of 23
1
Supplemental Text and Figures
E
xtracellular DNA
promotes
efficient extracellular electron transfer
by pyocyanin in
Pseudomonas aeruginosa
biofilms.
Authors
:
Scott H. Saunders
1
, Edmund C.M. Tse
2
, Matthew D. Yates
3
, Fernanda Jiménez
Otero
4
, Scott
A.
Trammell
3
, Eric D.A. Stemp
5
, Jacqueline K. Barton
2
*, Leonard M. Tender
3
* and Dianne K.
Newman
1
,6
*
2
Supplementary Text
DNA modified electrode controls
The purpose of carrying out the control experiments shown in
Fig. S4
using electrodes
modified using self
-
assembled monolayers (SAMs) of double
-
stranded DNA (
ds DNA
) was to
probe the electron transfer mechanism between the redox probe and the electrode surface. Four
sets of control experiments were conducted to probe whether DNA
-
mediated charge transport
(DNA
CT) occurs between PCN and the electrode via the base stack of
ds DNA
: (i) WM
-
vs.
MM
-
DNA, (ii) high
-
vs. low
-
density monolayers, (iii) PCN vs. RuHex, and (iv) scan rate
dependence studies.
(i) WM
-
vs. MM
-
DNA: Electrodes modified using SAMs of
ds DNA
tha
t are well
-
matched
(WM) or
contain a single base pair mismatch (MM) were prepared under identical conditions.
WM
-
DNA
electrodes and MM
-
DNA electrodes have similar physical properties but exhibit
different charge
transfer properties that occur through the
aromatic base stacks of DNA. If the
electron transfer process
between the redox probe and the electrode occurs via DNA CT, then a
decrease in the yield of DNA CT
is expected due to the perturbation to the base stack introduced
by the MM lesion. If the elec
tron transfer
process occurs via other modes such as physical
diffusion of a charge carrier or physical contact
between the redox probe and the electrode
surface, then no difference in the current measured would be
expected.
(ii) High
-
vs. low
-
density mon
olayers: Electrodes modified using high
-
and low
-
density SAMs
of
ds DNA
were used to probe the effect of the number of
ds DNA
on the charge transport
mechanism. A low
-
density
ds DNA
monolayer promotes electron transport within one
ds DNA
,
3
but
ds DNA
may adopt various conformations because the individual
ds DNA
are more distantly
spaced. A high
-
density
ds DNA
monolayer encourages
ds DNA
to align with each other in a more
orderly fashion due to more
ds DNA
present in a confined space. Regardless of a h
igh
-
or low
-
density monolayer, for the case of PCN
-
modified DNA
-
SAM, MM discrimination was observed,
suggesting that DNA CT is the major mode of electron transport between the electrode and the
redox probe.
(iii) PCN vs. RuHex: RuHex is a positively
-
char
ged species that binds electrostatically to the
negatively
-
charged phosphate backbone. RuHex also associates to the OH groups at the terminus
of the SAM surface passivated by backfilling using mercaptohexanol. Electrons tunnel through
the SAM from the elec
trode to the RuHex situated at the SAM terminus. Electrons can then be
transferred and be distributed among the Ruhex bound to the SAM surface and on the DNA
phosphate backbone. Residual RuHex present in solution could help with ET through hopping
and diff
usion in solution, and the contribution of solution
-
based ET to the current measured
depends on the concentration of RuHex used. Therefore a number of RuHex concentrations were
screened in order to optimize the conditions for this measurement. For species
that do not
participate in DNA CT, no MM discrimination will be observed. The RuHex experiments show
that the number of
ds DNA
on the SAM
-
modified electrodes is similar for both the WM and MM
cases. The observation of a sizable MM discrimination demonstrat
es that PCN bound on DNA
participates in DNA CT.
(iv) Scan rate dependence studies (10
-
5000 mV/s) were carried out to determine whether the
phenazine moiety was covalently attached to the DNA
-
modified electrodes. A linear relationship
4
between scan rates a
nd measured peak currents signified that phenazine was covalently attached.
D
ap
measurement theory
To measure
D
ap
, we compared
two
electrochemical measurements that depend on D
ap
in different ways.
By performing this comparison at multiple concentrations,
we could fit the data points to a line whose
slope can be defined by known parameters
, yielding
D
ap
.
D
phz
*
b
iofilms were soaked in 75
μ
M of PYO
and then transferred to fresh medium
lacking PYO
. The biofilm PYO concentration dropped over the
course of 45 min as equilibration with the medium occurred. Approximately 15 sets of scans were taken
during this time period.
The first measurement was square wave voltammetry (SWV
) (Fig. 5B).
The SWV peak current
(I
swv
)
is
defined in terms of concentration
of redox molecules (
C
) reacting directly (by physical
diffusion) or indirectly (through electron self
-
exchange reactions) with the electrode (
D
ap
)
, the area of
the electrode (
A
), the number
of electrons per redox reaction (
n
),
the F
araday constant (
F
), the pulse
width
(
t
p
) and a normalization constant (
). Not
e
that I
swv
depends on the square root of D
ap
.
#$%
=
푛퐹퐴퐶
,
./
,
/
(
푒푞
.
1
)
The second
measurement
was a generator
-
collector (GC)
measurement (Figure 5C).
The maximum
GC
current (I
gc
)
is
also
defined in terms of
C
, D
ap
,
n
, and
F
, but depends on a geometric factor (
S
)
(Boyd et
al., 2015)
as opposed to the electrode area. Not
e
that I
gc
depends directly on D
ap
.
89
=
푛퐹푆퐶
./
(
푒푞
.
2
)
5
A plot of experimentally determined
I
gc
vs. I
swv
values y
ields linear
relationships
for PYO in
biofilms
and in solution
(i.e.,
blank IDA) (Fig.
5D, S6), with
slope (
m
)
tha
t can be defined in terms I
gc
and I
swv
:
=
89
#$%
=
푛퐹푆퐶
./
푛퐹퐴퐶
,
./
,
/
(
푒푞
.
3
)
=
,
/
./
퐴휑
(
푒푞
.
4
)
This relationship enables determination of D
ap
from the experimentally deter
mined dependency of I
gc
to
I
swv
(i.e.,
m
) in terms of
known
experimental
parameters
(Fig. 5F)
.
Importantly, it provides a means of
determining D
ap
that is not dependent on knowing C, which for our system is unknown and changes over
time as PYO diffuses out of the biofilm.
./
=
(
푚퐴휑
)
?
?
/
(
푒푞
.
5
)
Note that because the slope of these plots is linear, it suggests that D
ap
is
constant
for a biofilm despite
how the concentration of PYO changes
as it diffuses.
We assume a constant D
ap
by fitting the data with
a line
ar model
.
D
loss
measurement theory
In order to
measure the physical diffusion coefficient of PYO molecules
lost from the
I
DA
-
grown biofilms, we
monitored the l
oss of
biofilm
PYO
over time
as it equilibrat
ed
with the
fresh
6
medium similar to an approach taken with polymer films
(White et al., 1982a)
.
To
quantify this
process,
we
used
successive SWV scans
acquired over 45min following transfer to fresh medium (the SWV
subset of the D
ap
data).
A 1
-
dimensional diffusion model was then applied to fit the decay of I
swv
yielding an estimate of
D
loss
.
C
onsidering
1
-
dimensional
diffusion of an initial mass
(
M
0
)
of a substance from a point source,
the
solution
of
Fick’s second law describing the
time
-
dependent
concentration gradient
is given by eq. 6
where
x
is distance nor
mal from the source,
A
is the cross
-
sectional area in 3D space, and
D
loss
is the
physical diffusion coefficient.
(
,
)
=
D
,
4
EF##
G
H
I
J
K
LMNN
O
(
푒푞
.
6
)
When the source is located at no
-
flux boundary such that the mass diffuses only to one side,
(
=
0
,
)
=
2
D
,
4
EF##
(
푒푞
.
7
)
w
here
C(x = 0,
t)
is the time dependent concentration at the surface of the no
-
flux boundary (e.g. the
IDA surfac
e).
To connect this model to the data, the concentration and initial mass can be defined in terms of the
measured SWV current (I
swv
). For SWV, the concentration,
C
, is given by:
7
(
)
=
#$%
(
)
,
/
푛퐹퐴
,
./
(
푒푞
.
8
)
Concentration can be defined as the mass per volume, so the initial mass,
M
0
, can be expressed in terms
of the effective volume probed by the electrode (
V
):
D
=
D
,
/
푛퐹퐴
,
./
(
푒푞
.
9
)
The initial current is defined as I
swv
(t = 0) = I
0
, which is experimentally estimated from the last I
swv
in
the PYO soak, before transfer to PYO
-
free medium. This is a conservative overestimate, because some
of the soak signal comes from the solution PYO.
Substituting the values for
C
(
eq. 8
) and
M
0
(
eq.
9
) into
equation
7
yields:
#$%
(
)
=
2
D
,
4
EF##
(
푒푞
.
10
)
The term
V
refers to the biofilm volume from which the mass of PYO was detected by the electrode. For
a 1D electrode process, the concentration gradient extends from the electro
de
-
solution interface (
C = 0
)
to the edge of the diffusion layer,
(
C = C
bulk
)
in a near linear fashion
(Bard et al., 1980)
. There is no
region where all of the mass is detected, but the electrode has detected one half of the mass in the
volume
×
, therefore the effective
V
can be defined:
=
퐴훿
2
(
푒푞
.
11
)
8
The diffusion layer,
, for a single potential step can be estimated by:
=
2
,
./
#
(
푒푞
.
12
)
where
t
s
is the amount of time that the driving potential is held. SWV is a se
ries of forward and reverse
potential steps for which we could not unequivocally define an equivalent t
s
value, however, we discuss
reasonable bounds in the assumptions section below.
Substituting into equation
11
yields
=
,
./
#
,
therefore equation
10 can be written as:
#$%
(
)
=
D
,
./
#
,
EF##
(
푒푞
.
13
)
And with that expression we can fit the decay of I
swv
(t) to a model of the form:
=
+
(
푒푞
.
14
)
w
here the coefficient,
b
, is described in known variables except for
D
loss
and
where
a
accounts for any
constant background signal.
=
D
,
./
#
,
EF##
(
푒푞
.
15
)
9
Therefore
,
D
loss
can be calculated from the fit
data as (Fig. 5F, S
7
):
EF##
=
D
?
./
#
?
(
푒푞
.
16
)
Model assumptions
The
D
loss
analysis described above is based on a number of assumptions. It assumes, for example,
that the biofilms are homogeneous. In reality, they are heterogeneous, containing many voids and
obstacles (e.g., cells
and exopolysaccharides) through which diffusion would not occur. We contend
heterogeneity would affect
D
loss
and D
ap
in a similar manner, as it reduces the biofilm volume in which
PYO resides. As such, we don’t expect biofilm heterogeneity to greatly affec
t the determination of
D
loss
from D
ap
. Importantly, our analysis assumes a single infinitely thick phase described by a single
D
loss
. In
reality, there are at least two phases, a thin biofilm adjoining an infinitely thick solution. If
D
loss
in
solution is
greater than in the biofilm, then at any instance the concentration gradient of PYO across the
biofilm will be steeper th
a
n predicted by the model.
S
ince the flux of PYO out of the biofilm at any
instance is proportional to the product of the gradient and
D
loss
,
a
s the model fits the rate of change of
PYO in the biofilm the assumption of a shallower gradient than the actual gradient is expected to result
in a calculated
D
loss
that is higher than the actual
D
loss
.
To estimate the scan time parameter, t
s
, we assumed that for the blank IDA
D
ap
=
D
loss
and solved
for t
s
that best fit the I
swv
decay for the blank IDA. We then used this value,
t
s
= 21ms
, to calculate
D
loss
for the biofilm IDAs. The scan time parameter is intended to estimate the thickness of
the diffusion layer
that is formed during a single potential step that drives the flux of the electrode reactant resulting in the
observed current. SWV is, however, a series of forward and reverse potential steps superimposed on a
10
series of forward potenti
al steps. Therefore, determining the effective scan time that describes the
change in thickness of the diffusion layer that occurs during the forward pulse contributing to I
swv
is
nontrivial (we only consider the forward pulse as the reverse scan partially
replenishes the electrode
reactant that is depleted during the forward scan). One approach to estimate t
s
is setting the SWV
expression equal to twice the Cottrell equation (eqs. 17 & 18), since I
swv
is the net current from the
forward and reverse pulses.
For a pulse
amplitude
of 0.025 V
and step increment of 0.001 V as used
here,
each potential step is
effectively
0.05 V
. The
potential at which the
I
swv
occurs
is the
formal
potential of the electrode reactant
and a
pplying the Nernst equation, the fraction
of electrode reactant in
the oxidized state at the electrode surface at the start of the
forward
potential step
generating I
swv
(E =
Eo'+ 0.025 V) is
82
.
6
% and at the end (E = Eo
-
0.025 V) is
17
.
4
%.
The
Cottrell
equation
assumes
the
potential step
drives a redox reaction in which 100% of the electrode accessible redox molecules at the
electrode go from oxidized to reduced (or vice versa)
.
R
eplac
ing
C in the Co
t
trell eq
uation
with 0.
655
x
C
to reflect the
fraction of
PYO
at the electrode surface that
changes oxidation state during the forward
potential step
of the SWV yields an estimate t
s
»
6ms.
#$%
=
푛퐹퐴퐶
,
./
,
/
=
2
/#
=
푛퐹퐴
0
.
655
×
,
./
,
#
(
푒푞
.
17
)
#
=
/
(
2
×
0
.
655
)
?
?
(
푒푞
.
18
)
As such, our t
s
estimate (21
ms) based on the blank D
ap
= D
loss
used to estimate D
loss
for the biofilm may
be an overestimate.
Noting that
D
loss
scales linearly with t
s
(eq. 16), this would conservatively
underestimate the difference between biofilm D
ap
and
D
loss
.
11
Parameters for
electrochemical calculations
The surface area of the electrode for SWV,
A
= 0.025cm
2
, and the geometric factor for GC,
S
=
18.4cm, were calculated for a blank IDA using the known concentration and D
ap
for ferrocene methanol
(Boyd et al., 2015)
. All quantified SWVs were acquired with a pulse amplitude of 25mV at a frequency
of 300Hz and an increment of 1mV. The SWV pulse time,
t
p
, is one half the square wave period (
½
*
1/300
= 1/600 sec). Peak separation from CV of PYO in solution indicated that it did not undergo the
full 2 electron reduction, but on average underwent electron transfer with
n
»
1.8. From these acquisition
parameters,
j
= 0.7 was inferred from a table of exis
tin
g values
(Lovri
ć
, 2010)
.
For the
D
loss
estimate, it
was assumed that
I
0
was the soak SWV peak current. The equivalent scan time,
t
s
, for the
D
loss
calculation is discussed above.
12
Supplemental Figures
Figure S1.
Colony biofilms images and controls. (A) Representative images of colony biofilms formed by WT
-
DKN, WT
-
MRP,
D
pelB
, and
D
phz
and
D
phz
* grown with each phenazine. (B) SEM image of cells at the
top and bottom (attached to
the
0.2
μ
m membrane
) of the colony biofilm.
(C) LC
-
MS quantification of phenazines from colony biofilms grown without an
underlying membrane. (D) Comparison of phenazines from WT colonies that were lysed with sonication or not. Statistical
test was a Welch’s single tailed t
-
test and the star denotes p<0.05. (E) Accumulated phenazine from three
D
phz
* colony
biofilms following three days of growth with synthetic phenazine (Day 3), and one day later after transfer to fresh agar (Day
4). PCA was not detected on Day 4. Same data
are shown in Fig. 1 H.