of 26
S
1
Supporting
Information
The Oxidation State of [4Fe4S] Clusters Modulates the DNA
-
Binding Affinity of DNA Repair
Proteins
Edmund C. M. Tse, Theodore J.
Zwang, Jacqueline K. Barton
*
Division of Chemistry and Chemical Engineering, California Institute of
Technology, Pasadena,
California 91125, United States
Table of Contents:
Supplementary Figures S1
-
13
Supplementary Tables S1
-
9
Supplementary Experimental Procedures
Supplementary References
S
2
Supplementary
Figures
& Tables
Figure S1
displays
cyclic
voltammograms (CVs)
of a DNA
-
modified Au electrode in
DNA
buffer with Endo
III
in various concentrations
in solution.
The current of the redox wave increases
as a function of increasing EndoIII concentration, further supporting the notion
1
that the redox
wave is likely related to EndoIII in solution. 1
M was chosen for bulk electrolysis to increase
oxidation yield.
Figure
S
1
.
Cyclic voltamm
etry of EndoIII on a DNA
-
modified electrode.
CVs in
DNA
buffer
(pH
7.0, 5 mM NaH
2
PO
4
, 50 mM NaCl
)
with 100 nM
EndoIII
(black), 1
M
EndoIII
(red), and 10
M
EndoIII
(blue) added at a scan rate of 100 mV/s.
E
mid
= +80 mV vs. NHE.
Square wave voltammetry (SWV) and differential pulse voltammetry (DPV)
eliminate
background capacitive current, thereby allo
wing the redox peak to b
e visualized with relative ease
(Figure S2).
S
3
(a)
(b
)
S
4
Figure S2
.
Square wave voltammetry (SWV) and differential pulse voltammetry (DPV) of EndoIII
on a
DNA
-
modified electrode.
(a) Cathodic and (b) anodic SWVs in
DNA
buffer
(pH 7.0, 5 mM
NaH
2
PO
4
, 50 mM NaCl
)
with 100 nM
EndoIII
(black), 1
M
EndoIII
(red), and 10
M
EndoIII
(blue) added. (C) Cathodic (red) and anodic (black) DPVs in
DNA
buffer
(pH 7.0, 5 mM NaH
2
PO
4
,
50 mM NaCl
)
with 10
M
EndoIII
added.
To determine
if the redox reaction of EndoIII on DNA
-
modified electrodes is a diffusionless
or a d
i
ffusive process, we conducted Randles
-
Sevcik analysis by measuring the peak current
responses at various scan rates while holding the concentration of EndoIII in solution
constant
(Figure S3). In Figure S4b, a
poor linear relationship is observed
for the
Randles
-
Sevcik plot of
Endo
III
with the peak current versus the scan rate,
indicating
that the redox event does not originate
from a surface
-
bound species and that Endo
III
does not
adsorb
very
strongly
on
to DNA
-
modified
electrodes
under
this condition.
By contrast, the peak currents of both the anodic and cathodic
peaks scale with the square root of the scan rate (Figure S4a), indicating that under these conditions
the redo
x reaction of EndoIII is in a diffusion
-
controlled regime.
Therefore, at concentrations at or
below 10
M, Endo
III
likely diffuses from the bulk solution to the electrode surface to undergo
redox reaction upon binding to
DNA and then diffuses back into the
bulk solution to allow for
freshly exposed DNA surfaces for other
copies of EndoIII
to
approach and
participate in
subsequent redox events.
(c)
S
5
Figure S3
a
.
Cyclic voltammetry of EndoIII on a
DNA
-
modified electrode with varying scan rates.
CVs in
DNA
buffer
(pH 7.0, 5 mM NaH
2
PO
4
, 50 mM NaCl
)
with 10
M
EndoIII
added at a scan
rate of 50 (black), 100 (red), 200 (blue), 400 (green), and 800 (orange) mV/s. 10
M is chosen as
the concentration to allow for clear visualization of the redox waves
.
Figure S3b
.
UV
-
visible
absorbance
spectr
a
of EndoIII
. The [4Fe4S] redox cofactor exhibits an
of
17,000 M
-
1
cm
-
1
at 410 nm.
2
Cluster loading was at least 70%.
S
6
Figure
S4
.
Randles
-
Sevcik plots of the peak currents of Endo
III.
Randles
-
Sevcik plots of the peak
currents of 10
M Endo
III
at various
scan rates recorded in Figure S3
a
versus (a) the square root
of the scan rate, and (b) the scan rate.
To check if the protein structure is altered by the bulk electrolysis process, we monitored
qualitatively the changes to the
secondary
structure of EndoIII by circular dichroism (CD).
(a)
(b)
S
7
Figure
S5a
. Bulk electrolysis in
DNA
buffer
(pH 7.0, 5 mM NaH
2
PO
4
, 50
mM NaCl
, black line)
with 1
M Endo
III
added (red
line
) held at
50 mV
vs. NHE
.
Only a small portion of the proteins
present in the solution was reduced, suggesting that the native or as
-
isolated oxidation state of the
protein is primarily in the reduced
[
4Fe4S]
2+
s
tate.
Figure
S5b
.
The influence of bulk electrolysis on EndoIII structure as monitored by circular
dichroism.
CD spectra of
DNA
buffer
(pH 7.0, 5 mM NaH
2
PO
4
, 50 mM NaCl
, black line)
with
native (green), reduced (blue), and oxidized (red)
Endo
III
(5
M
,
1000
L
)
added.
The slight
change in signal intensity is likely due to a change in
the
optical density
.
Measuring the ratio of
the CD signal at 222 nm to that at 208 nm is typically used to analyze
-
helic
al
proteins.
Coiled
Ratio(222/208)
1.15 ± 0.13
1.15 ± 0.15
1.16 ± 0.18
S
8
coils and
isolated
-
helix
exhibit
ratio
s
of 1.1
and
0.9
, respectively.
Here, native, reduced,
and
oxid
ized EndoIII exhibit ratios
of
c
a.
1.15, suggesting that EndoIII retains
its coiled coils and that
the overall structure of EndoIII is
not perturbed by bulk electrolysis.
Figure S6
. C
ontinuous wave (CW)
electron paramagnetic resonance (EPR) spectra of
EndoIII.
(a)
(c)
(a)
(b)
S
9
2
M (blue) and 50
M (black)
oxidized Endo
III prepared under anaerobic conditions
, (b)
reduced
(black) and oxidized
(red) 5
M
Endo
III exposed to ambient air, and (c)
p
ulse electron spin echo
envelope modulation (ESEEM) spectra of
reduced
(black) and oxidiz
ed (red) 20
M Endo
III
in
p
rotein buffer (20 mM NaH
2
PO
4
pH 7.5, 100 mM NaCl, 5% glycerol, 1 mM EDTA)
.
Instrument
settings: modulation amplitude = 10 G at 100 kHz; frequency = 9.37 GHz; microwave power =
4.7
mW; and temperature = 10 K.
In Figure S6, for
samples prepared under an N
2
atmosphere, an absence of an EPR signal with
a g value of 2.01 indicates that no detectable [3Fe4S]
1+
degradation product was generated. As a
positive control, for the oxidized EndoIII sample that was exposed to O
2
by passive d
iffusion in
ambient air, we observed an EPR signal at a g value of 2.01, which is indicative of the presence of
degraded [3Fe4S]
1+
clusters.
2
Endo
III
# of DNA strands
# of Proteins
Proteins:DNA
< 1
% Ox
84
120
1.43
33
% Ox
243
368
1.51
66
% Ox
129
229
1.78
> 99
% Ox
395
844
2.14
Table
S
1
.
Probing DNA
-
protein interactions using AFM
. Summary of the average number of
proteins bound on DNA versus the number of DNA strands counted.
Table S1 displays the
number of pUC19 DNA strands observed, the number of
EndoIII
bound
on the countable DNA strands, and the calculated ratio of Endo
III
:pUC19. The results show that
as
EndoIII
becomes mo
re and more oxidized, the
protein:DNA
ratio increases (Figure 2b).
Binding
affinity (K
D
) can be calculated from this set of AFM data using Eq. S1
-
3
.
%
푃푟표푡푒푖푛
푏표푢푛푑
표푛
퐷푁퐴
=
#
표푓
푃푟표푡푒푖푛푠
푏표푢푛푑
표푛
퐷푁퐴
푇표푡푎푙
#
표푓
푃푟표푡푒푖푛푠
=
[
]
[
]
+
퐸푞
.
1
Rearranging
Eq. S1 gives
Eq.
S2
.
=
[
]
%
푃푟표푡푒푖푛
푏표푢푛푑
표푛
퐷푁퐴
×
[
]
%
푃푟표푡푒푖푛
푏표푢푛푑
표푛
퐷푁퐴
퐸푞
.
2
S
10
푅푒푙푎푡푖푣푒
푏푖푛푑푖푛푔
푎푓푓푖푛푖푡푦
푏푒푡푤푒푒푛
푝푟표푡푒푖푛푠
푎푛푑
=
,
,
퐸푞
.
3
Figure S7
.
M
icroscale thermophoresis of
EndoIII. MST plots of (
a
) native and (
b
) oxidized
EndoIII.
E
lectrophoretic mobility shift assay
(EMSA) is commonly used to measure the binding
affinity between proteins and DNA. However, the EMSA experimental setup, even whe
n
conducted in an anaerobic chamber at
4
°
C, is not compatible with oxidized EndoIII. The [4Fe4S]
clusters of DNA repair proteins in the oxidized state degrade in the presence of O
2
. Efforts
including pre
-
running gel in degassed buffer and then switching
to freshly
-
degassed buffer prior
to protein loading were not successful.
T
he amount of O
2
generated by the Pt electrodes during the
gel running process resulted in
cluster
degradation. For an EMSA gel
-
shift experiment, about
2
0
pmol
(
10
L
per lane) of pro
tein is typically used.
The amount of O
2
generated per second during
a
gel
-
shift experiment
is about
10
mol
(
= 50 V
×
0.02 A
×
1 s
×
1 mol
/
96485 C
)
,
and t
he total
amount of O
2
generated
over the course of a 4
-
hour
EMSA gel
-
shift
experiment
is about
37
mmol.
Since the solubility of O
2
in
aqueous solution
is about 1.2
×
10
6
mol cm
-
3
, the
maximum
amount
of dissolved O
2
saturated in 1 L TBE buffer
is ca.
1.2 mmol.
The
O
2
diffusion coefficient
in aqueous
medium is about
1.9
×
10
5
cm
2
s
-
1
. The amount of O
2
liberated during a gel
-
shift experiment is
more than enough to saturate the running buffer with O
2
.
Accordingly,
oxidized EndoIII
likely
is
overwhelmed by O
2
generated
in situ
during the gel running process.
ITC m
easurements
s
how a
28
×
increase in the aff
inity of the predominantly oxidized protein sample for DNA when compared
to the predominantly reduced sample.
However, instrument limitation prevented ITC from being
(a)
(b)
S
11
operated anaerobically.
We therefore switched to using MST to determine the DNA binding af
finity
of EndoIII.
MST is a technique that allows for anaerobic determination of the binding affinity of
O
2
-
sensitive proteins such as FeS cluster biogenesis machinery to their substrates.
3
We utilized
MST to probe the binding affinity of EndoIII to DNA in different redox states. MST results show
that oxidized EndoIII with [4Fe4S]
3+
cluster binds 21
-
mer dsDNA ca. 550 times stronger than
EndoIII predominantly in the reduced state (Figure S7).
This comparison
may
understate the true
difference in affinity between the reduced and oxidized proteins, because it is unlikely that bulk
electrolysis has completely removed all oxidized protein from the reduced sample. The presence
of a small amount of
oxidized protein will
convolute
the affinity measurement in any sample, even
when the oxidized protein is in low concentration compared to the reduced protein, because the
oxidized protein has a significantly higher affinity for DNA. This
phenomenon infers
that the
values measured by
MST
that show a
550
×
change in affinity
between the two protein redox states
may
be obfuscated by a
change in the amount of oxidized protein present in the sample.
Calculating
the 99% confidence interval for the ratio of measur
ed affinities suggests that the oxidized sample
has between
276
×
and
1157
×
increase in affinity over the reduced sample.
S
12
Figure S
8
.
A
model describing the
electrostatic
interactions between EndoIII and DNA.
The crystal structure (PDB:1ORN) revealed that eight phosphates on the DNA backbone
interact with the amino acid residues of EndoIII. The average distance between the redox
-
active
metallocofactor and the eight phosphate groups is about 18
Å
. The neighborin
g phosphate groups
that do not directly interact with the protein binding surface are not crucial in this calculation
because (1) PE
1/d, (2) d increases for phosphate groups further away from the [4Fe4S] cluster,
and (3) the dielectric constant (
) b
etween the two point charges needs to take into account of the
intervening water molecules that have a
of
80.
The electrostatic nature of the binding surface
between EndoIII and DNA were predicted in another crystal structure (PDB: 2ABK).
4
A
crystallographic study demonstrated that a high
-
potential
iron
-
sulfur protein
(HiPIP) undergoes
minor conformational changes upon toggling the redox state of the [4Fe4S] cluster between 2+
and 3+.
5
The coordinates of the atoms in the
EndoIII
protein str
ucture for the [4Fe4S]
3+
case was
therefore
assumed to be the same as for the [4Fe4S]
2+
case.
The authors
in ref (5)
suggested that
the redox
conversion
at the
[4Fe4S] metallocofactor
induces
electrostatic potential
change at
the
S
13
protein
surface.
The conce
pt of redox
-
modulated binding affinity supports our proposed
electrostatic
model based on electrostatic interactions between DNA and [4Fe4S] DNA
-
processing
proteins. The modeled
PE is ca. 7 kcal/mol, while the energy difference estimated from
electrochemical studies is ca. 5 kcal/mol.
Similar modeled
PE is also obtained for other DNA
-
processing proteins containing [4Fe4S] clusters or
flavin cofactor
s,
such as MutY, Dna2, and
photolyase.
The model
PE for MutY also recapitulates the energy diff
erence between the 2+ and
3+ states computed from the redox potential shift obtained using electrochemical techniques.
6
This
result corroborates that our electrostatic model is general and is readily applicable to other DNA
-
processing enzymes carrying [4Fe
4S] metallocofactors.
Condition
# of Short
DNA
# of
Proteins on
Short DNA
Proteins:
Short DNA
# of Long
DNA
# of
Proteins on
Long DNA
Proteins:
Long DNA
< 1
% Ox
-
MM
280
496
1.77
49
174
3.55
< 1
% Ox
-
WM
139
185
1.33
31
49
1.58
33
% Ox
-
MM
140
192
1.37
23
55
2.39
33
% Ox
-
WM
372
428
1.15
69
100
1.45
66
% Ox
-
MM
352
594
1.69
49
134
2.7
3
66
% Ox
-
WM
375
473
1.26
76
129
1.70
> 99
%
Ox
-
MM
169
210
1.24
25
39
1.56
> 99
% Ox
-
WM
108
137
1.27
22
30
1.36
Table
S
2
. Number of DNA
and proteins counted in the AFM redistribution assay using Endo
III
oxidized
to various
extent.
S
14
Condition
r
dr
< 1
% Ox
-
MM
2.00
1.69
< 1
% Ox
-
WM
1.19
33
% Ox
-
MM
1.74
1.38
33
% Ox
-
WM
1.26
66
% Ox
-
MM
1.62
1.20
66
% Ox
-
WM
1.35
> 99
%
Ox
-
MM
1.26
1.17
> 99
% Ox
-
WM
1.07
Table
S
3
. Results of the AFM redistribution assay using Endo
III
oxidized to various extent
.
Table
S
2 shows the raw counting data. Table
S
3 utilizes the data in Table
S
2 and tabulates the
calculated
mismatched to well
-
matched differential as a function of the oxidation extent of Endo
III
.
The results show that as Endo
III
becomes more oxidized, the ability to differentiate between
WM
and MM
strands decreases (Figure
4
).
S
15
Figure
S9
.
Cyclic voltammetry
of DinG incubated with a
DNA
-
modified electrode.
CV
s
in
p
rotein
buffer (20 mM NaH
2
PO
4
pH 7.5, 100
mM NaCl, 5% glycerol, 1 mM EDTA, black line)
with
2
M
DinG
added (red
line
) at a s
can rate of 100 mV/s.
Figure
s
S9
-
10
displays CVs
, SWVs, and DPVs
of a DNA
-
modified Au electrode in
protein
buffer with and without DinG in solution. DinG exhibits a
reversible
redox wave with a midpoint
potential of about
+80 mV vs. NHE
,
a value
similar to that observed previously by our group.
7
,
8
O
nly one redox peak is observed in the anodic and cathodic scans,
indicating
that only one
electrochemical process is occurring.
S
16
Figure S
10
.
Square wave and differential pulse voltammetry of DinG incubated on a DNA
-
modified electrode.
(a) Cathodic
SWVs,
(b) anodic SWVs
, and (c) c
athodic and anodic
DPVs in
p
rotein buffer (20 mM NaH
2
PO
4
pH 7.5, 100 mM NaCl, 5% glycerol, 1 mM EDTA)
with 2
M
DinG added
.
(c)
(a)
(b)
S
17
Figure
S
11
.
Cyclic voltammetry of DinG incubated with a
DNA
-
modified electrode
at different
scan rates.
CVs
in
2
M DinG
in p
rotein buffer (20 mM NaH
2
PO
4
pH 7.5, 100 mM NaCl, 5%
glycerol, 1 mM EDTA)
at a scan rate of 25 (black), 50 (red), 100 (blue), 200 (green), 400 (orange),
and 800 (purple) mV/s.
Figure
S11
shows CVs of DinG at
a constant concentration but at varying scan ra
te.
Since the
contribution of the capacitance to the total current is significant, the Faradaic component is
extracted from the total peak current by removing the charging current prior to performing
Randles
-
Sevcik analysis.
S
18
Figure
S12
. Randles
-
Sevcik plots of
DinG electrochemistry. T
he
background
-
subtracted
peak
currents of 2
M DinG at various scan rates recorded in Figure
S11
were plotted
versus (a) the
square root of the scan rate,
and (b) the scan rate.
Figure S12a
shows the Randles
-
Sevcik plot of DinG with the capacitance component
of the
peak current removed
versus the square root of the scan rate. A linear correlation is observed for
both the cathodic and anodic adjusted peak currents of DinG with the square root of the scan rate,
indicating that the redox event involves a diffusive species.
By contrast
, Figu
re
S12b shows a
Randles
-
Sevcik plot of DinG with the charging
-
current
-
free peak current versus the scan rate. A
(a)
(b)
S
19
poor linear r
elationship is observed, corroborating that the redox
event does not originate from an
adsorbed
species and that DinG does not
asso
ciate
very
strongly to DNA
under
this condition.
Guided by the CV obtained in Figure
S9
and the knowledge gained from the electrochemical
characterization of Endo
III and DinG
,
we
conducted
bulk electrolysis on
DinG
at a constant
oxidizing potential (+250
mV vs
. NHE) and reducing potential (
50mV vs. NHE) for 25 min.
These
potentials are chosen based on the positions of the anodic and cathodic peaks in Figure S9 in order
to provide a thermodynamic driving force to generate cluster proteins in the [4Fe4S]
3
+
and
[4Fe4S]
2
+
states, respectively.
S
20
Figure S
1
3
. Bulk electrolysis
of DinG using a DNA
-
modified electrode. Bulk electrolysis
in
p
rotein buffer (20 mM NaH
2
PO
4
pH 7.5, 100
mM NaCl, 5% glycerol, 1 mM EDTA, black lines)
with 2
M DinG added (
blue
lines
) held at (a)
+2
50 mV and (b)
50 mV
vs. NHE
.
Upon application of a positive potential in the presence of DinG, the magnitude of current
recorded is larger than the buffer only case without DinG (Figure
S13
a, black and blue lines).
Interestingly, the
magnitude of current is very similar for both cases with and without DinG when
(a)
(b)
S
21
a negative pot
ential bias is applied (Figure S13
b, black and blue lines).
These
results suggest that
native DinG likely
contains a cluster that
exists primarily in the reduced
[4Fe4S]
2+
form. DinG is
oxidized at
+2
50 mV and reduced at
50 mV
vs. NHE
for all subsequent AFM studies.
Condition
# of Short
DNA
# of
Proteins on
Short DNA
Proteins:
Short DNA
# of Long
DNA
# of
Proteins on
Long DNA
Proteins:
Long DNA
< 1
% Ox
-
MM
1034
1003
0.97
50
86
1.72
< 1
% Ox
-
WM
781
765
0.98
47
50
1.06
> 99
%
Ox
-
MM
719
637
0.8
9
48
81
1.69
> 99
% Ox
-
WM
742
1077
1.45
4
9
126
2
.57
Table
S4
. Number of DNA and proteins counted in the AFM redistribution assay using
reduced
and oxidized
DinG.
Condition
r
dr
< 1
% Ox
-
MM
1.77
1.63
< 1
% Ox
-
WM
1.09
> 99
%
Ox
-
MM
1.9
0
1.0
8
> 99
% Ox
-
WM
1.77
Table
S5
. Results of the AFM redistribution assay using
reduced and oxidized DinG.
S
22
Condition
# of Short
DNA
# of
Proteins on
Short DNA
Proteins:
Short DNA
# of Long
DNA
# of
Proteins on
Long DNA
Proteins:
Long DNA
33
% Ox
-
MM
817
801
0.98
53
100
1.8
9
33
% Ox
-
WM
791
790
1.00
54
73
1.35
66
%
Ox
-
MM
542
471
0.87
63
104
1.65
66
% Ox
-
WM
775
865
1.12
49
80
1.6
3
Table
S6
. Number of DNA and proteins counted in the AFM redistribution assay using
33% and
66% oxidized samples which were prepared by mixing reduced DinG and oxidized EndoIII in a
2:1 and 1:2 ratios, respectively.
Condition
r
dr
33
% Ox
-
MM
1.92
1.42
33
% Ox
-
WM
1.35
66
%
Ox
-
MM
1.90
1.30
66
% Ox
-
WM
1.4
6
Table
S7
. Results of the AFM redistribution assay using
33% and 66% oxidized samples which
were prepared by mixing reduced DinG and oxidized EndoIII in a 2:1 and 1:2 ratios, respectively.
S
23
Condition
# of Short
DNA
# of
Proteins on
Short DNA
Proteins:
Short DNA
# of Long
DNA
# of
Proteins on
Long DNA
Proteins:
Long DNA
33
% Ox
-
MM
797
762
0.9
6
51
105
2.06
33
% Ox
-
WM
728
701
0.96
49
75
1.53
66
%
Ox
-
MM
647
747
1.15
43
104
2.42
66
% Ox
-
WM
470
936
1.99
46
158
3.43
Table
S8
. Number of DNA and proteins counted in the AFM redistribution assay using
33 % and
66 % oxidized samples which were prepared by mixing reduced EndoIII and oxidized DinG in a
2:1 and 1:2 ratios, respectively.
Condition
r
dr
33
% Ox
-
MM
2.15
1.35
33
% Ox
-
WM
1.5
9
66
%
Ox
-
MM
2.09
1.21
66
% Ox
-
WM
1.72
Table
S9
. Results of the AFM redistribution assay using
33 % and 66 % oxidized samples which
were prepared by mixing reduced EndoIII and oxidized DinG
in a 2:1 and 1:2 ratios, respectively.