Article
Extracellular DNA Promotes Efficient Extracellular
Electron Transfer by Pyocyanin in
Pseudomonas
aeruginosa
Biofilms
Graphical Abstract
Highlights
d
PYO and PCN bind extracellular DNA, which facilitates their
retention in biofilms
d
Electrode biofilms support fast PYO electron transfer and
slow PYO loss
d
Phenazines rapidly exchange electrons and are capable of
DNA charge transfer
in vitro
Authors
Scott H. Saunders, Edmund C.M. Tse,
Matthew D. Yates, ...,
Jacqueline K. Barton, Leonard M. Tender,
Dianne K. Newman
Correspondence
jkbarton@caltech.edu (J.K.B.),
leonard.tender@nrl.navy.mil (L.M.T.),
dkn@caltech.edu (D.K.N.)
In Brief
Phenazines are retained in biofilms
through binding to extracellular DNA, and
together these biofilm components
mediate efficient extracellular electron
transfer to support bacterial metabolism
Saunders et al., 2020, Cell
182
, 919–932
August 20, 2020
ª
2020 Elsevier Inc.
https://doi.org/10.1016/j.cell.2020.07.006
ll
Article
Extracellular DNA Promotes Efficient
Extracellular Electron Transfer by Pyocyanin
in
Pseudomonas aeruginosa
Biofilms
Scott H. Saunders,
1
Edmund C.M. Tse,
2,3
Matthew D. Yates,
4
Fernanda Jime
́
nez Otero,
5
Scott A. Trammell,
4
Eric D.A. Stemp,
6
Jacqueline K. Barton,
2,
*
Leonard M. Tender,
4,
*
and Dianne K. Newman
1,7,8,
*
1
Division of Biology and Biological Engineering, Caltech, Pasadena, CA, USA
2
Division of Chemistry and Chemical Engineering, Caltech, Pasadena, CA, USA
3
Department of Chemistry, University of Hong Kong, Hong Kong SAR, China
4
Center for Bio/Molecular Science and Engineering, Naval Research Laboratory, Washington, DC, USA
5
George Mason University, Fairfax, VA, USA
6
Department of Physical Sciences, Mt. Saint Mary’s University, Los Angeles, CA, USA
7
Division of Geological and Planetary Sciences, Caltech, Pasadena, CA, USA
8
Lead Contact
*Correspondence:
jkbarton@caltech.edu
(J.K.B.),
leonard.tender@nrl.navy.mil
(L.M.T.),
dkn@caltech.edu
(D.K.N.)
https://doi.org/10.1016/j.cell.2020.07.006
SUMMARY
Redox cycling of extracellular electron shuttles can enable the metabolic activity of subpopulations within
multicellular bacterial biofilms that lack direct access to electron acceptors or donors. How these shuttles
catalyze extracellular electron transfer (EET) within biofilms without being lost to the environment has
been a long-standing question. Here, we show that phenazines mediate efficient EET through interactions
with extracellular DNA (eDNA) in
Pseudomonas aeruginosa
biofilms. Retention of pyocyanin (PYO) and phen-
azine carboxamide in the biofilm matrix is facilitated by eDNA binding.
In vitro
, different phenazines can ex-
change electrons in the presence or absence of DNA and can participate directly in redox reactions through
DNA.
In vivo
, biofilm eDNA can also support rapid electron transfer between redox active intercalators.
Together, these results establish that PYO:eDNA interactions support an efficient redox cycle with rapid
EET that is faster than the rate of PYO loss from the biofilm.
INTRODUCTION
Microbial biofilms are ubiquitous in natural and engineered con-
texts, spanning plant roots to chronic human infections to anaer-
obic digestors (
Hall-Stoodley et al., 2004
;
Pandit et al., 2020
). As
biofilms develop, metabolic stratification occurs, driven by steep
concentration gradients of substrates, such as oxygen, that are
consumed by cells at the biofilm periphery faster than the sub-
strates can diffuse into the biofilm interior (
Stewart 2003
;
Stewart
and Franklin, 2008
;
Liu et al., 2015
). Indeed, oxidant limitation is a
generic challenge for cells that inhabit biofilm microenviron-
ments where electron donors are abundant, yet electron accep-
tors are not. One widespread strategy microbes employ to over-
come this challenge is to channel electrons derived from
intracellular metabolism to extracellular oxidants at a distance
(
Shi et al., 2016
). Known as ‘‘extracellular electron transfer’’
(EET), this process requires electron carriers to bridge the gap,
be they outer membrane-associated or extracellular cyto-
chromes (
Richter et al., 2009
;
Xu et al., 2018
;
Jime
́
nez Otero
et al., 2018
;
Nevin et al., 2009
), various structures called ‘‘nano-
wires’’ (
Steidl et al., 2016
;
Subramanian et al., 2018
;
Wang et al.,
2019
;
Reguera et al., 2005
;
Malvankar et al., 2011
), cable bacte-
ria conductive filaments (
Cornelissen et al., 2018
), or redox-
active small molecules (
Glasser et al., 2017a
). Although the
putative molecular components underpinning different EET pro-
cesses have been described in a variety of organisms, a detailed
understanding of how these components achieve EET remains
an important research goal across diverse systems.
In contrast to the rapid progress in understanding the molec-
ular basis by which cytochromes and nanowires facilitate EET,
less progress has been made on how soluble (physically diffu-
sive) electron shuttles facilitate EET beyond interactions at the
cell surface (
Light et al., 2018
;
Xu et al., 2016
;
Marsili et al.,
2008
). In part, this is due to the challenges involved in identifying
and studying small molecule metabolites within a complex extra-
cellular matrix. Accordingly, to study extracellular electron shut-
tling, we have chosen to work with a model system that employs
a relatively well studied and tractable set of shuttles called phen-
azines. Phenazines are colorful redox-active molecules pro-
duced by numerous microbial species including the bacterium,
Pseudomonas aeruginosa
(
Turner and Messenger, 1986
;
Chin-
cholkar and Thomashow, 2013
).
P. aeruginosa
strains are
ll
Cell
182
, 919–932, August 20, 2020
ª
2020 Elsevier Inc.
919
ubiquitous yet perhaps most well-known for their roles in chronic
infections where their growth as biofilms renders them antibiotic
tolerant and contributes to patient morbidity and mortality (
Cos-
terton et al., 1999
); importantly, phenazines support the develop-
ment of anoxic, antibiotic tolerant biofilm regions. Two dimen-
sional mapping of phenazine production in the agar underlying
colony biofilms has revealed that different phenazines localize
to distinct zones (
Bellin et al., 2014
,
2016
). Here, we set out to
better understand how phenazines facilitate EET within the
P. aeruginosa
biofilm matrix.
Intriguingly, although the
P. aeruginosa
biofilm matrix com-
prises a heterogeneous group of polymers (
Colvin et al., 2012
),
extracellular DNA (eDNA) from dead cells is a significant contrib-
utor (
Allesen-Holm et al., 2006
;
Jennings et al., 2015
;
Whitchurch
et al., 2002
), accounting for the majority of the matrix polymers in
some cases (
Steinberger and Holden, 2005
;
Matsukawa and
Greenberg, 2004
). Phenazines have long been known to interca-
late into double stranded DNA
in vitro
(
Hollstein and Van Gemert,
1971
). Recently, it was suggested that the phenazine pyocyanin
(PYO) can participate in DNA-mediated charge transfer
in vitro
(
Das et al., 2015
), and phenazine-eDNA interactions might facil-
itate biofilm EET (
Das et al., 2015
). Notably, the ability of PYO to
stimulate cell lysis (
Das and Manefield, 2012
) changes according
to the environment: when cells are oxidatively stressed (i.e.,
oxidant replete but reductant limited) and ATP limited, PYO is
toxic; whereas when they are reductively stressed (i.e., reductant
replete but oxidant limited), PYO promotes viability and biofilm
aggregate expansion (
Meirelles and Newman, 2018
;
Costa
et al., 2017
). This observation raises the intriguing possibility
that cell lysis by a small percentage of the population early on
might later promote EET once biofilms have developed anoxic
zones where extracellular electron shuttles support metabolism.
Although a variety of roles for eDNA in biofilms have been
proposed, including serving as a structural support, nutrient,
and/or genetic reservoir (
Flemming and Wingender, 2010
), to
our knowledge, that biofilm eDNA may facilitate EET has not
been demonstrated.
The current model of phenazine redox cycling in biofilms can
be broadly defined (
Figure 1
A). In anoxic regions, oxidized phen-
azines are reduced intracellularly by metabolic reactions that
support these cells (
Jo et al., 2017
;
Glasser et al., 2014
,
2017b
;
Wang et al., 2010
). These reduced phenazines are thought to
physically diffuse through the extracellular matrix toward the
Figure 1. Colony Biofilms Retain PYO and PCN
(A) Diagram of the phenazine redox cycle in a biofilm. Cells are shown as gray
rods, phenazines are shown as blue hexagons, electrons are shown as circles,
and the oxygen gradient is shown as the blue background.
(B) Structures of the three studied phenazines in their oxidized states pro-
duced by
P. aeruginosa
—phenazine carboxylate (PCA), phenazine carbox-
amide (PCN), and pyocyanin (PYO). All three phenazines (PHZ) undergo two
proton two electron reductions and the midpoint potentials are shown for the
reduction of each phenazine.
(C) Images of WT (top) and
D
phz
* (bottom) colony biofilms.
(D) Schematic of phenazine extractions from colony biofilms and agar. The
0.2
m
m membrane is shown as the dashed line.
(E) Biofilm and agar concentrations for PCA, PCN, and PYO from three WT
biofilms.
(F) The same data as (E), represented as retention ratios ([Biofilm]/[Agar]).
(G) Recovered phenazine concentrations from
D
phz
* colony biofilms grown
with different levels of synthetic phenazine in the underlying agar for 4 days.
The dashed gray line shows 1:1 (biofilm concentration:added phenazine).
(H) Accumulated phenazine from three
D
phz
* colony biofilms following 3 days
of growth with synthetic phenazine (day 3), and 1 day later after transfer to
fresh agar (day 4). Data are represented as the percentage day 4/day 3. PCA
was not detected (n.d.) on day 4.
In (E)–(H), values for individual biofilms are shown by open symbols, and lines
or bars represent the mean.
See also
Figure S1
.
ll
920
Cell
182
, 919–932, August 20, 2020
Article
oxic region where they react abiotically with molecular oxygen.
The oxidized phenazines then return to the anoxic region of the
biofilm to complete the redox cycle. Although studies have
begun to characterize the reactions on either side of the redox
cycle, very little is known about how phenazines operate in the
intervening extracellular matrix. Theoretical studies suggest
that physical diffusion of oxidized phenazine toward the biofilm
interior and reduced phenazine toward the biofilm periphery
may be fast enough to support the metabolism of the oxidant
limited cells (
Glasser et al., 2017a
;
Kempes et al., 2014
), but
these models have assumed the biofilm is a closed system in
which freely diffusing phenazines are not lost to the outer envi-
ronment. It is well established, however, that phenazines can
escape
P. aeruginosa
biofilms (
Ramos et al., 2010
), which, if
left unchecked, would greatly reduce the efficiency of phenazine
redox cycling. This is a key challenge these biofilms must solve.
This study investigates how phenazine electron transfer may
be reconciled with phenazine retention. Specifically, we ask:
are phenazines retained and if so to what extent? What mecha-
nisms of electron transfer are compatible with phenazine reten-
tion? Our motivation to answer these questions arises not only
from a desire to constrain the model of phenazine redox cycling
within
P. aeruginosa
biofilms, but more broadly, to identify a
potentially generalizable strategy for how diverse electron shut-
tles enable EET.
RESULTS
We studied three major phenazine derivatives made by
P. aeruginosa
strain UCBPP-PA14 (
Schroth et al., 2018
): phena-
zine carboxylate (PCA), phenazine carboxamide (PCN), and pyo-
cyanin (PYO); at the pH relevant for our experiments, the domi-
nant form of PCA is negatively charged whereas PCN and PYO
are uncharged (
Figure 1
B). Beyond studying wild-type (WT)-pro-
duced phenazines, we also determine the effects of individual
synthetic phenazines on a mutant that does not produce phena-
zines,
D
phz
(
D
phzA1
-
G1
,
D
phzA2
-
G2
), or on a mutant that is also
incapable of modifying PCA to make PCN and PYO,
D
phz
*
(
D
phz
,
D
phzMS
,
D
phzH
). Experiments were performed with
two different types of biofilms: macroscopic colony biofilms
grown on nutrient agar surfaces (
Figures 1
C and
S1
A) and micro-
scopic biofilms attached to the surfaces of an interdigitated
microelectrode array (IDA) suspended in liquid medium. Phena-
zine-dependent biofilm phenotypes operate similarly at both
scales (
Ramos et al., 2010
), so we selected the biofilm cultivation
method for any given experiment based on which was best
suited to answering our specific research question.
Colony Biofilms Retain PCN and PYO, but Not PCA
First, we sought to quantify phenazine retention by colony bio-
films (
Figures 1
C and 1D). We used liquid chromatography-
mass spectrometry (LC-MS) to quantify extracted endogenous
phenazines from the biofilms and compared their concentrations
to that in the underlying agar (
Figures 1
D–1F). Colony biofilms
could be cleanly separated from the agar, because they were
separated by a 0.2
m
m membrane filter, which did not affect
the results (
Figures S1
B and S1C). Overall, PCA, PCN, and
PYO concentrations varied by more than 10-fold in the biofilms
reaching concentrations of
15
m
M PCA,
400
m
M PCN, and
80
m
M PYO. Comparing biofilm to agar concentrations showed
that PCN and PYO were enriched in the biofilm 10-fold and
30-fold, respectively, while PCA reached similar concentrations
in the biofilm and the agar (
Figures 1
E and 1F). This suggested
that PCN and PYO were strongly retained by the biofilm and
PCA was not. Lysing resuspended biofilm cells by sonication
prior to phenazine quantification did not strongly affect the re-
sults (
Figure S1
D), indicating that the measured pools of phena-
zines were predominantly retained extracellularly rather than
intracellularly.
To test if differential phenazine retention requires endogenous
phenazines, we grew
D
phz
* colony biofilms with synthetic phen-
azines in the agar and quantified phenazines taken up by the bio-
film. Incubation with
R
50
m
M PYO resulted in >200
m
M PYO
accumulation in the biofilm (
Figure 1
G). PCN accumulated to a
lesser extent, and PCA biofilm uptake was minimal (<50
m
M)
even with 200
m
M added to the agar (
Figure 1
G).
D
phz
* colonies
transferred from phenazine agar to fresh agar after 3 days of
growth retained phenazines in the same pattern as the WT
over 24 h (
Figure 1
H), demonstrating that the observed phena-
zine retention does not depend on endogenous phenazine pro-
duction. WT colony biofilms exhibit relatively thick and smooth
morphologies that contain deep anoxic regions that are thought
to be supported by phenazine EET.
D
phz
* colony biofilms exhibit
different colony morphologies that are thin and highly wrinkled,
which is thought to be a physiological adaptation to maximize
surface area and oxygen penetration in the absence of phena-
zines as shown for
D
phz
(
Dietrich et al., 2013
). Notably, only in-
cubation of
D
phz
* colonies with exogenous PYO appreciably
complemented the colony wrinkling phenotype (
Figure S1
A),
demonstrating that retained PYO is used by the colony biofilm
for phenazine EET metabolism
in vivo
.
P. aeruginosa
colony bio-
films thus appear able to take up and use significant amounts of
exogenous PYO, and PCN to a lesser extent. These results pre-
dict that colony biofilms contain an extracellular component that
binds and effectively retains PYO and PCN, but not PCA.
Phenazines Differentially Bind Extracellular DNA
Because we measured considerable biofilm-retained PYO and
PCN that was insensitive to cell lysis by sonication, we hypothe-
sized these phenazines might bind an abundant component of
the biofilm extracellular matrix. The extracellular matrix in
P. aeruginosa
PA14 biofilms is known to be composed primarily
of two polymers: DNA from dead cells (eDNA) and the polysac-
charide, Pel (
Colvin et al., 2011
;
Das and Manefield, 2012
), and
recent work suggested that PYO intercalates into DNA
in vitro
(
Das et al., 2015
). To test the hypothesis that eDNA in the biofilm
matrix was responsible for binding phenazines, we quantified the
binding affinity of oxidized PCA, PCN, and PYO for a 29 base pair
double-stranded DNA (dsDNA) molecule
in vitro
using isothermal
titration calorimetry (
Figure 2
A). As expected, oxidized PCA
showed no detectable binding because it is negatively charged,
as is the phosphate backbone of DNA at pH 7. In contrast,
oxidized PCN (K
D
= 194
m
M; 95% confidence interval [CI] =
148–305
m
M) and PYO (K
D
=13
m
M; 95% CI = 6.5–49
m
M) both
bind dsDNA, and their dominant forms are neutral at pH 7 (
Costa
et al., 2017
). These results are consistent with ethidium bromide
ll
Cell
182
, 919–932, August 20, 2020
921
Article
displacement and microscale thermophoresis binding assays
(
Figures S2
A and S2B) (
Das et al., 2015
). Notably, these
in vitro
phenazine-DNA binding affinities correlate with their
in vivo
retention ratio ([biofilm]/[agar]), where PYO is retained in the bio-
film significantly more than PCN, and PCA is not retained.
Reduced PYO showed no change in endogenous fluorescence
upon addition of calf thymus DNA (
Figure S2
C), suggesting
that the DNA binding affinity of PYO is redox dependent.
To determine whether phenazine-eDNA binding occurs
in vivo
,
we treated 3-day-old WT biofilms with DNase I for 24 h. These
experiments were performed with DNase I spotted on tryptone
agar medium rather than its optimal buffer, as controls showed
that buffer alone significantly disturbed the biofilm (
Figures
S3
A–S3C). A previous study also observed that mature
P. aeruginosa
biofilms were minimally affected by DNase and
suggested that extracellular proteases may inactivate DNase
(
Whitchurch et al., 2002
). Despite a low activity for DNase under
these conditions, DNase-treated biofilms retained significantly
lower biofilm PCN and PYO concentrations than their untreated
counterparts; moreover, PCA concentration was unchanged
(
Figure 2
B).
P. aeruginosa
eDNA originates from the genomic
DNA of dead cells, is high molecular weight and bound by other
biomolecules (
Kavanaugh et al., 2019
); accordingly, complete
elimination of phenazine binding sites with DNase is not ex-
pected. We also compared the phenazine retention in the WT
to a Pel mutant (
D
pel
) and found that biofilms without Pel re-
tained significantly more PYO (
Figure 2
C). Because Pel, the
dominant exopolysaccharide in PA14 biofilms is known to bind
eDNA (
Jennings et al., 2015
), these results suggest that Pel
may partially block access to eDNA by PYO, although this re-
mains to be tested
in vitro
. Because eDNA and Pel are the pri-
mary matrix constituents of PA14 biofilms (
Colvin et al., 2011
;
Das and Manefield, 2012
), that PYO was well retained in
D
pel
biofilms indicates that eDNA is the dominant matrix component
responsible for its retention.
To further test if phenazine-eDNA binding occurs
in vivo
,we
competed phenazines against ethidium bromide, a well-known
DNA intercalator. Because PCN and PYO compete for DNA
binding sites with ethidium
in vitro
(
Figure S2
A), and ethidium
is largely excluded from cells (
Jernaes and Steen, 1994
), we
reasoned that these intercalators could compete for binding
sites in the biofilm eDNA. We grew
D
phz
* biofilms with 50
m
M
oxidized PCN or oxidized PYO and increasing amounts of
ethidium in the underlying agar.
Figure 2
D shows that increasing
concentrations of ethidium resulted in successively less PYO
accumulating in the biofilms, while PCN accumulated to a similar
Figure 2. Phenazines Interact with DNA
In Vitro
and
In Vivo
(A) Representative isothermal titration calorimetry (ITC) data for each phenazine injected into a solution of dsDNA (29 base pairs). Exothermic rea
ctions are
depicted as negative values. Integrated peak data was fit with a Bayesian model to calculate the K
d
(in base pairs DNA) with 95% confidence intervals (
Duvvuri
et al., 2018
).
(B) Biofilm phenazine concentrations for WT biofilms treated with or without DNase I in the underlying agar for 24 h (n = 3 per condition). Bars with asteri
sks denote
measurements that differ significantly (p < 0.05) by a Welch’s single-tailed t test.
(C) Phenazine retention ratios for WT and
D
pel
colony biofilms (n = 3 per condition). Statistical test same as in (B).
(D) Accumulated phenazine concentrations for
D
phz
* biofilms incubated with 50
m
M PCN or PYO and increasing concentrations of the competitive intercalator,
ethidium bromide, in the underlying agar (n = 2 per condition).
(E) eDNA quantified in six WT and
D
phz
* colony biofilms with the dye TOTO-1. Error bars show standard deviation from two technical replicates. Dashed lines
show calf thymus DNA standards, with concentrations back calculated for the biofilm volume.
In (B)–(D), values for individual biofilms are shown by point symbols, and bars represent the mean.
See also
Figures S1, S2, and S3
.
ll
922
Cell
182
, 919–932, August 20, 2020
Article
lower level in the presence of any amount of added ethidium.
These results demonstrate that more than 50% of biofilm-accu-
mulated PYO resides in binding sites that can be competitively
inhibited by ethidium, whereas more than 30% of the more
weakly bound PCN resides in binding sites that are saturated
at ethidium concentrations
R
10
m
M. Although these results
reveal eDNA binding sites for PYO and PCN, they also show frac-
tions that are ethidium insensitive. This suggests there may be
other binding sites for phenazines such as Pel-eDNA complexes,
cell surfaces, filamentous phage or extracellular proteins.
Regardless, the ability of ethidium to displace significant por-
tions of PCN and PYO indicates that these phenazines interca-
late into biofilm eDNA.
Confocal microscopy of WT and
D
phz
* colony biofilms with a
cell-impermeable dsDNA dye, TOTO-1 (
Okshevsky and Meyer,
2014
), showed abundant eDNA localized in dead cells and in be-
tween cells (
Figure S3
D). We quantified the bulk concentration of
eDNA in colony biofilms by incubating biofilm suspensions with
TOTO-1 and measuring dye fluorescence. Both WT and
D
phz
*
biofilm suspensions yielded large fluorescence values when
incubated with TOTO-1. These values fall within the range of
60–500
m
M base pairs dsDNA in the colony biofilms, when cali-
brated against standards of calf thymus DNA (
Figure 2
E). How-
ever, adding calf thymus DNA to the biofilm suspensions did
not yield the expected increase in dye fluorescence (
Figure S3
E),
which suggests that the dye may be partially inhibited by biofilm
components. Therefore, this order of magnitude estimate of bio-
film eDNA should be interpreted as a lower bound on the true
value. Given this estimate, the biofilm eDNA (>100
m
M base
pairs) is in excess of PYO (
80
m
M), but it may not be in excess
of PCN (>300
m
M). Due to its poor aqueous solubility, it is prob-
able that PCN crystallizes extracellularly at the observed biofilm
concentrations, which could lead to its measured retention (
Her-
nandez et al., 2004
). Together, our
in vivo
and
in vitro
results are
consistent with eDNA providing binding sites for oxidized PCN
and oxidized PYO in the biofilm extracellular matrix.
Constraints on Phenazine Electron Transfer
Mechanisms
In Vitro
and
In Vivo
Given that phenazines are differentially bound and retained in
biofilm eDNA, we next sought to constrain how electron transfer
might be achieved in this context. Previous research has shown
distinct localization patterns for different phenazines within bio-
films, with the lowest potential phenazines (e.g., PCA) in the inte-
rior, and the highest potential phenazine (e.g., PYO) at the oxic
periphery (
Bellin et al., 2014
,
2016
). To test whether electron
transfer could occur between these molecules in solution, we
mixed different oxidized and reduced phenazines under anoxic
conditions and monitored the absorbance spectra before and af-
ter mixing (
Figures 3
A and 3B). Because PYO exhibited the
largest changes in absorbance upon reduction, we monitored
different mixtures of PYO with PCA or PCN at 690 nm (unique
PYO absorbance maximum) starting 1 min after mixing, at which
point equilibrium had been achieved. Reactions proceeded as
expected from the redox potentials of the phenazines, where
PYO was almost completely reduced by the lower potential
PCA and PCN, and reduction of PCA and PCN by the higher po-
tential PYO was minimal (
Figures 3
B and
S4
A). In addition to es-
tablishing that electron transfer can occur between different
phenazines, given their similar structures, these results suggest
that electron transfer between unbound (physically diffusing) like
phenazines (e.g., PYO-PYO electron self-exchange) may also
contribute to EET. Moreover, reactions between reduced PCA
or PCN and oxidized PYO proceeded faster than oxidation of
any of these phenazines by molecular oxygen (
Figure 3
C),
because PYO
red
appeared in these reactions immediately upon
the addition of PCA
red
or PCN
red
in the presence of O
2
, with sub-
sequent oxidation of PYO
red
by O
2
proceeding more slowly. We
next wondered whether the presence of DNA would affect the
extent of PYO reduction. PCA or PCN fully reduce PYO in the
presence of DNA (
Figure 3
B). Because PCA does not bind
DNA, this result suggests electron transfer is occurring in solu-
tion between PCA in solution and unbound PYO. For PCN and
PYO that both bind DNA, it is also possible that electron transfer
is achieved by their unbound counterparts in solution. However,
it has long been known that DNA can facilitate electron transfer
between bound redox molecules (
Genereux and Barton, 2010
),
motivating us to test whether such a process could also occur
within our
P. aeruginosa
biofilms.
DNA facilitates charge transfer through the
p
-stacked base
pairs (
Genereux and Barton, 2010
), and recent studies have
shown that DNA charge transfer can occur over kilobase dis-
tances (
Tse et al., 2019
). Using single stranded DNA (ssDNA)
modified electrodes and hybridizing complementary DNA, a pre-
vious study suggested that PYO might be able to participate in
electron transfer via DNA (
Das et al., 2015
). However, this spe-
cific method likely produces heterogeneous ssDNA/dsDNA
monolayers, and the use of PYO in solution makes possible
direct electron transfer between PYO and the electrode, con-
founding DNA charge transfer signals. To avoid background
phenazine-electrode reactions, we revisited these experiments
with a carefully controlled dsDNA modified electrode and a
covalently tethered phenazine. Because it was not possible to
directly synthesize a tethered PYO derivative, we tethered
PCN, the other DNA binding phenazine, to a DNA strand (17
base pairs) through a flexible alkane linker (see
STAR Methods
).
We chose to work with the PCN-like tethered phenazine because
it was the simplest synthetic route and nearest possible repre-
sentation of a
P. aeruginosa
phenazine. To form dsDNA, the
complementary strand containing a thiol linker (17 base oligo)
was annealed to the PCN strand. The gold electrode was modi-
fied with the thiolated dsDNA according to standard protocols
(
Kelley et al., 1997
;
Slinker et al., 2010
,
2011
), yielding a packed
DNA monolayer where the distal end of each dsDNA molecule
contained a tethered PCN that can be either bound (intercalated
within dsDNA) or unbound while covalently attached (
Figure 3
D).
The efficiency of DNA charge transfer depends upon the integ-
rity of the
p
-stacking of base pairs within the dsDNA (
Genereux
and Barton, 2010
). Mismatched DNA base pairs stack less effi-
ciently, but otherwise do not affect DNA structure, thus providing
a convenient perturbation of DNA CT. Therefore, we used cyclic
voltammetry to compare electron transfer between PCN and
the electrode through well-matched dsDNA monolayers to
dsDNAs containing a single base mismatch (
Figure 3
D). We uti-
lized multiplexed DNA chips to facilitate replicate comparisons
between well-matched and mismatched DNA monolayers
ll
Cell
182
, 919–932, August 20, 2020
923
Article
(
Figures S4
B–S4D); measurements with a non-intercalating con-
trol probe showed that these different monolayers had very
similar surface coverages (
Figures S4
E and S4F) (
Slinker et al.,
2010
).
Figure 3
E shows that the mismatched construct yielded
diminished current in the phenazine redox peak, consistent
with the charge transfer being DNA-mediated; the presence of
Figure 3. Inter-Phenazine Electron Transfer and DNA Charge Transfer
(A) Diagram showing an electron transfer reaction in solution between a reduced phenazine and oxidized PYO and between reduced PYO and molecular oxyg
en.
(B) Reaction progress after 1 min measured at 690 nm for mixtures of phenazines shown in (A), compared to oxidized and reduced PYO alone. Each reaction w
as
performed in the presence and absence of calf thymus DNA. For each condition n = 3 and error bars are one standard deviation.
(C) PYO oxidation state measured at 690 nm over time (diagnostic for oxidized PYO) for different reactions in the presence of oxygen. Points are indivi
dual
measurements, lines are loess smoothed for each set of triplicate measurements.
(D) Schematic showing a DNA modified electrode with tethered PCN (green oval) and the expected electron transfer for well-matched duplexes (blue arro
w and
base pair) and duplexes containing a mismatch (orange arrow and base pair). Mismatched bases are less likely to be in a well stacked position, which is n
ecessary
for electron transfer through the DNA pi-stack. PCN is shown both intercalated into and outside of DNA to convey its reversible DNA binding.
(E) Representative cyclic voltammetry of the well-matched (wmDNA) and mismatched (mmDNA) constructs shown in (D) under anoxic conditions, acquire
dat
100 mV/s.
(F) Representative cyclic voltammetry of the well matched, mismatched, or no phenazine constructs under the aerobic conditions described in (D), ac
quired at
100 mV/s.
(G) Diagram of time resolved spectroscopy of the photoexcited electron donor Ru(phen)
2
dppz
2+
quenched by Rh(phi)
2
bpy
3+
with biofilm eDNA.
(H) Comparison of Ru(phen)
2
dppz
2+
fluorescence in the presence of a concentrated liquid
P. aeruginosa
culture and a resuspended biofilm containing eDNA.
Gray lines show background biological fluorescence before Ru(phen)
2
dppz
2+
was added. The color map is the same as (I).
(I) The background subtracted data from the biofilm panel of (H). The amount of Rh(phi)
2
bpy
3+
is color coded as quencher equivalents relative to Ru(phen)
2
dppz
2+
.
Dots are raw data, lines are fit bi-exponential decays.
See also
Figure S4
and
STAR Methods
.
ll
924
Cell
182
, 919–932, August 20, 2020
Article
the intervening mismatch inhibits DNA charge transfer. This
mismatch effect was consistent across replicate low density
(32%–54% decrease) and high density (36%–69% decrease)
DNA monolayers (
Figure S4
C;
STAR Methods
, ‘‘DNA modified
electrode controls’’). These results displayed mismatch attenua-
tion similar to that observed for well-characterized small mole-
cules shown to stack with the DNA duplex and carry out DNA-
mediated charge transfer (
Slinker et al., 2011
). Strikingly, in the
presence of oxygen, a classic voltammetric signal characteristic
of electrocatalysis was obtained (
Figure 3
F) centered on the
phenazine redox peak. Hence, the DNA-tethered phenazine is
able to accept electrons from the electrode through the DNA
p
-stack and then reduce oxygen in a catalytic fashion. Together,
these results demonstrate that a
P. aeruginosa
phenazine can
participate in DNA charge transfer
in vitro
.
To test if biofilm eDNA could support DNA charge transfer, we
incubated
D
phz
* colony biofilms suspended in PBS with well-
characterized intercalators that can perform DNA charge trans-
fer reactions, which can be monitored by time-resolved spec-
troscopy (
Arkin et al., 1996b
). The quenching of photoexcited
Ru(phen)
2
dppz
2+
by Rh(phi)
2
bpy
3+
is well-characterized and
known to occur by a redox mechanism (
Stemp et al., 1995
)
and not energy transfer (e.g., fluorescence resonance energy
transfer [FRET]), with both the forward and back electron trans-
fers occurring predominantly on the picosecond timescale (
Fig-
ures 3
G and
S4
G) (
Arkin et al., 1996a
). Both of these intercalators
bind to dsDNA more than an order of magnitude more tightly
than do phenazines. Moreover, Ru(phen)
2
dppz
2+
is luminescent
in aqueous solution when intercalated in dsDNA (or otherwise
protected from water), precluding a 2
nd
order reaction between
the complexes in solution (
Friedman et al., 1990
). Thus, in
time-resolved emission experiments on the nanosecond time-
scale, static quenching, where quenching is fast and occurs
without a change in Ru(phen)
2
dppz
2+
excited state lifetime, is
consistent with DNA-mediated charge transfer (
Arkin et al.,
1996b
), while slower dynamic quenching, which leads to a
change in emission lifetime, is consistent with a slower diffusive
process. We first compared the Ru(phen)
2
dppz
2+
signals of
liquid grown and biofilm suspensions (of the same optical den-
sity) to determine if the signal was specific to eDNA; the ruthe-
nium complex is not expected to be taken up by the cells
and bind to genomic DNA on the timescale of this experiment
(
Figure 3
H). We only observed ruthenium luminescence in
the presence of the biofilm suspension, consistent with ruthe-
nium luminescence being associated with binding to eDNA.
We then examined the pattern of quenching of Ru(phen)
2
dppz
2+
luminescence by Rh(phi)
2
bpy
3+
.
Figure 3
I shows static
quenching where the intensity of the Ru(phen)
2
dppz
2+
signal de-
creases, while the observed decay kinetics are unchanged (
Fig-
ure S4
H). Therefore, we conclude that biofilm eDNA can support
rapid DNA charge transfer between these two metal complexes
faster than the timescale for diffusion.
Electrode-Grown Biofilms Retain PYO Capable of
Extracellular Electron Transfer
Having investigated phenazine retention and electron transfer
separately, we next wanted to establish a system in which we
could monitor both of these processes simultaneously
in vivo
.
We took an electrochemical approach to achieve this, growing
P. aeruginosa
biofilms on interdigitated microelectrode elec-
trode arrays (IDA) (
Figure 4
A). Biofilms were grown by incubating
IDAs in bioelectrochemical reactors with planktonic cultures
under oxic conditions and stirring (
Figure 4
B). After 4 days,
with medium replaced daily, mature biofilms were transferred
to anoxic reactors with fresh medium for electrochemical mea-
surements. Confocal and SEM imaging revealed that the IDA
biofilms were heterogeneous in composition, but consistently
contained multicellular structures of live cells and abundant
eDNA (
Figures 4
C–4E,
S5
A, and S5B). Under these conditions,
the cells predominately produced PYO (and some PCA), as
measured by LC-MS (
Figure S5
C). Because PYO was the most
tightly binding phenazine in colony biofilms and
in vitro
,we
focused on this phenazine for the remainder of these
experiments.
Originally used to measure conductivity of abiotic materials,
the IDA has a 2-working electrode geometry and recently was
adapted to study EET through microbial biofilms (
Xu et al.,
2018
;
Yates et al., 2015
;
Boyd et al., 2015
;
Snider et al., 2012
).
Measurements are made by driving electron transfer between
the two electrode bands across a 5
m
m gap (
Figure 4
F), resulting
in EET that is decoupled from the cells’ metabolic activity. Spe-
cifically, we used a generator-collector (GC) strategy to measure
EET through the biofilm, where the ‘‘generator’’ electrode is
swept from an oxidizing potential (E = 0 mV versus Ag/AgCl) to
a reducing potential (E =
500 mV), while the ‘‘collector’’ elec-
trode maintains a fixed oxidizing potential (E = 0 mV). In this
GC arrangement, electron transfer into the biofilm from the
generator occurs as the potential of the generator is swept nega-
tively, reducing PYO (E
1/2
=
250 mV) at the biofilm/generator
interface. Electrons are conducted across the gap through the
biofilm due to EET, either by physical diffusion of PYO or other
mechanisms. The electron transfer at both the generator and
collector is measured as current (I
gc
) and plotted against the po-
tential of the generator electrode. Implicit in generating a sus-
tained I
gc
is recycling of the redox molecules supporting current.
Figure 4
G shows that WT and
D
phz
+ PYO biofilms supported
current (above background) across the 5-
m
m gap over a
3 min scan as the generator potential approached PYO’s redox
potential (
250 mV versus Ag/AgCl), consistent with PYO-medi-
ated EET; significantly,
D
phz
alone did not. Dependency of cur-
rent on the generator potential observed here is consistent with
PYO-mediated EET being a diffusive process. Reduction of PYO
at the generator and oxidation of PYO at the collector generates
a redox gradient that drives EET through the intervening biofilm,
resulting in current centered on the PYO midpoint potential that
saturates at strongly reducing generator potentials (
Snider et al.,
2012
). Here, the diffusive nature of EET would reflect either phys-
ical diffusion of PYO through the biofilm (reduced PYO from the
generator to collector, oxidized PYO from the collector to the
generator) or the effective diffusion of electrons through bound
PYO (
Strycharz-Glaven et al., 2011
).
To test whether these short-term GC measurements of biofilm
EET indicate long-term ability to support metabolic activity, we
poised the IDA electrodes at an oxidizing potential (+100 mV
versus Ag/AgCl) and monitored the current produced by the bio-
film over 4 days in the presence of 40 mM succinate as the
ll
Cell
182
, 919–932, August 20, 2020
925
Article