
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Scheme 2. Additional Substrates used in PAMO-catalyzed Asymmetric Sulfoxidation.
Table 8. Performance of Selected Purified PAMO Mutants as Catalysts in Asymmetric Sulfoxidation Reactions.
Entry
Substrate
1a
Variant
WT
Major Product
% Conversion
% ee
99
91
Config.
% Sulfone
1
2
2a
2a
2b
2b
2b
2b
2b
5a
5a
5a
5b
5b
5b
5b
99
99
81
S
R
S
S
S
S
S
R
R
S
S
S
R
R
13
11
4
1a
ZGZ-1
Q152I
Q152V
Q152S
Q152M
Q152A
WT
3
1b
97
99
96
99
97
42
95
91
4
1b
97
66
95
78
53
78
50
83
93
99
99
6
5
1b
6
6
1b
4
7
1b
5
8
4a
0
9
4a
P440L
I67A
24
0
10
11
12
13
14
4a
4b
4b
4b
4b
WT
9
4
I67A
94
99
94
15
9
ZGZ-1
ZGZ-2
8
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 12 of 28
15
16
17
4b
4b
4c
4c
4c
4c
4c
4c
7
P440F/A442F/L443D
5b
5b
5c
5c
5c
5c
5c
5c
8
99
99
35
95
90
16
R
R
S
20
11
1
2
3
4
5
6
7
8
P440F/A442N/L443I
WT
5
18
19
20
21
I67A
85
99
99
99
98
63
42
98
88
69
81
93
98
97
95
92
12
S
12
19
22
29
19
7
ZGZ-1
R
R
R
R
S
ZGZ-2
9
P440F/A442F/L443D
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
22
23
24
25
26
27
28
29
30
31
P440F/A442N/L443I
WT
7
P440L
8
97
93
96
91
R
R
S
0
7
ZGZ-1
8
18
10
0
7
A442D
WT
8
10
10
10
10
10
11
S
Q152S
11
94
95
94
96
S
0
Q152V
11
73
S
0
ZGZ-1
11
90
88
R
R
0
ZGZ-2
11
0
CONCLUSIONS AND PERSPECTIVES
in concert four (S)-selective mutations lead to high (R)-
selectivity. This prompted us to perform complete decon-
volution by generating all combinatorial double and triple
mutants. Here again strong cooperative non-additive
effects were observed. Using the experimental data, a
designed fitness pathway landscape describing all 24 up-
ward pathways leading from WT PAMO to the best mu-
tant ZGZ-2 was constructed. Local minima were observed
in 18 of the 24 trajectories, and in all of them pronounced
cooperative non-additive mutational effects were found.
The six energetically favored pathways that lack this fea-
ture likewise reveal synergistic mutational effects (non-
additivity).
In this study we have applied directed evolution to phenyl
acetone monooxygenase (PAMO), a thermally robust
Baeyer-Villiger monooxygenase, as a catalyst in the
asymmetric sulfoxidation of prochiral thioethers. Since
WT PAMO generally favors the formation of (S)-
sulfoxides, emphasis was on the evolution of reversed
eantioselectivity in favor of the (R)-products. This goal
was reached (up to 95% ee) by knowledge-driven iterative
saturation mutagenesis (ISM) with formation of high-
quality mutant libraries requiring a minimum of screen-
ing (the bottleneck of directed evolution), methyl-p-
methylbenzyl thioether serving as the model substrate.
WT PAMO shows 90% ee in favor of the (S)-sulfoxide,
while the best (R)-selective mutant ZGZ-2, characterized
by four point mutations I67Q/P440F/A442N/L443I, leads
to 95% ee, which means a change in free energy of ∆∆Gǂ =
16.4 KJ/mol. This mutant also displays high enantioselec-
tivity in the reaction of several other thioethers, and con-
sequently constitutes an alternative to other catalysts in
asymmetric sulfoxidation.1-4 The four new point muta-
tions do not impair the high thermostability of the en-
zyme, which is of practical importance. This also means
that, if necessary, ZGZ-2 can be used as a template in
future protein engineering studies.
In a previous study concerning the same type of analysis
involving the epoxide hydrolase from Aspergillus niger as
a catalyst in the hydrolytic kinetic resolution of a racemic
epoxide, we constructed an analogous fitness landscape
characterized by 5! = 120 pathways, about half of them
proved to be energetically favored due to the lack of local
minima.23 Thus, although ISM was applied in both cases
in the evolution of the respective best mutant, the two
landscapes differ somewhat in terms of the frequency of
local minima. This result should be taken as yet another
warning that experimental data derived from laboratory
evolution should not be used to draw conclusions regard-
ing the question of how many pathways in natural Dar-
winian evolution lead to fitter proteins.12,23 To avoid con-
fusion, it should be noted that fitness pathway landscapes
obtained by deconvolution of a given evolved mutant as
featured here and in a previous study23 are fundamentally
different from those that are constructed by studying
experimentally all theoretically possible pathways in an
ISM scheme.14c
An intriguing spinoff which emerged during our investi-
gation concerns the question of additive versus non-
additive mutational effects.12 Upon partial deconvolution
of
the
most
(R)-selective
mutant
I67Q/P440F/A442N/L443I (variant ZGZ-2) with genera-
tion of the four respective single mutants, it was discov-
ered that all of them display the opposite sense of enanti-
oselectivity in favor of the (S)-sulfoxide. This means that
ACS Paragon Plus Environment

Page 13 of 28
Journal of the American Chemical Society
4000 rpm for 30 min. 300 μL of the organic layer in each
The extreme synergistic mutational effects observed in
1
2
3
4
5
6
7
8
the present study support once more the hypothesis that
genetic changes in proteins are part of non-linear sys-
tems.12,23 Such effects are difficult to predict. Nevertheless,
once observed, we were able to rationalize them by a
theoretical analysis based on induced fit docking. Hope-
fully, insights of this type will someday lead to viable
models with predictive power.
well was transferred into 96 well plates and then filtered
through 96 well filter plates with 0.22 µL diameter (Acro-
Prep Advance 96 well Filter Plate). The organic layer was
transferred into 96 well glass-made plates and was sub-
jected to HPLC analysis. The mutants with improved
selectivity were picked out and the results were confirmed
by repeating the reaction in Eppendorf tubes. The muta-
tions in the gene of the mutants were confirmed by se-
quencing.
9
EXPERIMENTAL SECTION
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Enzyme Expression and Purification. The glycerol
stocks of PAMO WT and mutants were inoculated into 4
mL of LB media supplemented with 100 µg/mL carbenicil-
line. After overnight growth at 37 °C, 2.5 mL of culture
were transferred into 250 mL of TB media containing 100
µg/mL carbenicilline and 0.1g/100mL L-arabinose as in-
ducer at 30 °C with shaking at 200 rpm for 20 hours for
enzyme expression. Cell pellets were harvested by cen-
trifugation and washed once with 50 mM, pH 9.0 Tris-
HCl buffer. The pellets were suspended into 10 mL 50 mM
Tris-HCl buffer (pH 9.0) and were disrupted by soni-
cation. The supernatant containing the crude enzyme was
obtained by centrifugation at 4 °C and 12000 rpm for 40
min. After going through the filter, the supernatant was
subjected to affinity chromatography (GE Healthcare
HisTrap FF Crude column) for purification. The impurity
was removed by flushing the column with 25 mM imi-
zadol and the enzyme was eluted by 50 mM Tris-HCl
buffer containing 200 mM imizadol and 0.5 M NaCl. The
elution fractions were subjected to an Amicon Ultra-15
centrigugal filter for removal of salt and imizadol. The
purified enzymes were dissolved in 50 mM, pH 9.0 Tris-
HCl buffer and stored at -80 °C for use. The purity of the
enzyme was confirmed by SDS-PAGE (Figure S1 in Sup-
porting Information). The concentration of the purified
enzyme was determined by Bradford protein assay meth-
od, following the provided protocol.
Library Construction. The library A and B were created
by the QuikChange method with the plasmid of the WT
and corresponding mutants as a template. The primers
used in the library creation process are listed in the Sup-
porting Information (Table S3). The PCR reaction mixture
contained: 10×KOD Buffer (Novagen, 5 μL), Mg2SO4 (25
mM, 2 μL), dNTPs (2 mM each, 5 μL), template plasmid
(20 ng/μL, 0.5 μL), KOD Polymerase (1 U). The PCR pro-
gram contained an initial denaturation step at 95 °C for 3
min followed by 18 cycles of denaturation at 95 °C for 1
min, annealing at 72 °C for 1 min and elongation at 72 °C
for 14 min. The final elongation at 72 °C for 35 min was
then performed. The PCR product was digested by adding
5 units DpnI at 37 °C for 4 hours to remove the template
plasmid. The treated PCR product was transformed into
TOP 10 electrocompetent cells. The cells were spread on
the LB-agar plates supplemented with 100 µg/mL carbeni-
cilline for overnight incubation at 37 °C.
Library Screening. Individual colonies grown on the
agar plates were picked into 96 deep-well plates contain-
ing 800 µL of LB media supplemented with 100 µg/mL
carbennicilline. After overnight growth at 37 °C, with
shaking at 800 rpm, 150 µL of LB culture were transferred
into a 96-well plate containing 70 µL 70 % glycerol, and
mixed for preparation of glycerol stock plates. The glycer-
ol stock plates were stored in a -80 °C freezer. 10 μL of
each culture was used to inoculate a new plate containing
800 µL of TB media with 100 µg/mL of carbenicilline and
0.1 % arabinose as an inducer for enzyme expression. The
new inoculated plates were grown for an additional 20
hours to express the PAMO enzyme at 30 °C, with shaking
at 800 rpm. The cultures were centrifuged at 4000 rpm
and at 4 °C for 15 min to collect cells. After discarding the
supernatants, the cells in each well were suspended in
400 µL 50 mM, pH 9.0 Tris-HCl buffer supplemented with
1 mg/mL lysozyme and 5 unites of Dnase I. The plates
were incubated at 37 °C with shaking at 800 rpm for 3
hours, and then centrifuged at 4 °C and 4000 rpm for 40
min.
Site-Directed Mutagenesis. The PAMO mutants were
generated by using the QuikChange method,20 following
the protocol used in the library creation process. The only
difference here is that primers with the targeted codon
were used instead of degenerate primers. The primers
used are summarized in the Supporting Information (Ta-
ble S4). The targeted mutation in the gene of mutants was
confirmed by sequencing.
General Procedure for the Biotransformation of Sub-
strates.
2 mM of substrate was dissolved in 1 mL of 50 mM Tris-
HCl buffer (pH 9.0), with 2 % (v/v) acetonitrile as cosol-
vent. The reaction mixture contained 0.2 mM NADPH, 5
mM Glucose-6-phosphate, and 2 unites of Glucose-6-
phosphate dehydrogenase as the NADP+ regeneration
system. The reaction mixture was shaken at 1200 rpm and
30 °C for the times indicated. The reaction mixture was
vigorously shaken after adding 300 µL HPLC grade
chloromethane to extract the product. After centrifuga-
tion at 4 °C for 30 min, the organic layer was extracted by
syringe and filtered with 0.22 µm filter (Whatman). The
100 μL of each supernatant was transferred into a new 96
deep-well plate to prepare the reaction mixture. In each
well of the plate, 1 mL of reaction mixture was added,
containing 5 mM Glucose-6-phosphate, 2 U glucose-6-
phosphate dehydrogenase, 0.2 mM NADPH and 2 mM
substrate with 2 % (v/v) acetonitrile. The reaction plates
were incubated at 30 °C with shaking at 800 rpm for 6
hours. After 600 µL of dichloromethane was added into
each well, the reaction plates were centrifuged at 4 °C and
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 14 of 28
Mitka, M.; Kowalski, P. Tetrahedron 2005, 61, 8315–8327; d) Liao
samples were analyzed by HPLC equipped with Chiralcel
OB-H column to determine the conversion of sulfides and
the enantiomeric excesses of sulfoxides. The absolute
configuration of the sulfoxide products was determined
by comparison of their retention time with related litera-
ture or commercially available authentic compounds
(Table S6 in Supporting Information).
1
2
3
4
5
6
7
8
S.; List, B. Adv. Synth Catal. 2012, 354, 2363–2367; e) Legros, J.;
Bolm, C. Angew. Chem. Int. Ed. 2004, 43, 4225–4228 ; f) Legros, J.;
Bolm, C. Chem. Eur. J. 2005, 11, 1086–1092; g) Fukisaki, J.;
Matsumoto, K.; Matsumoto, K.; Katsuki, T. J. Am. Chem. Soc.
2011, 133, 56–61; h) Capozzi, M. A.; Capitelli, F.; Bottoni, A.; Cal-
varesi, M.; Cardellicchio, C. ChemCatChem 2013, 5, 210–219; i)
O’Mahoney, G. E.; Ford, A.; Maguire, A. R. J. Org. Chem. 2012, 77,
3288–3296; j) Newhouse, T. R.; Li, X.; Blewett, M. M.; Whitehead,
C.; Clare, M. C.; Corey, E. J. J. Am. Chem. Soc. 2012, 134, 17354–
17357.
[2] Recent key reviews and studies of organocatalytic asym-
metric sulfoxidation: a) Jurok, R.; Hodačová, J.; Eigner, V.;
Dvořáková, H.; Setnička, V.; Cibulka, R. Eur. J. Org. Chem. 2013,
7724–7738; b) Russo, A.; De Fusco, C.; Lattanzi, A. Chem-
CatChem 2012, 4, 901–916; c) Stingl, K. A.; Tsogoeva, S. B. Tetra-
hedron: Asymmetry 2010, 21, 1055–1074; c) Bäckvall, J.-E. In Mod-
ern Oxidation Methods; J.-E. Bäckvall Eds.; Wiley-VCH, Wein-
heim: 2010, pp. 277–313; d) Armstrong, A. In Enantioselective
Organocatalysis; Dalko, P. I. Eds.; Wiley-VCH, Weinheim: 2007,
pp. 403–424; e) Imada, Y.; Kitagawa, T.; Wang, H.-K.; Komiya,
N.; Naota, T. Tetrahedron Lett. 2013, 54, 621–624; f) Ménová, P.,
Cibulka, R. J. Mol. Catal. A 2012, 363–364, 362–370; g) Lindén, A.
A.; Hermanns, N.; Ott, S.; Krüger, L.; Bäckvall, J.-E. Chem. Eur. J.
2005, 11, 112–119; h) Imada, Y.; Iida, H.; Murahashi, S.-I. Angew.
Chem. Int. Ed. 2005, 44, 1704-1706.
[3] Reviews of structure, mechanism, protein engineering and
applications of BVMOs: a) Walsh, C. T.; Chen, Y.-C. J. Angew.
Chem. 1988, 100, 342–352; b) Orru, R.; Dudek, H. M.; Martinoli,
C.; Torres Pazmiño, D. E.; Royant, A.; Weik, M.; Fraaije, M. W. J.
Biol. Chem. 2011, 286, 29284–29291; c) Kayser, M. M. Tetrahedron
2009, 65, 947–974; d) Mihovilovic, M. D.; Drauz, K.; Gröger, H.;
May, O. Enzyme Catalysis in Organic Synthesis Vol. I–III, 3rd ed.,
Wiley-VCH, Weinheim, 2012, pp. 1439–1485; e) Mihovilovic, M.
D. Curr. Org. Chem. 2006, 10, 1265–1287; f) Wohlgemuth, R.
Engineering in Life Sciences 2006, 6, 577–583; g) Leisch, H.; Mor-
ley, K.; Lau, P. Chem. Rev. 2011, 111, 4165–4222; h) Hollmann, F.;
Arends, I. W. C. E.; Buehler, K.; Schallmey, A.; Bühler, B. Green
Chem. 2011, 13, 226–265; i) Balke, K.; Kadow, M.; Mallin, H.; Saß,
S.; Bornscheuer, U. T. Org. Biomol. Chem. 2012, 10, 6249–6265; j)
Zambianchi, F.; Pasta, P.; Colonna, S.; Gaggero, N.; Woodley, J.
M. Biotechnol. Bioeng. 2002, 78, 489-496.
Thermostability of PAMO Mutants. 40 µM of the puri-
fied mutant and WT enzyme were subjected to heating at
50 °C, 52 °C, 54 °C, 56 °C, 58 °C and 60 °C for 1 hour. The
samples were then put on ice for cooling for 10 min. The
residual activities of the enzymes were determined at 30
°C, as above.
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Kinetic parameters. The purified enzymes were sub-
jected to the determination of enzyme kinetic constants.
The activity of the enzyme was measured by monitoring
the consumption of NADPH at 25 °C and 340 nm
(ε340=6.22 mM-1 cm-1) during a fixed time. Each 1 mL
reaction system consisted of 50 mM, Tris-HCl (pH 9.0),
100 µM NAPDH, 0.1-0.5 mM substrate, 2% (v/v) acetoni-
trile, 0.05-0.5 µM purified enzyme. The measurements
were performed by using the Molecular Devices Spectra-
max. The data from the spectrophotometric assay were
used to calculate the kinetic constants of enzymes using
the Michaelis-Menten equation.
Molecular docking calculations. Docking was per-
formed using the Induced Fit Docking procedure devel-
oped by Schrödinger.25 The crystal structure of PAMO
containing FAD and NADP was used (PDB entry 2YLR).15
The receptor model was build using the Protein Prepara-
tion Wizard.26 Protonation states were determined using
PROPKA.27 Prior to docking, each protein structure was
energy minimized using the OPLS2005 force field,28 with
all heavy atoms restrained to an RMSD of 0.3 Å.
Supporting Information. Reaction data from additional
PAMO mutants, primers used for library creation and site-
directed mutagenesis, chiral HPLC separation conditions,
thermodynamic analysis for deconvolution of mutants, and
detailed results from Induced Fit Docking calculations. This
material is available free of charge via the Internet at
http://pubs.acs.org.
[4] Reviews and key papers of BVMO-catalyzed asymmetric
sulfoxidation: a) Matsui, T.; Dekishima, Y.; Ueda, M. Appl.
Microb. Biotechnol. 2014, 98, 7699–7706; b) Riebel, A.; de
Gonzalo, G.; Fraaije, M. W. J. Mol. Catal. B 2013, 88, 20–25; c)
Bisagni, S.; Summers, B.; Kara, S.; Hatti-Kaul, R.; Grogan, G.;
Mamo, G.; Hollmann, F. Top. Catal. 2014, 57, 366–375.
[5] Polyak, I.; Reetz, M. T.; Thiel, W., J. Am. Chem. Soc., 2012,
134, 2732–2741.
[6] Reetz, M. T.; Daligault, F.; Brunner, B.; Hinrichs, H.;
Deege, A. Angew. Chem. Int. Ed. 2004, 43, 4078–4081.
AUTHOR INFORMATION
Corresponding Author
reetz@mpi-muelheim.mpg.de
[7] a) Fraaije, M. W.; Wu, J.; Heuts, D. P. H. M.; van
Hellemond, E. W.; Spelberg, J. H. L.; Janssen, D. B. Appl. Micro-
biol. Biotechnol. 2005, 66, 393–400; b) Zambianchi, F.; Fraaije, M.
W. Carrea, G.; de Gonzalo, G.; Rodriguez, C.; Gotor, V.; Ottolina,
G. Adv. Synth. Catal. 2007, 349, 1327–1331; c) Rodriguez, C.; de
Gonzalo, G.; Torres Pazmiño, D. E.; Fraaije, M. W.; Gotor, V.
Tetrahedron: Asymmetry 2008, 19, 197–203; d) Zhang, Z.-G.;
Parra, L. P.; Reetz, M. T. Chem. Eur. J. 2012, 18, 10160–10172.
[8] WT PAMO as a sulfoxidation catalyst: de Gonzalo, G.; To-
rres Pazmiño, D. E.; Ottolina, G.; Fraaije, M. W. Tetrahedron
Asymm. 2005, 16, 3077–3083.
Author Contributions
The manuscript was written through contributions of all
authors.
Funding Sources
Financial support by the Max-Planck-Society and the
Arthur C. Cope foundation is gratefully acknowledged.
REFERENCES
[1] Recent key reviews and studies of transition metal cata-
lyzed asymmetric sulfoxidation: a) Kagan, H. B. In Organosulfur
Chemistry in Asymmetric Synthesis; Toru, T., Bolm, C., Eds.;
Wiley-VCH, Weinheim: 2009, pp. 1–29; b) Pelllissier, H. Tetrahe-
dron 2006, 62, 5559–5601; c) Kaczorowska, K. B.; Kolarska, Z.;
[9] Malito, E.; Alfieri, A.; Fraaije, M. W.; Mattevi, A. Proc. Natl.
Acad. Sci. USA 2004, 101, 13157–13162.
ACS Paragon Plus Environment

Page 15 of 28
Journal of the American Chemical Society
[10] Torres Pazmiño, D. E.; Snajdrova, R.; Rial, D. V.; Mihovi-
lovic, M. D.; Fraaije, M. W. Adv. Synth. Catal. 2007, 349, 1361–
1368.
[23] Reetz, M. T.; Sanchis. J. ChemBioChem 2008, 9, 2260–
2267.
[24] Polyak, I.; Reetz, M. T.; Thiel, W. J. Phys. Chem. B 2013,
1
2
3
4
5
6
7
8
9
[11] Reviews of directed evolution: a) Directed Evolution Li-
brary Creation; Gillam, E. M. J., Copp, J. N., Ackerley, D. F., Eds.;
2nd ed., Methods in Molecular Biology, Humana Press, Totowa,
2014; b) Widersten, M. Curr. Opin. Chem. Biol., 2014, 21, 42–47; c)
Siloto, R. M. P.; Weselake, R. J. Biocatal. Agric. Biotechnol., 2012,
1, 181–189; d) Goldsmith, M.; Tawfik, D. S. Curr. Opin. Struct. Biol.
2012, 22, 406–412; e) Cobb, R. E.; Chao, R.; Zhao, H. AIChE J,
2013, 59, 1432–1440; f) Reetz, M. T. Directed evolution of en-
zymes. In Enzyme Catalysis in Organic Synthesis; Drauz, K.,
Dröge, H., May, O., Eds.; Wiley-VCH: Weinheim, 2012; Vol. 1, pp
119–190; g) Bommarius, A. S.; Blum, J. K.; Abrahamson, M. J.
Curr. Opin. Chem. Biol., 2011, 15, 194−200; h) Quin, M. B.;
Schmidt-Dannert, C. ACS Catal., 2011, 1, 1017−1021; i) Brustad, E.
M.; Arnold, F. H. Curr. Opin. Chem. Biol., 2011, 15, 201−210; j)
Jäckel, C.; Hilvert, D. Curr. Opin. Biotechnol., 2010, 21, 753−759;
k) Turner, N. J. Nat. Chem. Biol., 2009, 5, 567–573; l) Lutz, S.;
Bornscheuer, U. T. Protein Engineering Handbook; Wiley-VCH:
Weinheim, 2009; Vol. 1–2.
[12] Reetz, M. T. Angew. Chem. Int. Ed. 2013, 52, 2658–2666.
[13] Directed evolution studies of PAMO as the catalyst in
asymmetric Baeyer-Villiger reactions: a) Parra, L. P.; Agudo, R.
Reetz, M. T. ChemBioChem 2013, 14, 2301–2309; b) Zhang, Z.-G.;
Roiban, G.-D.; Acevedo, J. P.; Polyak, I.; Reetz, M. T. Adv. Synth.
Catal. 2013, 355, 99–106; c) Wu, S.; Acevedo, J. P.; Reetz, M. T.
Proc. Natl. Acad. Sci. U. S. A. 2010, 107, 2775–2780.; d) Reetz, M.
T.; Wu, S. Chem. Commun. 2008, 5499–5501.
[14] a) Reetz, M. T. Angew. Chem. Int. Ed. 2011, 50, 138–174; b)
Acevedo-Rocha, C. G.; Kille, S.; Reetz, M. T. In Directed Evolu-
tion Library Creation: Methods and Protocols, 2nd Edition;
Ackerley, D., Copp, J., Gillam, E., Eds.; Methods in Molecular
Biology; Humana Press: Totowa, 2014; c) Gumulya, Y.; Sanchis,
J.; Reetz, M. T. ChemBioChem 2012, 13, 1060-1066.
[15] Orru, R.; Dudek, H. M.; Martinoli, C.; Torres Pazmino, D.
E.; Royant, A.; Weik, M.; Fraaije, M. W.; Mattevi, A., J. Biol.
Chem., 2011, 286, 29284–29291.
[16] a) Tang, L.; Gao, H.; Zhu, X.; Wang, X.; Zhou, M.; Jiang, R.
BioTechniques 2012, 52, 149–158; b) Kille, S.; Acevedo-Rocha, C.
G.; Parra, L. P.; Zhang, Z.-G.; Opperman, J. J.; Reetz, M. T.;
Acevedo, J. P. ACS Synth. Biol. 2013, 2, 83–92.
[17] Patrick, W. M.; Firth, A. E. Biomol. Eng. 2005, 22, 105–112.
[18] a) Carter, D. C.; Winter, G.; Wilkinson, A. J.; Fersht, A. R.
Cell 1984, 38, 835–840.; b) Wells, J. A. Biochemistry 1990, 29,
8509–8517.; c) see also Mildvan, A. S.; Weber, D. J.; Kuliopulos,
A. Arch. Biochem. Biophys. 1992, 294, 327–340.
117, 4993–5001.
[25] (a) Schrödinger Suite 2014-2 Induced Fit Docking proto-
col; Glide version 6.3, Schrödinger, LLC, New York, NY, 2013;
Prime version 3.6, Schrödinger, LLC, New York, NY, 2014. (b)
Farid, R.; Day, T.; Friesner, R. A.; Pearlstein, R. A., Bioorg. & Med.
Chem., 2006, 14, 3160–3173. (c) Sherman, W.; Day, T.; Jacobson,
M. P.; Friesner, R. A.; Farid, R., J. Med. Chem., 2006, 49, 534–553.
(d) Sherman, W.; Beard, H. S.; Farid, R., Chemical Biology &
Drug Design, 2006, 67, 83–84.
[26] (a) Schrödinger Suite 2014-2 Protein Preparation Wizard;
Epik version 2.8, Schrödinger, LLC, New York, NY, 2014; Impact
version 6.3, Schrödinger, LLC, New York, NY, 2014; Prime ver-
sion 3.5, Schrödinger, LLC, New York, NY, 2014. (b) Sastry, G.M.;
Adzhigirey, M.; Day, T.; Annabhimoju, R.; Sherman, W., J. Com-
put. Aid. Mol. Des., 2013, 27(3), 221–234.
[27] (a) Li, H.; Robertson, A. D.; Jensen, J. H., Proteins, 2005,
61, 704–721. (b) Bas, D. C.; Rogers, D. M.; Jensen, J. H., Proteins,
2008, 73, 765–783. (c) Olsson, M. H. M.; Søndergard, C. R.; Ros-
tkowski, M.; Jensen, J. H., J. Chem. Theory Comput., 2011, 7, 525–
537. (d) Søndergaard, C. R.; Olsson, M. H. M.; Rostkowski, M.;
Jensen, J. H., J. Chem. Theory Comput., 2011, 7, 2284–2295.
[28] Banks, J. L.; Beard, H. S.; Cao, Y.; Cho, A. E.; Damm, W.;
Farid, R.; Felts, A. K.; Halgren, T. A.; Mainz, D. T.; Maple, J. R.;
Murphy, R.; Philipp, D. M.; Repasky, M. P.; Zhang, L. Y.; Berne,
B. J.; Friesner, R. A.; Gallicchio, E.; Levy. R. M., J. Comp. Chem.
2005, 26, 1752–1780.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
[19] Early examples of non-additive mutational effects: a) Ser-
rano, L.; Horovitz, A.; Avron, B.; Bycroft, M.; Fersht, A. R. Bio-
chemistry 1990, 29, 9343–9352; b) Huang, Z.; Wagner, C. R.;
Benkovic, S. J. Biochemistry 1994, 33, 11576–11585; c) Wagner, C.
R.; Huang, Z.; Singleton, S. F.; Benkovic, S. J. Biochemistry 1995,
34, 15671–15680.
[20] Recent examples of non-additive mutational effects: a)
Reetz, M. T.; Prasad, S.; Carballeira, J. D.; Gumulya, Y.; Bocola,
M. J. Am. Chem. Soc. 2010, 132, 9144–9152; b) Zheng, H.; Reetz,
M. T. J. Am. Chem. Soc. 2010, 132, 15744–15751; c) Sandström, A.
G.; Wikmark, Y.; Engström, K.; Nyhlén, J.; Bäckvall, J.-E. Proc.
Natl. Acad. Sci. USA 2012, 109, 78–83; d) Bartsch, S.; Kourist, R.;
Bornscheuer, U. T. Angew. Chem. 2008, 120, 1531–1534; e)
Istomin, A. Y.; Gromiha, M. M.; Vorov, O. K.; Jacobs, D. J.; Live-
say, D. R. Proteins Struct. Funct. Bioinf. 2007, 70, 915–924.
[21] Hogrefe, H. H.; Cline, J.; Youngblood, G. L.; Allen, R. M.
BioTechniques 2002, 33, 1158–1165.
[22] Reetz, M. T.; Carballeira, J. D.; Peyralans, J.; Höbenreich,
H.; Maichele, A.; Vogel, A. Chem.-Eur. J. 2006, 12, 6031–6038.
ACS Paragon Plus Environment

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

Page 17 of 28
Journal of the American Chemical Society
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Scheme 1. Model Reaction Used in all Protein Engineering Experiments.
37x7mm (300 x 300 DPI)
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 18 of 28
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Scheme 2. Additional Substrates used in PAMO-catalyzed Asymmetric Sulfoxidation.
83x43mm (300 x 300 DPI)
ACS Paragon Plus Environment

Page 19 of 28
Journal of the American Chemical Society
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 1. Potential randomization sites (blue) in PAMO. The flavin and NADP cofactors are displayed in
orange and yellow, respectively. Based on the 2YLX crystal structure of PAMO.
127x101mm (300 x 300 DPI)
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 20 of 28
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 2. ISM scheme in the directed evolution of PAMO as a catalyst in the asymmetric sulfoxidation of
substrate 1b with preferential formation of (R)-2b
363x259mm (72 x 72 DPI)
ACS Paragon Plus Environment

Page 21 of 28
Journal of the American Chemical Society
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 3. The 14 possible mutants in the evolutionary protein space between WT and ZGZ-2. Lines indicate
only one trajectory of the 24 combinatorial possibilities to go from WT to ZGZ-2 (Q=a, F=b, N=c, I=d).
84x26mm (150 x 150 DPI)
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 22 of 28
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 4. Fitness pathway landscape showing the 24 pathways leading from WT PAMO (bottom) to best (R)-
selective variant ZGZ-2, a typical trajectory lacking local minima (green pathway) and one having local
minima (red) being featured.
254x142mm (72 x 72 DPI)
ACS Paragon Plus Environment

Page 23 of 28
Journal of the American Chemical Society
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 5. Fitness pathway landscape in the frontal view of Figure 4 of all 24 trajectories leading from WT
PAMO to variant ZGZ-2 characterized by four point mutations. Green notations indicate energetically favored
pathways, whereas red notations represent disfavored trajectories having local minima. Letters in red in the
dendrogram denote a local minimum after the introduction of this mutation.
420x264mm (72 x 72 DPI)
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 24 of 28
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 6. Thermodynamic cycle (Eq. 1) highlighting the interaction of point mutations and sets of mutations
involved at every stage along the energetically favored pathway a→b→c→d toward variant ZGZ-2 as a
catalyst in the sulfoxidation of substrate 1b.
257x168mm (72 x 72 DPI)
ACS Paragon Plus Environment

Page 25 of 28
Journal of the American Chemical Society
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 7. Thermodynamic cycle (Eq. 1) highlighting the interaction of point mutations and sets of mutations
involved at every stage along the energetically disfa-vored pathway d→a→b→c toward variant ZGZ-2 as a
catalyst in the sulfoxidation of substrate 1b.
299x179mm (72 x 72 DPI)
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 26 of 28
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 8. Oxidation of 1b by PAMO. Oxidation at the blue and green lone pairs will lead to formation of R-
and S-2b, respectively.
302x237mm (72 x 72 DPI)
ACS Paragon Plus Environment

Page 27 of 28
Journal of the American Chemical Society
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 9 Induced fit docking results for substrate 1b in (a) WT PAMO; (b) P440F; (c) L443I; (d)
P440F/L443I; (e) I67Q/P440F/L443I; (f) I67Q/P440F/A442N/L443I mutants. The mutated residues are
displayed in yellow. Nonꢀpolar hydrogen atoms are omitted for clarity. ꢀ∆∆G_^‡ values are given in KJ/mol
and r(OꢀS) in Å.
416x228mm (72 x 72 DPI)
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 28 of 28
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
90x47mm (300 x 300 DPI)
ACS Paragon Plus Environment