S
1
SUPPORTING INFORMATION
Tryptophan to Tryptophan Hole Hopping in an Azurin Construct
Martin Melčák,
a,b
Filip Šebesta,
a,c
Jan Heyda,
a,b
Harry B. Gray,*
,d
Stanislav Záliš,*
,a
Antonín
Vlček*
,a,e
a
J. Heyrovsk
ý
Institute of Physical Chemistry, Czech Academy of Sciences, Dolejškova 3, CZ
-
182
23 Prague, Czech Republic
b
Department of Physical Chemistry, University of Chemistry and Technology Prague, Technická
5, CZ
-
166 28 Prague, Czech Republic
c
Dep
artment of Chemical Physics and Optics, Faculty of Mathematics and Physics, Charles
University, Ke Karlovu 3, CZ
-
121 16 Prague, Czech Republic
d
Beckman Institute, California Institute of Technology, Pasadena, California 91125, United States
e
Department o
f Chemistry, Queen Mary University of London, E1 4NS London, U.K.
S
2
Contents
Fig
.
S1
-
2
:
Ground
-
state MM/MD trajectories.
S
3
-
4
Fig. S3:
Typical Re126W124W122Cu
I
structures
S5
Fig. S4:
Unreactive CS1 MM/MD trajectories
S6
Fig. S5
-
6:
Charge an
d spin
QM/MM/MD
trajectories
S6
-
7
Fig. S7:
Distances and angles along
QM/MM/MD
reactive trajectories
S8
Fig. S8:
Visualization of an indole surface for calculating electrostatic potentials
S8
Fig. S9
-
11:
Trajectories of electrostatic potentials
S9
-
11
Fig. S12:
Visualization of proximal volume shells used to calculate water
proximal distribution functions
S12
Fig. S13
-
14:
Water proximal radial distribution functions
S
13
-
14
Fig. S15:
Trajectories of protein
-
generated electrostatic
potentials
S15
Fig. S16:
Trajectories of W122
-
SAL distances
S16
Fig. S17:
Evolution of the W122 indole
-
NH
–
OH
2
distances
S17
Fig. S18:
Trajectories of
Re
–
-
generated electrostatic potentials
S
1
7
Fig. S19:
Characteristics of a typical unreactiv
e "
in
" trajectory
S
1
8
Fig. S20:
Electrostatic potentials along a typical unreactive "
in
" trajectory
S
1
9
Fig. S21
-
22:
Distributions of electrostatic potentials over unreactive "
in
" and "
out
"
trajectories
S20
-
2
1
Fig. S23:
Structures showing W124,
Re
–
, and Q107 orientations in reactive and unreactive
trajectories
S
2
2
Fig. S24:
Characteristics of a typical "
out
" trajectory
S
2
3
Fig. S25:
Temporal evolution of "
in
"
-
MLCT and "
in
"
-
CS1 populations
S
2
4
Section
S
1.
Computational details
S
2
5
S
1.1.
General procedure
S
2
5
S
1.2.
Classical MM/MD simulations
S
2
5
S
1.3.
QM/MM
/
MD simulations
S
2
6
S
1.4.
CAM
-
B3LYP test calculations
S
2
7
S
1.5.
Electronic coupling
S
2
7
S
1.6.
Electrostatic potentials
S
2
8
S
1.7.
Proxim
al volume, coordination number, and distribution function
S
2
9
S
1.8.
Atomic charges for the CS1 and CS2 states
S30
References
S35
S
3
Figure S1.
G
rou
n
d
-
state
M
M/M
D
trajectories of
Re126W
124
W
122
Cu
I
.
Black: the shortest
distance between atoms of
W122 and W124
indole
aromatic rings.
Red: The shortest distance
between atoms of W124 and dmp aromatic rings.
Starting structures for CS1 and CS2 simulations
were taken from the first three trajectories, as sho
wn in Figure 2.
S
4
Figure S
2
.
G
rou
n
d
-
state
MM/MD trajectories of the shortest CO
-
W124(indole) distances in
Re126W
124
W
122
Cu
I
.
Black
,
green
:
equatorial CO ligands.
Red:
axial CO. The trajectories are in
the same order as in Figure S1.
S
5
Figure S3.
Selected
ground
-
state
Re126W
124
W
122
Cu
I
structures.
A
:
C
rystal structure
1
is a
pro
to
typal reactive configuration
.
B
-
D show that the indole
-
indole
distance and orientation vary
in both conformations.
The
Re
-
H
12
6
-
L125
-
W12
4
-
G12
3
-
W122
unit is
shown
as
a stick
representation.
A:
Crystal structure, "
in
"
C: "
out
"
B: "
in
"; W124
-
W122 distorted
D: "
out
"; W124
-
W122 distorted
S
6
Figure S
4
.
Top four panels:
Typical CS1 MM/MD unreactive trajectories of
Re126W
124
W
122
Cu
I
.
Black: shortest distance between W122 and W124 aromatic
-
ring atoms. Red: shortest distance
between W124(indole) and dmp atoms. Top
-
left trajectory (2
-
3.3) stays the whole time in the
"
in
" conformation (indole
-
indole mean distance 4.0 Å; indole
-
dmp mean di
stance = 3.4 Å). The
other three trajectories show conversion to the "
out
" conformation.
Bottom: A 10
-
ns trajectory showing short
-
lived "
out
"
"
in
" conversions.
Figure S5.
Left and middle: CS1
QM/MM/MD
charge trajectories (A, D) showing reverse ET to *Re
with a predominantly
*(dmp)
electronic structure
. Right: a typical
unreactive "
in
"
-
CS1
QM/MM/MD
charge trajectory (2
-
3.3).
S
7
Figure S
6
.
E
volution of
spins at the two indoles (W124, W122), Re(CO)
3
,
and dmp along CS1 UKS
QM/
MM/
MD trajectories.
Top row: Reactive trajectories showing conversion to CS2. Bottom left
and middle: Trajectories showing
reverse
ET
to
*Re
o
f
predominant
3
* IL(dmp) character
.
Bottom right:
T
ypical unreactive
trajectory
of a C
S1 state in the "
in
" conformation (2
-
3.3)
.
L
etters
specify the starting ground
-
state structures (Figure
2
).
S
8
Figure
S7.
Distances and angles
along reactive trajectories
. Top: Shortest
indole
-
indole (red) and
W122
-
indole
–
dmp
distances (black). From Figure 7. Middle: Center
-
center indole
-
indole (red)
and W122
-
indole
–
dmp distances (b;ack)(from Figure 7). Bottom: Center
-
center distances (red)
and angles between the two indoles.
Figure S8.
Example of an indole surface used
in electrostatic potential calculations
.
Further details are provided in Section S1.5.
S
9
Figure S
9
.
Temporal evolution of charge
and potentials along trajectory C.
Lef
t
-
t
op
: Difference between
W122 and W124
charges
(
q,
blue). Left
-
middle: E
lectrostatic
potentials
at W124 and W122
generated by all atoms in the system
including
the other indole
.
Left
-
bottom:
E
lectrostatic potentials
at W124 and W122
generated by all atoms in the system
except the
other
indole
(
(124) and
(122) as used in the
text)
.
Right:
Differences between charges (
q,
blue) and electrostatic potentials at the two indoles.
Red: Total potentials generated by all atoms
including the other indole
. Black: Potentials
generated by all atoms without the other indole
(
)
.
The dotte
d vertical line marks the time when the charges at W124 and W122 bec
a
me equal for
the first time.
P
otential trajectories with and without indoles follow the same trends
; and
t
he
y
both cross
at
the onset of the ET region. (To calculate potentials at W124
except
the
other
indole, W122 was
removed from
the system
,
and vice versa.
In this way, we excluded electrostatic effects of the
shifting charge.
)
S
10
Figure S
10
.
Temporal evolution of charge
s
and
electrostatic
potentials along
trajector
y
B.
Left
-
t
op:
Difference
of electrostatic potentials at W124 and W122 generated by all atoms in the
system except the
other
indole (
,
black), by the solvent (
(solvent),
red), and by the protein
except the
other
indole
(
(prot),
green, shifted by
-
3.5 V (B) for
clarity). T
he charge difference
between W122 and W124
(
q)
is shown in
blue.
Left
-
b
ottom: Solvent
-
generated potentials at W124 and W122.
The dotted vertical line marks the time when the charges at W124 and W122 bec
a
me equal for
the first time.
Right:
Electrostatic potentials at W124 and W122 generated by all atoms (top) and except the
other indole (bottom).
Note that potential trajectories with and without the other indole follow similar trends, especially
that
they both cross at the onset of the ET r
egion.
(To calculate potentials at W124
without
the
other indole, W122 was
removed from
the system, and vice versa.
T
his way, we have excluded
electrostatic effects of the shifting charge.)
S
11
Figure S
1
1
.
Temporal evolution of charge
s
and electrostatic
potentials along trajectory E.
Left
-
t
op: Difference of electrostatic potentials at W124 and W122 generated by all atoms in the
system except the other indole (
,
black), by the solvent (
(solvent), red), and by the protein
except the other indole (
(pr
ot), green, shifted by
-
3.5 V (B) for clarity). The charge difference
between W122 and W124 (
q) is shown in blue.
Left
-
b
ottom: Solvent
-
generated potentials at W124 and W122.
The dotted vertical
line marks the time when the charges at W124 and W122 became
equal for
the first time.
Right: Electrostatic potentials at W124 and W122 generated by all atoms (top) except the other
indole (bottom).
Note that potential trajectories with and without the other indole follow similar trends
; and
they
both cross at the
onset of the ET region. (To calculate potentials at W124 without the other
indole, W122 was
removed from
the system, and vice versa. This way, we have excluded
electrostatic effects of the shifting charge.)
S
12
Figure S12.
Non‐overlapping proximal volume shells around individual fragments used to
calculate water proximal radial distribution functions g(
r
) and coordination numbers.
2
-
4
Top
-
left:
Shells around the two indoles, which compete for water among themselves but not with other
fragments. Top
-
right: Shell around dmp that does not compete with any other fragment. Shells
around the three CO ligands that compete for water am
ong themselves but not with other
fragments.
dmp
W124
W122
CO
S
13
Figure S13.
Water coordination numbers of W122 and W124 calculated from CS1 and CS2 parts
of the reactive trajector
ies
(
clockwise from
top
-
left
) and averaged MM/MD trajectories of CS1
and CS2 (bottom
-
lef
t
).
Each water molecule was assigned solely to its closest residue. The
corresponding
non‐overlapping proximal volume shells are depicted in Figure
S12.
S
14
Figure S1
4
.
Water proximal distribution functions of W122 and W124 calculated from CS1 and
CS2 parts of the
trajectories (clockwise from top
-
left)
and averaged MM/MD trajectories of CS1
and CS2 (bottom
-
left
).
Each water molecule was assigned solely to its closest res
idue.
The
corresponding
non‐overlapping proximal volume shells are depicted in Figure
S12.
S
15
Figure S1
5
.
Electrostatic potential at W124 (red) and W122 (black) generated by the protein
.
Green: potential
at W122 generated by the
S118A119L120
segment.
("Protein" contains
Re
–
but not the other indole.)
S
16
Figure S1
6
.
Top: Shielding of W122 by the SAL segment. S118
light
-
blue
, A119
yellow
, L120
pink
.
(See also Figure 8.)
Bottom:
Shortest distances between
W122
and individual residues of the
SAL
segment. ET starts
in the region of increasing distances to all three residues
(blue dotted line)
. Hydrogen atoms were
included in distance calculations.
W122
S
17
Figure S17.
Evolution of the W122 indole
-
NH
–
OH
2
distance for the water molecule making a
bri
dge to the A119 amide
-
O atom (red
, water #369
)
,
to the first water molecule of the bridge to
the S118 amide
-
O atom (
black
,
water #281
)
; and to a water molecule #457
(blue)
that approaches
in the direction from W124
.
Calculated along trajectory C.
The A119
amide
-
O
H
2
O
HN
-
W122 bridge
emerged
ca. 400 fs before the ET onset and stayed
stable through the ET and CS2 regions of the trajectory. The S118 amide
-
O
H
2
O
H
2
O
HN
-
W122 chain also formed ca. 400 fs before the ET onset
.
The
(
122
-
solv) decrease was
reinforced
by another H
2
O molecule (#457, blue) that approached in the direction from W124. The S118
amide
-
O
H
2
O
H
2
O
HN
-
W122 chain
opened up
late
r
in the ET region
when water #281 moved
away
,
together with #457
, as SAL shifted away from W122
. However
, the combined electrostatic
field of these two molecules kept the
(
122
-
solv) low. Both these water molecules
started
moving back toward W122
after ~1250 fs (#457) and ~1500 fs (#281), helping to drive the ET to
completion and stabilizing
CS2
at
2000 fs
.
Figure S1
8
.
Electrostatic potential at W124 (red) and W122 (black) generated by
Re
–
along
trajectory C.
S
18
Figure S19.
Characteristics of the u
nreactive trajectory 2
-
3
.
3
, where
Re
–
keeps the "
in
"
orientation and the indoles are relatively close to
each other.
T
op
-
left
:
Charges at molecular fragments and CS1
-
CS2 electronic coupling
B
ottom
-
left
:
Shortest indole
-
indole and W122
-
dmp distances (H atoms not considered)
T
op
-
right
: Proximal radial water distribution functions around W124 and W122
indoles
.
Inset: Water coordination number up to 3 Å.
Compare with Figure S14
.
Bottom
-
r
ight: Shortest distances between the W122 indole and SAL. Compare with Figure S16.
S
19
Figure S20.
Electrostatic potentials calculated from the unreactive tr
ajectory 2
-
3.3, where
Re
–
keeps the "
in
" orientation and the indoles are relatively close to each other.
Left
-
top:
Distributions of
(124) and
(122) over the 3 ps
QM/MM/MD
trajectory. The maxima
are separated by ca. 0.8 V.
(124) is more negative, stabili
zing W124
•
+
(CS1).
Left
-
bottom:
The d
ifference between potentials at the two indoles,
(124)
-
(122) along the
QM/MM/MD
trajectory stays <0.5, that is
always
below the 1.1 V level where the ET occurs.
Right
-
top
: Distributions of
(124
-
solv) and
(122
-
solv) over the 3 ps
QM/MM/MD
trajectory. The
maxima are separated by ca.
1
.
7
V.
(124
-
solv
) is more negative, stabilizing W124
•
+
(CS1).
Right
-
bottom
:
(124
-
solv) and
(122
-
solv) along the 3 ps
QM/MM/MD
trajectory.
(124
-
solv) <<
(122
-
solv) all the t
ime, stabilizing W124
•
+
(CS1).
The solvent
-
generated potentials never
equalized in the course of the 3 ps simulation.
S
20
Figure
S2
1
.
Distributions of electrostatic potentials at W124 (top) and W122 (bottom) indoles
over "
in
" (2
-
3.3) and "
out
" (2
-
2.4) unreactive
QM/MM/MD
trajectories
.
The width of the
potential range is the same in each column.
Left column: On going from "
in
" to "
out
" conformation, the distribution of total potentials at both
indoles shifts lower but more at W124, resulting in
a more stable CS1 relative to CS2.
Middle column: On going from "
in
" to "
out
" conformation, the distribution of solvent
-
generated
potential
s
at W124
shifts
slightly
higher
than at W122
. Hence, solvation stabilizes
in
-
CS1
slightly
more than
out
-
CS1.
Right
column
:
On going from "
in
" to "
out
" conformation, the distribution of protein
-
generated
potentials at both indoles shifts lower but much more at W124,
resulting in strong CS1
stabilization relative to CS2.
Protein
-
generated electrostatic potential is resp
onsible for the
overall out
-
CS1 stabilization indicated by the left column.
S
21
Figure
S2
2
.
Distributions of electrostatic potentials at W124 (
left
) and W122 (
right
) indoles
over
"
in
" (2
-
3.3) and "
out
" (2
-
2.4) unreactive
QM/MM/MD
trajectories generated by
Re
–
and Q107.
The potential range is the same in each panel.
Electrostatic potential
s
generated at W124 by
Re
–
and, to a lesser extent, Q107 are lower in the
"
out
" than the "
in
" form.
Re
–
and, to a lesser extent, Q107 are the
main contributors to the
out
-
CS1 stabilization relative to "
in
" by the protein
-
generated potential
s
shown in
Figure S22.
S
22
Figure S2
3
.
Relative orientations of the W124 indole (light blue),
Re
–
(green) and Q107 (yellow).
Top
-
left: unreactive "
in
"
QM/MM/MD
trajectory 2
-
3.3 at 2010 fs.
Top
-
right: unreactive "out"
QM/MM/MD
trajectory 2
-
2.4 at 2010 fs.
Bottom: reactive
QM/MM/MD
trajectory C at 900 fs (ET start time).
The Q107 peptide O
-
atom is relatively close to W124 in all cases. The amide side chain is tilted
toward the dmp
•
–
in the reactive (C) as well as unreactive "
in
" cases. In the "
out
" form, it is tilted
toward W122. The corresponding O
-
W122 distance (4.2 Å to
nearest C
-
atom) is shorter than in
unreactive "
in
" and reactive cases.
Positively charged H
-
atoms of one of the dmp
•
–
CH
3
groups are very close to W122 in C and
unreactive "
in
" (2.6
-
2.7 Å) and much farther in "
out
" (9.2 Å)
. Moreover, one of the C
O
ligands
in
the "
out
" form point
s
toward
W124
. T
he closest
distance between the negatively charged (ca.
-
0.
17
e) O atom and W12
4
is
4.2 Å.
T
hese
structural differences
result in
a much smaller
electrostatic potential at W124 generated by
Re
–
in the "
out
" th
an "
in
" conformer.
"
out
"
unreactive "
in
"
reactive C at 900 fs
S
23
Figure S
24
.
Characteristics of the unreactive trajectory 2
-
2.4, where
Re
–
assumes
the "
out
"
orientation and the indoles are relatively close to each other.
Top
-
left: Charges at molecular fragments and CS1
-
CS2 electronic coupling
Bottom
-
left: Shortest indole
-
indole and W122
-
dmp distances (H atoms not considered)
Top
-
right: Proximal radial water distribution functions around W124 and W122 indoles.
Inset: Water coordination number
s
up
to 3 Å. Compare with Figure S14.
Bo
ttom
-
right: Shortest distances between the W122 indole and SAL. Compare with Figure S16.
Generally, unreactive "
out
" and "
in
" trajectories are very similar. W124 is more solvated in the
second solvation sphere, W122 less in the case of the "
out
" form. C
ompare with Figure S19.
S
24
Figure S25.
Temporal evolution of "
in
"
-
MLCT and "
in
"
-
CS1
populations
from all calculated
MM/MD trajectories.
The "
in
" geometry was defined as dmp
–
W124 < 5 Å.
S
25
S1.
Computational details
S1.1. General procedure.
The simulation protocol is summarized in the main text, Figure 1. In order to realistically
simulate the evolution of CS1 and CS2 states by QM/MM, we prepared the initial thermalized
(300 K) geometries by mimicking the experimentally established mechanism
(Figure 1). We first
performed 12.5 ns MM/MD
NpT
simulations with GS parameters to generate a set of
independent geometries of the protein native state, sampling different sidechain conformations
and solvent distributions. Next, starting from hundreds of r
andomly selected geometries, we
performed 1 ns long MM/MD simulations with MLCT FF parameters, followed by 1 ns long
MM/MD simulations with CS1 FF parameters. Simulation lengths were restricted to 1 ns since
long propagation of excited states (significantl
y longer than their lifetimes) by classical force
-
fields may lead to population of unrealistic (over
-
relaxed) geometries that would bias both the
ensemble averages and available reactive pathways. The final geometries and velocities of MM
simulations serve
d as inputs for excited
-
state UKS QM/MM/MD runs. Structural parameters
were monitored in all
simulations (see below). UKS simulations were used to study the ET
reactivity by following the charge and spin transfer between
Re
, W124, and W122. The same
proced
ure was applied to investigate the CS2 state, whose MM/MD simulations started from CS1
MM/MD structures (Figure 1, main text).
S1.2. Classical MM/MD simulations
We performed classical MM/MD simulations using AMBER 14 software and parameters
5
for the MM part of the system.
In order to properly parametrize the
QM
-
region, we employ
ed a
unique set of parameters that we have previously developed
4
for [Re(imidazole)(CO)
3
(dmp)]
+
.
Atomic charges for CS1 and CS2 triplet states of the QM part were obtained as described in
Section S1.8. T
he
values
are summarized in Table S1. For the rest of the protein, the
ff14SB
modifications
of
parm10
parameters
6, 7
were employed
.
The vicinity of the Cu
I
atom was not
investigated in this work. To keep a realistic geometry, we restrained the Cu
-
ligand distances
using data from Table 1 of ref.
8
The SPC/E model was used for
explicit water surroundings
9
and
two Na
+
cations
were added to
neutralize the system
.
Since the Terachem
-
Amber QM/MM does