Molecular Gating of an Engineered Enzyme Captured in Real Time

Article abstract of DOI:10.1021/jacs.8b09848

Enzyme engineering tends to focus on the design of active sites for the chemical steps, while the physical steps of the catalytic cycle are often overlooked. Tight binding of a substrate in an active site is beneficial for the chemical steps, whereas good accessibility benefits substrate binding and product release. Many enzymes control the accessibility of their active sites by molecular gates. Here we analyzed the dynamics of a molecular gate artificially introduced into an access tunnel of the most efficient haloalkane dehalogenase using pre-steady-state kinetics, single-molecule fluorescence spectroscopy, and molecular dynamics. Photoinduced electron-transfer-fluorescence correlation spectroscopy (PET-FCS) has enabled real-time observation of molecular gating at the single-molecule level with rate constants (k_on = 1822 s^-1, k_off = 60 s^-1) corresponding well with those from the pre-steady-state kinetics (k_-1 = 1100 s^-1, k₁ = 20 s^-1). The PET-FCS technique is used here to study the conformational dynamics in a soluble enzyme, thus demonstrating an additional application for this method. Engineering dynamical molecular gates represents a widely applicable strategy for designing efficient biocatalysts.

Full text of DOI:10.1021/jacs.8b09848

Subscriber access provided by YORK UNIV
Article
Molecular Gating of an Engineered Enzyme Captured in Real Time
Piia Kokkonen, Jan Sykora, Zbynek Prokop, Avisek Ghose, David Bednar, Mariana Amaro, Koen
Beerens, Sarka Bidmanova, Michaela Slanska, Jan Brezovsky, Jiri Damborsky, and Martin Hof
J. Am. Chem. Soc., Just Accepted Manuscript • DOI: 10.1021/jacs.8b09848 • Publication Date (Web): 03 Dec 2018
Downloaded from http://pubs.acs.org on December 3, 2018
Just Accepted
“Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted
online prior to technical editing, formatting for publication and author proofing. The American Chemical
Society provides “Just Accepted” as a service to the research community to expedite the dissemination
of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in
full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully
peer reviewed, but should not be considered the official version of record. They are citable by the
Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore,
the “Just Accepted” Web site may not include all articles that will be published in the journal. After
a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web
site and published as an ASAP article. Note that technical editing may introduce minor changes
to the manuscript text and/or graphics which could affect content, and all legal disclaimers and
ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or
consequences arising from the use of information contained in these “Just Accepted” manuscripts.
is published by the American Chemical Society. 1155 Sixteenth Street N.W.,
Washington, DC 20036
Published by American Chemical Society. Copyright © American Chemical Society.
However, no copyright claim is made to original U.S. Government works, or works
produced by employees of any Commonwealth realm Crown government in the course
of their duties.

Page 1 of 31
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
TOC
84x45mm (300 x 300 DPI)
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 2 of 31
1
2
Molecular Gating of an Engineered Enzyme Captured in Real Time
3
4
5
6
7
Piia Kokkonen#1,2, Jan Sykora#3, Zbynek Prokop#1,2 Avisek Ghose3, David Bednar1,2, Mariana Amaro3, Koen
Beerens1, Sarka Bidmanova1,2, Michaela Slanska1, Jan Brezovsky1, Jiri Damborsky1,2*, Martin Hof3*
8
9
1
Loschmidt Laboratories, Department of Experimental Biology and Research Centre for Toxic Compounds in the
Environment RECETOX, Faculty of Science, Masaryk University, Kamenice 5/A13, 625 00 Brno, Czech Republic
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
2
International Clinical Research Center, St. Anne's University Hospital Brno, Pekarska 53, 656 91 Brno, Czech
Republic
3 J. Heyrovsky Institute of Physical Chemistry of the ASCR, v. v. i., Dolejskova 3, 182 23 Prague 8, Czech
Republic
# Equal contribution
*To whom correspondence should be addressed. E-mail: jiri@chemi.muni.cz, martin.hof@jh-inst.cas.cz
Current affiliation AG: Center for Nanobiology and Structural Biology of the Institute of Microbiology ASCR, v.v.i.,
Zámek 136, Nove Hrady 373 33, Czech Republic; JB: Laboratory of Biomolecular Interactions and Transport,
Institute of Molecular Biology and Biotechnology, Faculty of Biology, Adam Mickiewicz University, Umultowska
89, 61-614 Poznań, Poland
ABSTRACT
Enzyme engineering tends to focus on the design of active sites for the chemical steps, while the physical
steps of the catalytic cycle are often overlooked. Tight binding of a substrate in an active site is beneficial
for the chemical steps, whereas good accessibility benefits substrate binding and product release. Many
enzymes control the accessibility of their active sites by molecular gates. Here we analyzed the dynamics
of a molecular gate artificially introduced into an access tunnel of the most efficient haloalkane
dehalogenase using pre-steady-state kinetics, a single-molecule fluorescence spectroscopy and molecular
dynamics. Photoinduced electron-transfer – fluorescence correlation spectroscopy (PET-FCS) has enabled
real-time observation of molecular gating at single molecule level with the rate constants (kon = 1822 s-1,
koff = 60 s-1) corresponding well with those from the pre-steady-state kinetics (k-1 = 1100 s-1, k1 = 20 s-1).
PET-FCS technique is used here to study the conformational dynamics in a soluble enzyme, thus
demonstrating an additional application for this method. Engineering dynamical molecular gates represents
a widely applicable strategy for designing efficient biocatalysts.
1
ACS Paragon Plus Environment

Page 3 of 31
1
Journal of the American Chemical Society
INTRODUCTION
2
3
4
5
6
7
8
9
Enzymes are highly efficient biocatalysts with outstanding specificity and selectivity. Enzymatic
reactions take place in their active sites which are frequently hidden inside the protein core. The active sites
are connected to the bulk solvent through the transport tunnels. The importance of tunnels in enzymatic
activity, substrate specificity, and stereoselectivity has been explored in a number of studies using X-ray
crystallography, steady-state and pre-steady-state kinetics and molecular modeling.1,2 Beyond
complementarity between the substrate and binding site, tunnels offer another level of discrimination for
molecules going in and out of the enzyme active cavity. Tunnels discriminate the preferred substrates or
cofactors, reduce the access of solvents that may disturb the catalysis, and prevent the escape of reactive
intermediates before the reaction is finished. The orchestration of enzymatic reactions commonly requires
a precise order of several individual steps and the dynamic character of transport tunnels can play a key role
in the catalytic efficiency. Many enzymes employ molecular gates to control the access of small molecules
into their catalytic sites, in a similar fashion as ion channels control the access of ions into the cell. These
gates can adopt open and closed states and thus grant or deny the access of substrates, ions and water
molecules into the catalytic site, reflecting the requirements of the catalytic cycle.2
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
The low efficiency of de novo designed biocatalysts relative to those that have evolved naturally
demonstrates that our understanding of enzymatic catalysis is still incomplete.3 Especially, the dynamic
processes that control the function of enzymes are only recently being taken into account in the enzyme
design protocols.4,5 Great progress has been achieved in computational protein design of the
substrate/intermediate complementarity with the active site.6,7 Considerably less attention has been paid to
the role of ligand transport and the dynamics of the access pathways, even though many times the rate-
limiting step of enzymatic catalysis is not the chemical one. In the past few years, the design of access
pathways has gained popularity and as it is offering a complementary strategy to the design of the catalytic
2
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 4 of 31
1
2
sites.1–2,8–11 Rational designing of the tunnels and dynamical dates in enzymes is still challenging, however,
3
4
5
because their motions are currently insufficiently understood.
6
7
8
9
Previously we have described the most active haloalkane dehalogenase LinB86, carrying four
mutations Trp140Ala+Phe143Leu+Leu177Trp+Ile211Leu.12,13 The variant was constructed from one of the
less active variants, LinB32, which had the main access tunnel blocked by an introduced tryptophan residue
Leu177Trp.12 Even though the kinetics of LinB32 indicated improvement in a chemical step, the poor
product transport substantially hindered its catalytic activity.13 The mutation of three additional residues
(Trp140Ala+Phe143Leu+Ile211Leu) opening a novel auxilliary tunnel lead to LinB86 which has the
highest catalytic efficiency with the substrate 1,2-dibromoethane ever reported for this enzyme family. The
newly opened de novo tunnel is used for ligand transportation to a lower extent than the natural one, but the
decreased intramolecular contacts enabled the helix surrounding the main tunnel to move in a gate-like
fashion. To verify the enhanced dynamics of this region and to elucidate the mechanism of the superior
catalytic performance of LinB86, we combined the data from pre-steady-state kinetics, single-molecule
fluorescence spectroscopy and molecular modelling.
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 kinetic analyses identified the effects on the individual catalytic steps of 1,2-dibromoethane
conversion by all three enzymes: wild-type LinBwt, LinB32 and LinB86.A The detailed kinetics of enzyme-
product interactions predicted a significant role of isomerization dynamics involved in the superior catalytic
behavior of LinB86. Subsequently, photoinduced electron transfer - fluorescence correlation spectroscopy
(PET-FCS) and molecular simulations provided real-time observation and a structural view of the gating
dynamics. In this study, the PET-FCS was used to monitor the conformational dynamics of a soluble
enzyme; previously the methodology has been utilized to study transmembrane proteins or protein
folding.14,15 In the broader context of protein science, this study demonstrates how PET-FCS technique can
A In our previous work, LinB32 was referred as LinB-ClosedW and LinB86 as LinB-OpenW.13 The naming
scheme has been changed due to the new observations reported in this study which show that LinB86 adopts
both open and closed states and thus referring to it as LinB-OpenW would be confusing for the readers.
3
ACS Paragon Plus Environment

Page 5 of 31
1
2
3
4
5
6
7
Journal of the American Chemical Society
be used to better understand the impact of structural dynamics on function or selectivity of a protein. The
results led to a detailed insight into the molecular gating, which is instrumental for future de novo design of
biocatalysts with balanced chemical and physical steps of the catalytic cycle.
8
9
RESULTS
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
Pre-steady-state kinetics. The kinetic model for the reaction of LinBwt, LinB32 and LinB86 with
1,2-dibromoethane was globally fitted to pre-steady-state kinetic data including rapid quench burst, multiple
turnover stopped-flow fluorescence analyses and steady-state initial rate kinetic measurements
(Supplementary Information I). The analysis provided individual rate and equilibrium constants for the
minimal kinetic pathway including: (i) substrate binding, (ii) cleavage of carbon-halogen (SN2 reaction)
bond leading to the formation of an alkyl-enzyme intermediate, (iii) hydrolysis of the alkyl-enzyme
intermediate (and reaction) and (iv) product release (Figure 1A). In comparison to LinBwt, LinB32 displays
improvement of the first (SN2) chemical step, but decreased affinity for substrate and a significant slowdown
of product release. The LinB86 variant reconstituted the substrate binding equilibrium to the wild-type level
and simultaneously retained the improvement of the first chemical step introduced in LinB32 by the
Leu177Trp mutation. Strikingly, the LinB86 acquired additional efficiency for the second (AdN) chemical
step and extensively accelerated the transport of the product out of the active site (Supplementary Figures
1-3). The four mutations introduced to LinB86 provided effective balance among all steps involved in the
catalytic cycle with 1,2-dibromoethane. The product release remains the rate-limiting step for all three
variants (Supplementary Table 1).
4
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 6 of 31
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: Pre-steady-state kinetics. (A) The equilibrium and rate constants for individual catalytic steps
for the reaction of LinBwt (black), LinB32 (red) and LinB86 (blue) with 1,2-dibromoethane at 37oC and
pH 8.6. The scheme depicts the minimal kinetic model, where S stands for a substrate, P for a product, E
for an enzyme, ES for a non-covalent enzyme-substrate complex, EI for a covalently bound alkyl-enzyme
intermediate and EP for an enzyme-product complex. KA is equilibrium association constant for the ES
complex formation, KD is equilibrium dissociation constant (KD = 1/KA), k2 is rate constant for cleavage of
carbon-halogen bond (SN2), k3 is rate constant for hydrolysis of the alkyl-enzyme intermediate (AdN) and
k4 is rate constant for product release. The catalytic mechanism is adopted from Verschueren et al., 1993.16
(B) Schematic illustration of mutations and conformational states of LinB variants (top) and kinetics of
close-open enzyme isomerization (bottom). The bulky “W” introduced to the main access tunnel of LinB32
(red) stabilizes the closed state. For LinB86 (blue), the mutations depicted as “LAL” opened de novo
auxilliary tunnel that primarily does not maintain ligand transport, but greatly enhances the gating
dynamics. The dependence of observed rates (kobs) on ligand concentration recorded upon binding and
dissociation experiments of bromide ion with LinB32 (red) and LinB86 (blue) is plotted in the graph. The
rate of enzyme close-open isomerization representing the steps k1 and k-1 were derived from the global fit
of binding and dissociation data to the kinetic model described in the Supplementary Scheme 2.
The kinetics of bromide binding and dissociation provided information about the dynamical
behavior of individual mutants. The binding kinetics of LinB32 and LinB86 show a major difference in
comparison to the wild-type enzyme: both variants show complex kinetic behavior (Supplementary
information II), while a single step binding of bromide reaching rapid equilibrium has been described for
LinBwt.17 The decrease in observed rates (kobs) with increasing ligand concentration determined for LinB32
5
ACS Paragon Plus Environment

Page 7 of 31
1
2
3
4
5
6
7
Journal of the American Chemical Society
and LinB86 (Figure 1B) is indicative of the presence of a slow conformational step preceding the binding.
The kinetic model suggests that LinB32 and LinB86 exist in solution as a mixture of two conformers.18
Interestingly, LinB86 is a highly flexible protein and undergoes the open-close transition more than two
orders of magnitude faster than the rather rigid LinB32 (Figure 1B, Supplementary Figure 4).
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
For both proteins, the rate of forward isomerization is significantly slower than the rate for the
reverse step, favoring the closed conformation of the free enzyme. The equilibrium ratio between the two
conformational states changes in the presence of halide ions. Bromide stabilizes the open conformation,
which represents approximately 20% and 50% of the total enzyme population for LinB32 and LinB86,
respectively. Interestingly, the apparent dissociation constants (290 ± 20 mM for LinBwt, 210 ± 20 mM for
LinB32 and 180 ± 60 mM for LinB86) determined from the steady-state fluorescence show that only the
kinetics, not the overall affinity, has been altered by modification of the access tunnels (Supplementary
Figure 5).
To test the hypothesis that Leu177Trp mutation is important for the SN2 chemical step of the
catalytic cycle, we conducted a back-mutation experiment of LinB86+Trp177Leu (Supplementary
information section IV). The steady-state turnover obtained for LinB86+Trp177Leu (kcat = 3.8 ± 0.1 s-1) is
significantly decreased in comparison to the turnover number observed for LinB86 (kcat = 57 ± 3 s-1) and
even LinBwt (kcat = 12 ± 4 s-1).13 Thus, the Leu177Trp mutation blocking main tunnel is important for the
catalytic efficiency, which is further improved by the mutations in the engineered auxilliary tunnel that lead
to enhanced dynamics in LinB86.
A single-molecule PET-FCS. We have applied the concept of PET–FCS14,19–21 to observe the open-
close gating dynamics in LinB86. FCS is a single-molecule fluorescence fluctuation spectroscopy method
that provides information on protein dynamics on nanosecond to millisecond time-scales (Figure 2A). The
photochemical process of PET is used as an “on and off” switch for the fluorescence signal of a fluorophore,
thus creating the fluorescence fluctuations detected by FCS. The switch is activated by the structural
6
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 8 of 31
1
2
3
4
5
6
7
8
changes provoked by the dynamical gate, and consequently the FCS data reflects the rate of the gating
process (Figure 2BC). The prerequisite for a successful PET-FCS study is proper design of the “on and off”
switch. The chosen Atto-oxazine fluorophores Atto655 and AttoOxa12 can be quenched, i.e., turned off,
via PET from tryptophan.22,23 This enabled us to study the dynamic gating of LinB86 in relation the
introduced Trp177. A computational study has been performed to design and construct the best system for
the single molecule PET-FCS. The modelling identified suitable sites along the LinB86 sequence for the
cysteine replacement, which can be subsequently labelled with thiol reactive Atto variants (Supplementary
Information III and IV). The following parameters were evaluated for the mutable residues: (i) solvent
accessibility, (ii) backbone flexibility and (iii) stability. According to these parameters, four LinB86
mutants were constructed, purified and experimentally characterized (Asp149Cys, Gln172Cys, Leu179Cys
and Ala269Cys). The protein stability tests identified LinB92 variant (Gln172Cys) as the most suitable
candidate for the PET-FCS study. An additional variant of LinB92 with back-mutated Trp177 and denoted
as LinB99 (Gln172Cys+Trp177Leu) was prepared for control experiments to exclude non-specific
contributions to the PET-FCS data. Two thiol reactive dyes Atto655 (Atto 655-Maleimide) and AttoOxa12
(Atto Oxa12-Maleimide) were selected for labelling of both LinB92 and LinB99 (Supplementary Table 4).
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
7
ACS Paragon Plus Environment

Page 9 of 31
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 2: Single molecule spectroscopy. (A) Scheme depicting the principle of single molecule FCS
experiment. The fluorescence of freely diffusing molecule (orange) is monitored during its transit through
the observation volume (in green). (B) Schematic illustration of the dynamic behavior of LinB92-Atto655,
i.e., LinB86 labelled with Atto dye at the position 172, which can adopt open and closed conformations.
When labelled with Atto655 or AttoOxa12, the closed conformation of LinB92-Atto655 maintains the
fluorescence dye away from Trp177 (depicted as W) preventing the PET. Thus, the closed state of LinB92-
Atto655 is fluorescent. In contrast, the open state of LinB92-Atto655 enables the Atto dye to get into close
proximity of Trp177, resulting in PET and quenching of the dye. The open state is therefore non-fluorescent.
The respective rate constants determined by the single molecule experiments are depicted. (C) Simplified
cartoon of time trace of the fluorescence intensity fluctuations recorded in FCS experiment (raw data trace
is depicted as Supplementary Figure 8 in the Supporting information). Transition to the open state leads to
a reduction in the fluorescence intensity (in green), while the opposite transition recovers the fluorescence
intensity (in red). These fluctuations caused by the conformational transition are projected in the recorded
autocorrelation function G(τ), which enables the determination of the rate constants of the transitions
between the closed and open states. (D) FCS autocorrelation curves G(τ) obtained for LinB92-Atto655
(black) and LinB99-Atto655 (red). LinB99 carries out back mutation (Trp177Leu) and serves as the control
sample to exclude non-specific contributions to the PET-FCS data. The inset in linear scale highlights the
8
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 10 of 31
1
2
3
4
observed difference in the PET component between LinB92-Atto655 and LinB99-Atto655 on the (sub)-
millisecond time scale. (E) PET relaxation time constant τ1 obtained for LinB92-Atto655 (black) and
LinB99-Atto655 (red) at different bromide concentrations.
5
6
7
8
9
The FCS autocorrelation functions G(τ) obtained for the enzymes labelled with Atto655 (Figure 2D)
show a complex behavior with a dominant component on the millisecond time scale that is attributed to the
diffusion of the protein. For LinB92-Atto655 an additional component, τ1, is detected on the (sub)-
millisecond time scale, while the control variant LinB99-Atto655 lacks this contribution to G(τ) (Figures
2D and 2E, Supplementary Table 5). Therefore, it is rational to causally link the τ1 component in LinB92-
Atto655 to the transition between the fluorescent and the non-fluorescent states (induced by PET due to the
conformational change). Single-molecule PET-FCS experiments carried out in solutions of the labelled
LinB92-Atto655 and LinB99-Atto655 at various bromide concentrations (Figure 2E, Supplementary Table
5) confirmed the conclusion drawn from the pre-steady-state kinetic data: the presence of bromide ions
changes the kinetics of transition between the open and closed states of LinB86. For low bromide
concentrations a slowing down of τ1 is apparent, however, at bromide concentrations higher than 50 mM,
τ1 is apparently reduced until the difference between LinB92-Atto655 and its control variant LinB99-
Atto655 is no longer visible (Figure 2E). Careful inspection of the autocorrelation functions and kinetic
data indicates that the conformational dynamics is shifted to longer millisecond timescales at elevated
bromide concentrations. This overlays the component τ1 with the diffusion part of G(τ), τD, leading to an
apparent speeding up of τ1. Analogous behavior is also observed for LinB92-AttoOxa12 and LinB99-
AttoOxa12 (Supplementary Table 6). The reproducibility with a different dye supports the conclusion that
the τ1 component of the FCS data reflects the “on and off” switch of fluorescence caused by the PET process,
which itself is brought upon by the conformational change of the enzyme during the molecular gating.
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
Molecular dynamics simulations. The molecular dynamics simulations of LinB86 and LinB92-
Atto655 revealed that the gating mechanism includes a large movement of one of the cap domain helices
9
ACS Paragon Plus Environment

Page 11 of 31
1
Journal of the American Chemical Society
(Figure 3). Since both LinB86 and LinB92-Atto655 displayed very similar opening in the simulations, we
conclude that the Atto655 dye does not disturb the dynamics of the gating process (Supplementary Table
9). In the open conformation, the Atto655 probe binds to the mouth of the access tunnel leading to the active
site (Figure 3). Consequently, the fluorophore of the Atto655 probe is in a close vicinity of Trp177
promoting the PET process. This open conformation corresponds to the non-fluorescent state. In contrast,
the closed state of LinB92-Atto655 reduces the interaction possibilities for Atto655 and Trp177, which
hinders PET and leads to the fluorescent state of the probe. Thus, a two-state model can be applied for the
equilibrium between the two distinct conformations of open non-fluorescent and closed fluorescent protein
states (Figure 2B). The opening of the enzyme increases the main access tunnel bottleneck from 1.3 Å to
1.9 Å, and the passage of the ligands through the open state displays lower energy barriers (Supporting
Information Table 9 and Section VII).
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
10
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 12 of 31
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: Molecular dynamics simulations. Individual snapshots of the two states observed for the
LinB92-Atto655 complex in the molecular dynamics simulations. (A) The fluorescent probe Atto655 (cyan)
is attached to the residue Cys172 and in the closed conformation it cannot reach the Trp177 (blue). (B) In
the open conformation the fluorescence of the Atto655 probe is quenched by the interaction with tunnel-
blocking Trp177 (blue). A flexible helix of the cap domain (in red) represents the gating element, which is
undertaking periodical conformational transitions leading to open and closed states. The conformations are
defined by the distance of the Cα atoms of residues Trp177 and Asp147 (start of the red helix), which is the
metrics used in the Markov state model construction. The closed state resembles the crystal structure of
LinB86 (PDB ID 5LKA) and is the predominant conformation in the simulations with the equilibrium
probability >98% in both the LinB92-Atto655 and LinB86 simulations (Supplementary Table 9).
Application of the two-state model from the molecular dynamics simulations to the PET-FCS results
allowed the calculation of the koff and kon rate constants (Table 1), which coincided well with the k-1 and k1
rate constants determined by kinetic measurements (Supplementary Table 2). These experimental results
show that the fluorescent closed state is preferred over the non-fluorescent open one. This corresponds also
to the molecular dynamic simulations where the equilibrium probability of the open state was 1.2 ± 0.5%
11
ACS Paragon Plus Environment

Page 13 of 31
1
Journal of the American Chemical Society
in the LinB92-Atto655 simulations and 2.2 ± 0.1% in the LinB86 simulations (Supplementary Table 9).
The kon and koff values obtained from the PET-FCS experiments are dependent on bromide concentrations
in good correspondence with the pre-steady-state kinetic data (Supplementary Table 2). Taken together,
these data confirm dynamics of a molecular gate is not significantly disturbed by the attachment of the PET
probe and that the molecular gating is an essential part of the catalytic cycle of LinB86.
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
Table 1. Estimated kon and koff rate constants characterizing the transition between the non-fluorescent and
fluorescent state of LinB92-Atto655. The values were calculated from parameters α1 and τ1 (Supplementary
Table 5) assuming a two-state model. The data for AttoOxa12 are summarized in Supplementary Table 7.
kon (s-1)
1822±488
899±283
≤ 600*
koff (s-1)
60±31
28±16
≤ 20*
LinB92-Atto655
LinB92-Atto655 + 50 mM Br-
LinB92-Atto655 + ≥ 100 mM Br-
*The lower threshold of the rate constants was qualitatively estimated by assuming the PET component is
overlapping with the diffusion, i.e., being on the order of 2 ms.
DISCUSSION
The modifications of the LinB variants presented in this study are in the access tunnels of the enzyme
and they affect the transport of the ligands. The kinetic results revealed that LinB32 and LinB86 have a
more complex dynamical profile than LinBwt. Leu177Trp mutation in LinB32 causes a tighter packing of
the tunnels that reduces its dynamics while enhancing the chemical step of the catalysis. On the other hand,
the additional mutations Trp140Ala+Phe143Leu+Ile211Leu in LinB86 lowered interactions in the cap
domain of the enzyme. This engineering step re-established the dynamical gate-like movement of the cap
domain helix that controls the transport of substrates, products and solvent molecules. Such molecular
gating elements and large structural conformational changes are common among many proteins.24–26
The rate constants that characterize the equilibrium between the two conformational states of
LinB86 agree for all the three different methodologies applied. Closed state of the enzyme LinB86 is the
predominant one. Both the pre-steady-state kinetics and the PET-FCS experiments displayed slowing of the
12
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 14 of 31
1
2
3
4
5
6
7
enzyme’s dynamical behavior at elevated concentrations of bromide ions. The obtained data indicates that
the gating dynamics is affected significantly by the concentration of this reaction product and confirms that
both methods are indeed looking at the same phenomenon.
8
9
Single-molecule experiments have been previously conducted to study the conformations of
transmembrane proteins and chaperones, and to observe protein folding and denaturation pathways.19,20,27–
34 Here we have shown that the single-molecule PET-FCS is also an excellent method for studying gating
and ligand transport in enzymes. Using protein engineering, we have tuned the dynamics of the main tunnel
(occurring in micro- to millisecond time scale) which turned out to be crucial for the ligand transport in
LinB32 and LinB86. Would it be possible to modify the enzyme structure so that the open state is more
predominant in order to further increase accessibility of buried active site, as the product release is still the
rate-limiting step? There is a good chance that increasing the dynamics of the protein would increase the
activity even further, however, it is known that promoting flexibility of enzymes generally leads to problems
with their stability.35
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
We have increased the flexibility of the cap domain from LinB32 to LinB86 by mutating larger
amino acid residues into smaller ones with the intention of creating a de novo access tunnel.7 Based on the
results reported in our previous study, the newly opened auxilliary tunnel is used in ligand transport in
LinB86, even though the main tunnel is still the preferred transport pathway.13 Here we show that the
unintentional facilitation of the cap domain movements and the enhanced dynamics of the main tunnel
explain the increased enzymatic efficiency more than auxiliary tunnel, even though this new tunnel is
functional and used by the ligands. Excellent agreement of kinetic data and PET-FCS experiments in the
transitions between the open and closed states let us believe that modified conformational dynamics is the
mechanism for improved catalytic efficiency of LinB86. Back mutation introduced to LinB86+Trp177Leu
protein resulted in a significant drop in activity, suggesting that introduced gating element is composed of
both Trp177 and a flexible helix of the cap domain. This gating element is undertaking periodic open-closed
13
ACS Paragon Plus Environment

Page 15 of 31
1
2
3
4
5
6
7
Journal of the American Chemical Society
transitions, thus balancing physical and chemical steps of the catalytic cycle. While the current trend in
enzyme engineering is to design the best environment for stabilization for the transition state, the access
tunnels and their gating elements represent a natural hotspots for mutagenesis, especially in enzymes with
buried active sites and the rate-limitation in the physical steps of the catalytic cycle.1,2,36
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
CONCLUSIONS
We have recently demonstrated that introduction of a de novo tunnel in the enzyme with buried
active site can provide a biocatalyst with superior catalytic performance.7 The main objective of this study
was to carry out in-depth analysis of the effect of mutations on functional properties and protein dynamics
using several experimental and theoretical techniques. The introduction of a bulky tryptophan residue in
the access tunnel of LinB lead to the variant LinB32, which displayed blocked access tunnel and slower
dynamical movements of the cap domain. This engineering step had an adverse effect on the product release,
while it at the same time increased the rates of the chemical steps with the model substrate 1,2-
dibromoethane. The mutation of three additional residues in de novo tunnel resulted in LinB86, which is
the most catalytically efficient haloalkane dehalogenase variant ever. This variant benefited both from the
increased rate of the chemicals step, as LinB32, and the re-established dynamical movements of the cap
domain that allows better transportation of substrates and products. Here, we have captured the dynamical
reasons for the superior activity of LinB86 using single-molecule PET-FCS in combination with pre-steady-
state kinetics and molecular dynamics simulations. We have revealed the molecular mechanisms of the
excellent catalytic behavior of LinB86 and the usability of the PET-FCS methodology for the determination
of dynamical movements of proteins. Moreover, we demonstrate for the first time that PET-FCS is a
powerful tool for monitoring the conformational dynamics of a soluble enzyme. Since the dynamical cap
domain is one of the determining factors for the activity and specificity shared within the haloalkane
dehalogenase enzyme family,37,38 we believe that presented data can initiate a future single-molecule studies
14
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 16 of 31
1
2
3
4
5
6
7
8
deciphering the role of dynamics to enzymatic catalysis. Moreover, our results illustrate the effective use
of the combination of methodologies to study protein dynamical behavior and offer a solid base for studying
protein structures with engineered dynamics. Engineering the gating mechanisms in enzymes with buried
active sites opens new possibilities in protein engineering outside perfecting the active site
complementarity.
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
METHODS
Rapid-quench analysis: Rapid quench flow experiments were performed at 37°C in a 100 mM
glycine buffer pH 8.6 using the rapid quench flow instrument model QFM 400 (BioLogic, France). The
reaction was started by rapid mixing of 75 µl enzyme with 75 µl substrate solution and quenched with 100
µl 0.8 M H2SO4 after time intervals ranging from 5 ms to 1 s. The quenched mixture was directly injected
into 0.5 ml of ice-cold diethyl ether with 1,2-dichloroethane as the internal standard. After extraction, the
diethyl ether layer containing non-covalently bound substrate and alcohol product was collected, dried on
a short column containing anhydrous Na2SO4 and analyzed on gas chromatograph Agilent 7890 (Agilent,
USA) equipped with capillary column DB-FFAP (30m x 0.25mm x 0.25μm, Phenomenex) and connected
with mass spectrometer Agilent 5975C (Agilent, USA). The amount of halide in the water phase was
measured by ion chromatograph 861 Advanced Compact IC equipped with METROSEP A Supp 5 column
(Metrohm, Switzerland).
Stopped-flow analysis: The experiments were performed by using the stopped-flow instrument
SFM-300 (BioLogic, France) combined with MOS-200 spectrometer equipped with a Xe arc lamp. The
dead time of the instrument was 0.4 ms. Fluorescence emission from tryptophan residues was observed
through a 320 nm cut-off filter upon excitation at 295 nm. All reactions were performed at 37°C in a 100
mM glycine buffer of pH 8.6. Multiple turnover fluorescence traces were recorded upon rapid mixing of an
15
ACS Paragon Plus Environment

Page 17 of 31
1
Journal of the American Chemical Society
enzyme with 0-4 mM 1,2-dibromoethane, each trace shows the average of seven individual experiments.
Bromide binding was measured upon rapid mixing of enzyme with bromide to a final concentration of 0-1
M. The bromide dissociation was analyzed under rapid dilution of enzyme pre-incubated with 50-300 mM
bromide and consequently mixed 1:1 with pure buffer. To detect the rapid equilibrium phase during bromide
dissociation, enzyme pre-incubated with 150 mM bromide was rapidly diluted by mixing 1:1 with a buffer
containing 0, 50, 100 and 150 mM bromide.
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
Steady-state kinetics: The initial rates of reaction of LinB variants with 1,2-dibromoethane were
measured using a VP-ITC isothermal titration microcalorimeter (MicroCal, USA) at 37°C in a glycine
buffer pH 8.6. The micro-calorimeter reaction mixture vessel was filled with 1.4 mL of enzyme solution at
a concentration of 0.004-0.01 mg/mL. The substrate solution was titrated at 150 s intervals in the reaction
mixture. Each injection increased the substrate concentration until the reaction rate became saturated. A
total of 28 injections were carried out during the titration. The reaction rates reached after every injection
(in units of thermal power) were converted to enzyme activity. The substrate concentration was verified by
the gas chromatography (Finnigan, USA).
Data analysis and statistics: The kinetic data were fitted using dynamic kinetic simulation program
KinTek Global Kinetic Explorer version 6.0 (KinTek, USA). Data fitting used numerical integration of rate
equations from an input model searching a set of parameters that produce a minimum χ2 value using
nonlinear regression based on the Levenberg-Marquardt method.39,40 The residuals were normalized by
sigma value for each data point. The thermodynamic cycle was enforced to maintain a net free energy
change of zero when fitting data. The standard error was calculated from the covariance matrix during
nonlinear regression in globally fitting all fluorescence traces for each binding and dissociation
experiment.41 Detailed information for particular kinetic data analysis and solving specific kinetic models
are described in Supplementary Information.
16
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 18 of 31
1
2
3
4
5
6
7
8
Construction of the mutants for the labelling: Expression of the different LinB variants was achieved
using E. coli BL21(DE3) containing the created expression vectors. LB medium with ampicillin (1 l) was
inoculated with 5 ml freshly grown pre-culture. The bacteria were grown at 37 °C, till optical density
measured at 600 nm reached approximately 0.6. Then, the cells were cooled below 20 °C and 0.5 mM IPTG
was added and growth continued at 20 °C. After overnight growth, cells were harvested using centrifugation
for 10 min at 8000 rpm. A washing step was carried out using approximately 125 ml Wash Buffer. Cells
were harvested by centrifugation (10 min at 8000 rpm). Cell pellets were dissolved in approximately 40 ml
of Purification Buffer A and then frozen at -80 °C. Disruption was achieved by sonication (4*8 min, 0.3
cycle, 85 amplitude). Crude extract was centrifuged (20-60 min at 14000 rpm) to remove cell debris. Cell
free extracts were filtered through a Rotilabo-Spritzenfilter (0.45 μm) to remove remainders of cell debris.
Samples were purified using metal affinity chromatography with a small adaptation of step wise increased
imidazole concentration for washing (10 ml 4%, 10 ml 6% and 15 ml 10%). Finally, imidazole was removed
by a double dialysis step in 50 mM phosphate buffer, pH 7.5. Protein samples were filtered through a sterile
0.22 μm Rotilabo-Spritzenfilter for storage. Concentration of prepared LinB variants was determined in
triplicates using Bradford reagent and the purity was checked by SDS-PAGE.
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
Enzyme labelling: Labelling of enzymes (LinB92 and LinB99) with Atto-655-Maleimide and
AttoOxa12-Maleimide (Atto-Tec GmbH, Germany) was performed in native buffer (50 mM potassium
phosphate buffer, pH 7.4 at 20°C). The protein solution (0.2 mg/ml) was mixed with cysteine-reactive Atto
analogues added from their lyophilized stock at molar ratio of 1:1. After incubation at 10oC for 1 h, the
unreacted label was removed by the desalting column (Zeba Spin Column, Fisher Scientific). The collected
fraction was then further dialyzed overnight against 50 mM phosphate buffer at 6°C (MWCO 14 kDa) to
remove traces of the free dye. In parallel, for FCS measurement, labelled enzyme was further diluted,
o
generally 10000x, in phosphate buffer and dialyzed against 100 mM Glycine buffer (pH 8.6 at 20 C)
overnight at 6°C (MWCO 14 kDa).42
17
ACS Paragon Plus Environment

Page 19 of 31
Journal of the American Chemical Society
1
2
3
4
5
6
7
8
9
Kinetics of the labelled enzymes: The activity of LinB variants was tested with 1,2-dibromoethane
at 37°C. The activity was examined in 10 ml 100 mM glycine buffer, pH 8.6 with 10 μl of substrate in 25-
ml Reaction Flasks closed by Mininert Valves. The reactions were initiated by addition of 100 μl enzyme
with concentration of approximately 0.2 mg/ml. Samples (1 ml) were taken at 3-5 min time intervals until
20 min and immediately mixed with 35% nitric acid to stop the enzymatic reaction. Increasing concentration
of reaction product was measured spectrofotometrically at 460 nm using the Iwasaki protocol with mercuric
thiocyanate and ferric ammonium sulphate using a microplate reader Sunrise (Tecan, Switzerland).43
Dehalogenation activities were quantified by the slope of the relationship between the product concentration
and the time.
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
UV-VIS absorption and steady-state fluorescence measurement: Absorption spectra were recorded
on Shimadzu UV-2600 spectrophotometer at the speed of 150 nm/min. Steady state fluorescence spectra
were measured on Jobin-Yvon Fuorolog 3 spectrometer at the speed 100 nm/min with the emission and
excitation bandwidth set equally to 2 nm. Excitation wavelength coincided with FCS experiment (635 nm).
Signal was corrected for background noise. Fluorescence steady state anisotropy was measured with
inserted polarizers. Four intensities of polarized fluorescence (IVV, IVH, IHV, IHH) were recorded, where "V"
and "H" in the subscripts represent the vertical and horizontal directions, and the first and second subscript
denotes the direction plane of the polarization in the excitation and emission arm, respectively. The steady-
state anisotropy is calculated as follows:
퐼푉푉 ― 퐺퐼푉퐻
(1)
푟푠푡 =
퐼푉푉 + 2퐺퐼푉퐻
in which G is the instrumental correction factor given by the observed ratio, IHV/IHH.
Far-UV Circular dichroism (CD) spectra: Far-UV CD spectra of the labelled and non labelled
enzymes were measured in native buffer in 0.2 mm quartz cuvette at 10°C on Jasco 1500 CD
spectrophotometer. Concentration of enzyme was adjusted to avoid pile up effect on the detector. Measured
18
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 20 of 31
1
2
CD data were expressed as mean residue elipticity [Θ]MR bearing the units of deg.cm2.dmol-1 according to
3
4
equation (2):
5
6
7
8
100훩
(2)
[훩]푀푅
=
푙푐푛
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
where Θ is acquired ellipticity (mdeg), l is cuvette path length (mm), c is molar protein concentration, and
n is the number of peptide bonds in the protein.44
FCS setup and data analysis: FCS data acquisitions were performed in glass-bottom Lab-Tek
chambers (ThermoFisher). To minimize adsorption of proteins on glass surface, chambers were incubated
with 0.01% poly-L-lysine solution (15-20 min) followed by washing 20 times with deionized water. Prior
to measurements, labelled protein was further diluted to 1 nM concentration in 100 mM glycine buffer, pH
= 8.6. All experiments were carried out at 20°C. FCS measurements were performed on a standard inverse
fluorescence microscope (Olympus IX71, combined with Microtime 200, PicoQuant, Germany) equipped
with a diode laser as the continuous excitation source (635 nm, LDH-D-C635 PicoQuant,Germany).
Collimated laser beam was guided into a water-immersion 60x objective (NA 1.2, Olympus, Japan) by a
dichroic mirror (z635, Chroma filters, USA). The back aperture of the objective was underfilled to achieve
enlarged focal volume (its FWHM approximately 8 times larger than the Airy disk). The average laser
power was adjusted to be 185 μW leaving the objective plane. The fluorescence signal was collected by the
same objective passing through the dichroic mirror onto a 150 μm pinhole. The emission signal was further
guided through a band-pass filter (Semrock, FF01-697/58-25, USA), separated by 60/40 beam splitter plate
and imaged onto the active area of two avalanche photodiodes (Perkin-Elmer APD, SPCM-AQRH,
Canada), which were synchronized to circumvent the dead-time of the electronics and after-pulsing effects.
The signal was recorded in the pt2 mode using a Picoharp 300 device (PicoQuant, Germany).45 The data
were analyzed using PicoQuant Symphotime software yielding the cross-correlation curves G(τ), which
19
ACS Paragon Plus Environment

Page 21 of 31
1
Journal of the American Chemical Society
were further fitted, using in-house written Origin and Matlab scripts, with equation (3) describing both
2
3
4
diffusion and PET components:
5
6
7
8
9
―1
휏
휏
∑
1 + 훼 exp ( ― )
푛
휏푛
푛
(3)
퐺(휏) = (1 + )
(
)
(
)
휏퐷
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
where, ρ and τD represent the diffusion time and corresponding amplitude; and αn and τn stand for the PET
relaxation amplitudes and corresponding PET time constants. The data were further evaluated assuming a
two-state transition between fluorescent and non-fluorescent conformational states. Validity of the
assumption was supported by molecular dynamics simulations. Rate constants kon and koff were calculated
according to:
푘표푓푓
(4)
(5)
훼1 =
휏1 =
푘표푛
1
푘표푓푓 + 푘표푛
where, kon corresponds to the transition from the non-fluorescent form to the fluorescent one, while rate
constant koff describes the opposite process.
Computational design of a system for single molecule analysis: Sequences of six experimentally
characterized haloalkane dehalogenases DhaA, LinB, DrbA, DmbC, DhlA and DmbB, were used as queries
for PSI-BLAST searches against nr database of NCBI (for details see Supplementary Information IV).46,47
A multiple sequence alignment was constructed and used for estimation of the level of conservation at
individual sites within the haloalkane dehalogenase HLD-II subfamily. Evolutionary variable spots for Cys
replacement were visually inspected and their solvent accessibility, a fluctuation quantified by B-factors
and the distance to Trp177 were assessed.
Computational parameters for the Cys-Atto-655 complex: Using Avogadro 1.0.3 program,48 the
structure of the Cys-Atto655 residue (Supplementary Figure 9) was prepared in the extended α-helical
conformation of backbone atoms, and the eight lowest energy conformations of PET probe were identified
20
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 22 of 31
1
2
3
4
5
6
7
8
for each backbone conformation with weighted rotor search keeping the backbone atoms fixed. Next, the
Cys-Atto655 residue was capped with N-methylamide (NME) and acetyl (ACE) residues and the geometry
of all 16 selected conformations were energy minimized with b3lyp/6-31G* using Gaussian09 program
revision D.01.49 The two lowest energy conformations for each backbone conformations were used for
fitting the partial atomic charges of the novel Cys-Atto655 residue using RESP ESP charge derive (R.E.D.)
server 2.050,51 with HF/6-31G* level of theory and Gaussian09 program included in the R.E.D. server. The
charges on the Cys-Atto655 residue were derived using RESP-A1A charge model and multi-conformation
multi-orientation RESP fit.52 During the fitting procedure, the charges on the capping NME and ACE
residues were constrained to zero and the charges on the four atoms of peptide bond (Atom IDs 1, 2, 4 and
5) were constrained to the values corresponding to the respective atoms of the electro-neutral residues from
the Cornell et al. force field.53 The atom types were derived by analogy with the Cornell et al. force field
with the following exceptions: (i) the parameters for the aromatic oxygen (atom ID 84) were obtained from
the study of VanBeek et al. 54, (ii) the parameters for S6 atom type from the GAFF force field55 were used
for the sulfonate sulfur (Atom ID 50), and (iii) the missing parameters for the sp2 nitrogens (Atom IDs 20,
32, 58 and 72) were generated by parmchk2 program form AMBERTools 16 suite.56 The assigned atom
types and derived partial charges of Cys-Atto655 residue are summarized in Supplementary 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
Molecular dynamics simulations: The crystal structure of LinB86 (PDB ID 5LKA) was downloaded
from RSCB Protein Data Bank.13 The hydrogen atoms were added to the structure with the H++ web
server.46 The water molecules from the crystal structure, which did not overlap with the protonated
structure, were retained. The Cys-Atto655 residue was manually inserted to the position 172 using PyMol
so that the fluorophore was not interacting with any protein atoms at the start of the simulations. The
resulting structure was solvated using the tLeap module of AmberTools 14 in a cubic water box of TIP3P
water molecules so that all protein atoms were at least 10 Å from the surface of the box.56,57 Cl- and Na+
ions were added to neutralize the charge of the protein and to get a final concentration of 0.1M. The system
21
ACS Paragon Plus Environment

Page 23 of 31
1
Journal of the American Chemical Society
was equilibrated using the Equilibration_v1 module of High Throughput Molecular Dynamics (HTMD)
and the simulations were carried out in AceMD. The system was first minimized using conjugate-gradient
method for 500 steps. Then the system was heated and minimized as follows: (i) 500 steps (2 ps) of NVT
thermalization to 310 K with the Berendsen barostat with constraints on all heavy atoms of the protein, (ii)
625000 steps (2.5 ns) of NPT equilibration with Langevin thermostat with 1 kcal*mol-1*Å-2 constraints on
all heavy atoms of the protein and (iii) 625000 steps (2.5 ns) of NPT equilibration with the Langevin
thermostat without constraints. Holonomic constraints were set on all hydrogen-heavy atom bond terms and
the mass of hydrogen atoms was scaled with factor 4, enabling the 4 fs timestep.58–61 The simulations at
310 K employed periodic boundary conditions, using the particle mesh Ewald method for treatment of
interactions beyond 9 Å cut-off, electrostatic interactions suppressed >4 bond terms away from each other
and the smoothing and switching functions of van der Waals and electrostatic interactions were enabled to
avoid drastic changes at the cut-off from 7.5 Å to 9 Å. HTMD with AceMD were used to perform adaptive
sampling of the LinB92 system.62 The 20 ns production runs were started with the files resulting from the
equilibration and they employed the same settings as the last step of the equilibration. Adaptive sampling
was run first using the distance of the PET probe and the Trp177 as the reaction coordinate, then sampling
the χ2 dihedral angle of the Trp177, the binary contact map between Atto655 probe and Trp177 atoms and
finally the RMSD of the backbone Cα atoms. The total simulation time used in calculations was 8720 ns
saved with 0.1 ns intervals. The distance between Trp177 and Asp147 was used as the metric for the Markov
model construction. The data was clustered using MiniBatchKmeans algorithm to 200 clusters. The implied
timescale plots of Markov state model at various lag times were constructed and a lag time of 10 ns was
selected for the two-state Markov model construction (Supplementary Figure 10). To estimate the errors in
the data, 1000 individual 2-state Markov models were constructed with random 50% bootstrapped data
using the previous parameters. Simulations and analysis with similar parameters were calculated for LinB86
without the Atto-655 probe to ensure that observed protein conformational change is not caused by the
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
attachment of PET probe. These simulations were started from LinB86 crystal structure (PDB ID 5LKA),13
22
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 24 of 31
1
2
3
4
5
6
7
8
followed the previously described equilibration schemes and the adaptive sampling was conducted on the
RMSD of the backbone Cα atoms. Total simulation time was 3920 ns and the construction of the Markov
state models and bootstrapping were performed similarly to those described for LinB92-Atto655 complex
(Supplementary Table 9).
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
ACKNOWLEDGEMENTS
The authors thank Gaspar Pinto and Ondrej Vavra (Masaryk University) for conducting CaverDock
calculations. The authors further thank the Grant Agency of the Czech Republic (GA16-07965S, and GA16-
06096S), the Czech Ministry of Education (LQ1605, LM2015051, CZ.02.1.01/0.0/0.0/16_013/0001761,
LM2015047, and LM2015055) and the European Union (720776). Computational resources were supplied
by the Ministry of Education, Youth and Sports of the Czech Republic under the Projects CESNET (Project
No. LM2015042) and CERIT-Scientific Cloud (Project No. LM2015085).
SUPPORTING INFORMATION
Details of pre-steady-state kinetics of 1,2-dibromoethane conversion, pre-steady state kinetics of bromide
binding and dissociation, stopped-flow fluorescence analysis of 2-bromoethanol conversion, steady-state
kinetics of LinB86+W177L, single molecule PET-FCS results of enzyme dynamics and computational
design, molecular dynamics simulations and ligand binding simulations.
REFERENCES
(1) Prokop, Z.; Gora, A.; Brezovsky, J.; Chaloupkova, R.; Stepankova, V.; Damborsky, J. Engineering
of Protein Tunnels: Keyhole-Lock-Key Model for Catalysis by the Enzymes with Buried Active
Sites. In Protein Engineering Handbook, Vol. 3.; Weinheim: Wiley-VCH, 2012; pp 421–464.
(2) Gora, A.; Brezovsky, J.; Damborsky, J. Gates of Enzymes. Chem. Rev. 2013, 113 (8), 5871–5923.
(3) Huang, P.-S.; Boyken, S. E.; Baker, D. The Coming of Age of de Novo Protein Design. Nature
2016, 537 (7620), 320–327.
(4) Maria-Solano, M. A.; Serrano-Hervás, E.; Romero-Rivera, A.; Iglesias-Fernández, J.; Osuna, S.
Role of Conformational Dynamics in the Evolution of Novel Enzyme Function. Chem. Commun..
2018, 54 (50), 6622–6634.
23
ACS Paragon Plus Environment

Page 25 of 31
1
2
3
Journal of the American Chemical Society
(5) Campbell, E. C.; Correy, G. J.; Mabbitt, P. D.; Buckle, A. M.; Tokuriki, N.; Jackson, C. J.
Laboratory Evolution of Protein Conformational Dynamics. Curr. Opin. Struct. Biol. 2018, 50, 49–
57.
4
5
6
7
8
(6) Chen, Z.; Zeng, A.-P. Protein Engineering Approaches to Chemical Biotechnology. Curr. Opin.
Biotechnol. 2016, 42, 198–205.
(7) Swint-Kruse, L. Using Evolution to Guide Protein Engineering: The Devil Is in the Details.
Biophys. J. 2016, 111 (1), 10–18.
9
(8) Romero‐Rivera, A.; Iglesias‐Fernández, J.; Osuna, S. Exploring the Conversion of a D-Sialic Acid
Aldolase into a L-KDO Aldolase. Eur. J. Org. Chem. 2018, 2018 (20–21), 2603–2608.
(9) Liskova, V.; Bednar, D.; Prudnikova, T.; Rezacova, P.; Koudelakova, T.; Sebestova, E.;
Smatanova, I. K.; Brezovsky, J.; Chaloupkova, R.; Damborsky, J. Balancing the Stability-Activity
Trade-Off by Fine-Tuning Dehalogenase Access Tunnels. ChemCatChem 2015, 7 (4), 648–659.
(10) Li, G.; Yao, P.; Gong, R.; Li, J.; Liu, P.; Lonsdale, R.; Wu, Q.; Lin, J.; Zhu, D.; Reetz, M. T.
Simultaneous Engineering of an Enzyme’s Entrance Tunnel and Active Site: The Case of
Monoamine Oxidase MAO-N. Chem. Sci. 2017, 8 (5), 4093–4099.
(11) Yan, X.; Wang, J.; Sun, Y.; Zhu, J.; Wu, S. Facilitating the Evolution of Esterase Activity from a
Promiscuous Enzyme (Mhg) with Catalytic Functions of Amide Hydrolysis and Carboxylic Acid
Perhydrolysis by Engineering the Substrate Entrance Tunnel. Appl. Environ. Microbiol. 2016, 82
(22), 6748–6756.
(12) Biedermannová, L.; Prokop, Z.; Gora, A.; Chovancová, E.; Kovács, M.; Damborsky, J.; Wade, R.
C. A Single Mutation in a Tunnel to the Active Site Changes the Mechanism and Kinetics of
Product Release in Haloalkane Dehalogenase LinB. J. Biol. Chem. 2012, 287 (34), 29062–29074.
(13) Brezovsky, J.; Babkova, P.; Degtjarik, O.; Fortova, A.; Gora, A.; Iermak, I.; Rezacova, P.; Dvorak,
P.; Smatanova, I. K.; Prokop, Z., Chaloupkova, R.; Damborsky, J. Engineering a de Novo Transport
Tunnel. ACS Catal. 2016, 6 (11), 7597–7610.
(14) Sauer, M.; Neuweiler, H. PET-FCS: Probing Rapid Structural Fluctuations of Proteins and Nucleic
Acids by Single-Molecule Fluorescence Quenching. In Fluorescence Spectroscopy and
Microscopy; Methods in Molecular Biology; Humana Press, Totowa, NJ, 2014; pp 597–615.
(15) Li, X.; Xing, J.; Qiu, Z.; He, Q.; Lin, J. Quantification of Membrane Protein Dynamics and
Interactions in Plant Cells by Fluorescence Correlation Spectroscopy. Mol. Plant 2016, 9 (9),
1229–1239.
(16) Verschueren, K. H.; Seljée, F.; Rozeboom, H. J.; Kalk, K. H.; Dijkstra, B. W. Crystallographic
Analysis of the Catalytic Mechanism of Haloalkane Dehalogenase. Nature 1993, 363 (6431), 693–
698.
(17) Prokop, Z.; Monincová, M.; Chaloupková, R.; Klvaňa, M.; Nagata, Y.; Janssen, D. B.; Damborský,
J. Catalytic Mechanism of the Haloalkane Dehalogenase LinB from Sphingomonas Paucimobilis
UT26. J. Biol. Chem. 2003, 278 (46), 45094–45100.
(18) Fersht, A. Measurement and Magnitude of Individual Rate Constants. In Structure and mechanism
in protein science; W. H. Freeman and Co.: New York, NY, 1998; pp 147–149.
(19) Doose, S.; Neuweiler, H.; Sauer, M. Fluorescence Quenching by Photoinduced Electron Transfer:
A Reporter for Conformational Dynamics of Macromolecules. Chemphyschem 2009, 10 (9–10),
1389–1398.
(20) Schulze, A.; Beliu, G.; Helmerich, D. A.; Schubert, J.; Pearl, L. H.; Prodromou, C.; Neuweiler, H.
Cooperation of Local Motions in the Hsp90 Molecular Chaperone ATPase Mechanism. Nat. Chem.
Biol. 2016, 12 (8), 628–635.
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
(21) Ries, J.; Schwarze, S.; Johnson, C. M.; Neuweiler, H. Microsecond Folding and Domain Motions
of a Spider Silk Protein Structural Switch. J. Am. Chem. Soc. 2014, 136 (49), 17136–17144.
24
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 26 of 31
1
2
3
4
5
6
7
8
(22) Vaiana, A. C.; Neuweiler, H.; Schulz, A.; Wolfrum, J.; Sauer, M.; Smith, J. C. Fluorescence
Quenching of Dyes by Tryptophan: Interactions at Atomic Detail from Combination of Experiment
and Computer Simulation. J. Am. Chem. Soc. 2003, 125 (47), 14564–14572.
(23) Doose, S.; Neuweiler, H.; Sauer, M. A Close Look at Fluorescence Quenching of Organic Dyes by
Tryptophan. Chemphyschem 2005, 6 (11), 2277–2285.
(24) Hanoian, P.; Liu, C. T.; Hammes-Schiffer, S.; Benkovic, S. Perspectives on Electrostatics and
Conformational Motions in Enzyme Catalysis. Acc. Chem. Res. 2015, 48 (2), 482–489.
(25) Narayanan, C.; Bernard, D. N.; Doucet, N. Role of Conformational Motions in Enzyme Function:
Selected Methodologies and Case Studies. Catalysts 2016, 6 (6):81, 1-17.
(26) Kohen, A. Role of Dynamics in Enzyme Catalysis: Substantial versus Semantic Controversies. Acc.
Chem. Res. 2015, 48 (2), 466–473.
(27) Ding, B.; Hilaire, M. R.; Gai, F. Infrared and Fluorescence Assessment of Protein Dynamics: From
Folding to Function. J. Phys. Chem. B 2016, 120 (23), 5103–5113.
(28) Kleckner, I. R.; Foster, M. P. An Introduction to NMR-Based Approaches for Measuring Protein
Dynamics. Biochim. Biophys. Acta 2011, 1814 (8), 942–968.
(29) Orozco, M. A Theoretical View of Protein Dynamics. Chem. Soc. Rev. 2014, 43 (14), 5051–5066.
(30) Lu, H. P. Sizing up Single-Molecule Enzymatic Conformational Dynamics. Chem. Soc. Rev. 2014,
43 (4), 1118–1143.
(31) Yang, H.; Luo, G.; Karnchanaphanurach, P.; Louie, T.-M.; Rech, I.; Cova, S.; Xun, L.; Xie, X. S.
Protein Conformational Dynamics Probed by Single-Molecule Electron Transfer. Science 2003,
302 (5643), 262–266.
(32) Blank, K.; De Cremer, G.; Hofkens, J. Fluorescence-Based Analysis of Enzymes at the Single-
Molecule Level. Biotechnol. J. 2009, 4 (4), 465–479.
(33) Rogers, J. M. G.; Polishchuk, A. L.; Guo, L.; Wang, J.; DeGrado, W. F.; Gai, F. Photoinduced
Electron Transfer and Fluorophore Motion as a Probe of the Conformational Dynamics of
Membrane Proteins: Application to the Influenza A M2 Proton Channel. Langmuir 2011, 27 (7),
3815–3821.
(34) Neuweiler, H.; Johnson, C. M.; Fersht, A. R. Direct Observation of Ultrafast Folding and
Denatured State Dynamics in Single Protein Molecules. Proc. Natl. Acad. Sci. U. S. A. 2009, 106
(44), 18569–18574.
(35) Miller, S. R. An Appraisal of the Enzyme Stability‐activity Trade‐off. Evolution 2017, 71 (7),
1876–1887.
(36) Kingsley, L. J.; Lill, M. A. Substrate Tunnels in Enzymes: Structure-Function Relationships and
Computational Methodology. Proteins 2015, 83 (4), 599–611.
(37) Koudelakova, T.; Chovancova, E.; Brezovsky, J.; Monincova, M.; Fortova, A.; Jarkovsky, J.;
Damborsky, J. Substrate Specificity of Haloalkane Dehalogenases. Biochem. J. 2011, 435 (2), 345–
354.
(38) Tyzack, J. D.; Furnham, N.; Sillitoe, I.; Orengo, C. M.; Thornton, J. M. Understanding Enzyme
Function Evolution from a Computational Perspective. Curr. Opin. Struct. Biol. 2017, 47, 131–139.
(39) Levenberg, K. A Method for the Solution of Certain Non-Linear Problems in Least Squares. Q.
Appl. Math. 1944, 2 (2), 164–168.
(40) Marquardt, D. An Algorithm for Least-Squares Estimation of Nonlinear Parameters. J. Soc. Ind.
Appl. Math. 1963, 11 (2), 431–441.
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
(41) Johnson, K. A.; Simpson, Z. B.; Blom, T. Global Kinetic Explorer: A New Computer Program for
Dynamic Simulation and Fitting of Kinetic Data. Anal. Biochem. 2009, 387 (1), 20–29.
(42) Kim, Y.; Ho, S. O.; Gassman, N. R.; Korlann, Y.; Landorf, E. V.; Collart, F. R.; Weiss, S. Efficient
Site-Specific Labeling of Proteins via Cysteines. Bioconjug. Chem. 2008, 19 (3), 786–791.
(43) Iwasaki, I.; Utsumi, S.; Ozawa, T. New Colorimetric Determination of Chloride Using Mercuric
Thiocyanate and Ferric Ion. Bull. Chem. Soc. Jpn. 1952, 25 (3), 226–226.
25
ACS Paragon Plus Environment

Page 27 of 31
1
2
3
Journal of the American Chemical Society
(44) Kelly, S. M.; Price, N. C. The Use of Circular Dichroism in the Investigation of Protein Structure
and Function. Curr. Protein Pept. Sci. 2000, 1 (4), 349–384.
(45) Macháň, R.; Kapusta, P.; Hof, M. Statistical Filtering in Fluorescence Microscopy and
Fluorescence Correlation Spectroscopy. Anal. Bioanal. Chem. 2014, 406 (20), 4797–4813.
(46) Altschul, S. F.; Madden, T. L.; Schäffer, A. A.; Zhang, J.; Zhang, Z.; Miller, W.; Lipman, D. J.
Gapped BLAST and PSI-BLAST: A New Generation of Protein Database Search Programs.
Nucleic Acids Res. 1997, 25 (17), 3389–3402.
4
5
6
7
8
9
(47) NCBI Resource Coordinators. Database Resources of the National Center for Biotechnology
Information. Nucleic Acids Res. 2018, 46 (D1), D8–D13.
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
(48) Hanwell, M. D.; Curtis, D. E.; Lonie, D. C.; Vandermeersch, T.; Zurek, E.; Hutchison, G. R.
Avogadro: An Advanced Semantic Chemical Editor, Visualization, and Analysis Platform. J.
Cheminformatics 2012, 4 (1), 17.
(49) Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.; Robb, M. A.; Cheeseman, J. R.;
Scalmani, G.; Barone, V.; Petersson, G. A.; Nakatsuji, H.; Caricato, M.; Li, X.; Hratchian, H. P.;
Izmaylov, A. F.; Bloino, J.; Zheng, G.; Sonnenberg, J. L.; Hada, M.; Ehara, M.; Toyota, K.;
Fukuda, R.; Hasegawa, J.; Ishida, M.; Nakajima, T.; Honda, Y.; Kitao, O.; Nakai, H.; Vreven, T.;
Montgomery, Jr., J. A.; Peralta, J. E.; Ogliaro, F.; Bearpark, M.; Heyd, J. J.; Brothers, E.; Kudin, K.
N.; Staroverov, V. N.; Kobayashi, R.; Normand, J.; Raghavachari, K.; Rendell, A.; Burant, J. C.;
Iyengar, S. S.; Tomasi, J.; Cossi, M.; Rega, N.; Millam, J. M.; Klene, M.; Knox, J. E.; Cross, J. B.;
Bakken, V.; Adamo, C.; Jaramillo, J.; Gomperts, R.; Stratmann, R. E.; Yazyev, O.; Austin, A. J.;
Cammi, R.; Pomelli, C.; Ochterski, J. W.; Martin, R. L.; Morokuma, K.; Zakrzewski, V. G.; Voth,
G. A.; Salvador, P.; Dannenberg, J. J.; Dapprich, S.; Daniels, A. D.; Farkas, Ö.; Foresman, J. B.;
Ortiz, J. V.; Cioslowski, J.; Fox D. J. Gaussian 09 Rev. D.01; Wallingford, CT, 2009.
(50) Dupradeau, F.-Y.; Pigache, A.; Zaffran, T.; Savineau, C.; Lelong, R.; Grivel, N.; Lelong, D.;
Rosanski, W.; Cieplak, P. The R.E.D. Tools: Advances in RESP and ESP Charge Derivation and
Force Field Library Building. Phys. Chem. Chem. Phys. 2010, 12 (28), 7821–7839.
(51) Vanquelef, E.; Simon, S.; Marquant, G.; Garcia, E.; Klimerak, G.; Delepine, J. C.; Cieplak, P.;
Dupradeau, F.-Y. R.E.D. Server: A Web Service for Deriving RESP and ESP Charges and Building
Force Field Libraries for New Molecules and Molecular Fragments. Nucleic Acids Res. 2011, 39,
W511-517.
(52) Bayly, C. I.; Cieplak, P.; Cornell, W.; Kollman, P. A. A Well-Behaved Electrostatic Potential
Based Method Using Charge Restraints for Deriving Atomic Charges: The RESP Model. J. Phys.
Chem. 1993, 97 (40), 10269–10280.
(53) Cornell, W. D.; Cieplak, P.; Bayly, C. I.; Gould, I. R.; Merz, K. M.; Ferguson, D. M.; Spellmeyer,
D. C.; Fox, T.; Caldwell, J. W.; Kollman, P. A. A Second Generation Force Field for the
Simulation of Proteins, Nucleic Acids, and Organic Molecules. J. Am. Chem. Soc. 1995, 117 (19),
5179–5197.
(54) VanBeek, D. B.; Zwier, M. C.; Shorb, J. M.; Krueger, B. P. Fretting about FRET: Correlation
between κ and R. Biophys. J. 2007, 92 (12), 4168–4178.
(55) Wang, J.; Wolf, R. M.; Caldwell, J. W.; Kollman, P. A.; Case, D. A. Development and Testing of a
General Amber Force Field. J. Comput. Chem. 2004, 25 (9), 1157–1174.
(56) Case, D. A.; Betz, R. M.; Botello-Smith, W.; Cerutti, D. S.; Cheatham III, T. E.; Darden, T. A.;
Duke, R. E.; Giese, T. J.; Gohlke, H.; Goetz, A. W.; Homeyer, N.; Izadi, S.; Janowski, P. A.; Kaus,
J.; Kovalenko, A.; Lee, T. S.; LeGrand, S.; Li, P.; Lin, C.; Luchko, T.; Luo, R.; Madej, B.;
Mermelstein, D.; Merz, K. M.; Monard, G.; Nguyen, H.; Nguyen, H. T.; Omelyan, I.; Onufriev, A.;
Roe, D. R.; Roitberg, A.; Sagui, C.; Simmerling, C. L.; Swails, J.; Walker, R. C.; Wang, J.; Wolf,
R. M.; Wu, X.; Xiao, L.; York, D. M.; Kollman, P. A., AMBER16. University of California, San
Francisco, 2016.
26
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 28 of 31
1
2
3
4
5
6
7
8
(57) Jorgensen, W. L.; Chandrasekhar, J.; Madura, J. D.; Impey, R. W.; Klein, M. L. Comparison of
Simple Potential Functions for Simulating Liquid Water. J. Chem. Phys. 1983, 79 (2), 926.
(58) Harvey, M. J.; De Fabritiis, G. An Implementation of the Smooth Particle Mesh Ewald Method on
GPU Hardware. J. Chem. Theory Comput. 2009, 5 (9), 2371–2377.
(59) Feenstra, K. A.; Hess, B.; Berendsen, H. J. C. Improving Efficiency of Large Time-Scale
Molecular Dynamics Simulations of Hydrogen-Rich Systems. J. Comput. Chem. 1999, 20 (8), 786–
798.
(60) Hopkins, C. W.; Le Grand, S.; Walker, R. C.; Roitberg, A. E. Long-Time-Step Molecular
Dynamics through Hydrogen Mass Repartitioning. J. Chem. Theory Comput. 2015, 11 (4), 1864–
1874.
(61) Naritomi, Y.; Fuchigami, S. Slow Dynamics in Protein Fluctuations Revealed by Time-Structure
Based Independent Component Analysis: The Case of Domain Motions. J. Chem. Phys. 2011, 134
(6), 065101.
(62) Doerr, S.; De Fabritiis, G. On-the-Fly Learning and Sampling of Ligand Binding by High-
Throughput Molecular Simulations. J. Chem. Theory Comput. 2014, 10 (5), 2064–2069.
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
27
ACS Paragon Plus Environment

Page 29 of 31
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
338x190mm (96 x 96 DPI)
ACS Paragon Plus Environment