1
Title:
Prevalence and correlates of phenazine resistance in culturable bacteria from a dryland wheat field
1
2
Running title:
Phenazine resistance in wheat rhizosphere community
3
4
Authors:
Elena K. Perry
1
, Dianne K. Newman
1,
2
*
5
6
Affiliations:
7
1
Division of Biol
ogy and Biological Engineering, California Institute of Technology, Pasadena, CA
8
91125, USA
9
2
Division of Geological and Planetary Sciences, California Institute of Technology, Pasadena, CA 91125,
10
USA
11
12
Contact information:
13
E
-
mail:
dkn@caltech.edu
14
T
e
l
:
626
-
395
-
3543
15
Ab
s
t
r
a
c
t
16
P
he
na
z
i
ne
s
a
r
e
a
c
l
a
s
s
of
b
a
c
t
e
r
i
a
l
l
y
-
pr
od
uc
e
d
r
e
dox
-
a
c
t
i
ve
na
t
u
r
a
l
a
nt
i
b
i
o
t
i
c
s
t
ha
t
h
a
ve
17
de
m
ons
t
r
a
t
e
d
p
ot
e
nt
i
a
l
a
s
a
s
us
t
a
i
na
bl
e
a
l
t
e
r
n
a
t
i
ve
t
o
t
r
a
d
i
t
i
o
na
l
p
e
s
t
i
c
i
de
s
f
o
r
t
h
e
bi
o
c
on
t
r
o
l
o
f
18
f
u
nga
l
cr
o
p
di
s
e
a
s
e
s
.
H
ow
e
ve
r
,
t
h
e
p
r
e
v
a
l
e
nc
e
o
f
ba
c
t
e
r
i
a
l
r
e
s
i
s
t
a
nc
e
t
o
a
g
r
i
c
ul
t
u
r
a
l
l
y
-
r
e
l
e
va
n
t
19
phe
n
a
z
i
ne
s
i
s
po
o
r
l
y
und
e
r
s
t
o
od
,
l
i
m
i
t
i
n
g
bo
t
h
t
h
e
un
de
r
s
t
a
nd
i
n
g
of
h
ow
t
he
s
e
m
ol
e
c
ul
e
s
m
i
gh
t
20
s
ha
pe
r
hi
z
os
p
he
r
e
b
a
c
t
e
r
i
a
l
c
o
m
m
un
i
t
i
e
s
a
nd
t
he
a
bi
l
i
t
y
t
o
pe
r
f
or
m
r
i
s
k
a
s
s
es
s
m
en
t
f
o
r
of
f
-
t
a
r
ge
t
21
ef
f
ec
t
s
.
H
e
r
e
,
w
e
de
s
c
r
i
be
pr
o
f
i
l
e
s
o
f
s
us
c
e
pt
i
b
i
l
i
t
y
to
t
h
e
a
n
t
i
f
ung
a
l
a
ge
nt
phe
na
z
i
n
e
-
1
-
c
a
r
b
ox
yl
i
c
22
aci
d
(
P
C
A
)
a
c
r
os
s
m
o
r
e
t
h
a
n
10
0
ba
c
t
e
r
i
a
l
s
t
r
a
i
ns
i
s
ol
a
t
e
d
f
r
om
a
w
he
a
t
f
i
e
l
d
w
he
r
e
PC
A
p
r
o
du
c
e
r
s
23
a
r
e
i
nd
i
ge
no
us
a
n
d
a
bu
nda
nt
.
We
f
i
nd
t
ha
t
Gr
a
m
-
pos
i
t
i
ve
ba
c
t
e
r
i
a
a
r
e
t
ypi
c
a
l
l
y
m
o
r
e
s
e
ns
i
t
i
v
e
to
24
P
C
A
t
ha
n
G
r
a
m
-
ne
ga
t
i
ve
ba
c
t
e
r
i
a
,
bu
t
t
h
a
t
t
he
r
e
i
s
al
s
o
s
i
gn
i
f
i
c
a
n
t
va
r
i
a
b
i
l
i
t
y
i
n
s
us
c
e
pt
i
b
i
l
i
t
y
b
ot
h
25
w
i
t
hi
n
a
nd
a
c
r
os
s
phy
l
a
.
P
h
e
na
z
i
n
e
-
re
s
i
s
t
a
n
t
s
t
ra
i
n
s
ar
e
m
or
e
l
i
k
e
l
y
t
o
b
e
i
s
ol
a
t
e
d
f
r
o
m
t
he
w
he
a
t
26
.
CC-BY-NC-ND 4.0 International license
available under a
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
The copyright holder for this preprint
this version posted November 24, 2021.
;
https://doi.org/10.1101/2021.11.23.469799
doi:
bioRxiv preprint
2
rhizosphere, where PCA producers
are
also more abundant, compared to bulk soil.
Furthermore,
27
PCA toxicity is pH
-
dependent for most susceptible strains
and broadly correlates with PCA
28
reduction rates,
suggesting
that uptake and redox
-
cycling are im
portant determinants of
phenazine
29
toxicity
.
Our results shed light on which
classes
of bacteria are most likely to be susceptible to
30
phenazine toxicity
in acidic or neutral soils
. In addition,
the
taxonomic and phenotypic divers
ity of
31
our
strain
collection
represents a valuable resource
for future
studies on the role of natural antibiotics
32
in shaping wheat rhizosphere communities
.
33
Importance
34
Microbial communities
c
o
n
t
r
i
b
u
t
e
t
o
crop health
i
n
i
m
p
o
r
t
a
n
t
w
a
y
s
.
F
o
r
e
x
a
m
p
l
e
,
p
henazine
35
metabolites are a
class of redox
-
active molecules
m
a
d
e
b
y
d
i
v
e
r
s
e
s
o
i
l
b
a
c
t
e
r
i
a
that
u
n
d
e
r
p
i
n
the
36
biocontrol of wheat and o
ther crops.
T
h
e
i
r
p
h
y
s
i
o
l
o
g
i
c
a
l
f
u
n
c
t
i
o
n
s
a
r
e
n
u
a
n
c
e
d
:
i
n
s
o
m
e
c
o
n
t
e
x
t
s
37
t
h
e
y
a
r
e
t
o
x
i
c
,
i
n
o
t
h
e
r
s
,
b
e
n
e
f
i
c
i
a
l
.
While much is known about phenazine production and the effect
38
of phenazines on producing strains,
o
u
r
a
b
i
l
i
t
y
t
o
p
r
e
d
i
c
t
h
o
w
p
h
e
n
a
z
i
n
e
s
m
i
g
h
t
s
h
a
p
e
t
h
e
39
c
o
m
p
o
s
i
t
i
o
n
o
f
e
n
v
i
r
o
n
m
e
n
t
a
l
m
i
c
r
o
b
i
a
l
c
o
m
m
u
n
i
t
i
e
s
i
s
p
o
o
r
l
y
c
o
n
s
t
r
a
i
n
e
d
;
t
h
a
t
p
h
e
n
a
z
i
n
e
40
prevalence in
t
h
e
rhizosphere
is predicted to increase in
a
r
i
d
soils
a
s
t
h
e
c
l
i
m
a
t
e
c
h
a
n
g
e
s
p
r
o
v
i
d
e
s
a
n
41
i
m
p
e
t
u
s
f
o
r
f
u
r
t
h
e
r
s
t
u
d
y
.
A
s
a
s
t
e
p
t
o
w
a
r
d
s
g
a
i
n
i
n
g
a
p
r
e
d
i
c
t
i
v
e
u
n
d
e
r
s
t
a
n
d
i
n
g
o
f
p
h
e
n
a
z
i
n
e
-
l
i
n
k
e
d
42
m
i
c
r
o
b
i
a
l
e
c
o
l
o
g
y
,
w
e
document the effects of phenazines on diverse bacteria that were co
-
isolated
43
from a wheat rhizosphere
a
n
d
i
d
e
n
t
i
f
y
c
o
n
d
i
t
i
o
n
s
a
n
d
p
h
e
n
o
t
y
p
e
s
t
h
a
t
c
o
r
r
e
l
a
t
e
w
i
t
h
h
o
w
a
s
t
r
a
i
n
44
w
i
l
l
r
e
s
p
o
n
d
t
o
p
h
e
n
a
z
i
n
e
s
.
45
46
Introduction
47
Diverse microorganisms
produce
natural
antibio
t
ics
that can kill or inhibit the growth of other
48
microbes
(Bérdy, 2012; Granato
et al.
, 2019)
. Several
such
compounds have been commercialized
49
as antimicrobial drugs
for the treatment of infections
, beginning with penicillin in the 19
40s
50
(Aminov, 2010; Hutchings
et al.
, 2019)
.
Unfortunately, the s
elective pressure
s
exerted by the
51
widespread
use of
antibiotics
in medicine and agriculture
have
led to
worrisome
increases in the
52
prevalence of antimicrobial resistance
among human and animal pathogens
over the past several
53
decades
(Davies and Davies, 2010; Manyi
-
Loh
et al.
, 2018)
.
Yet
while th
is
repercussion
of
human
54
.
CC-BY-NC-ND 4.0 International license
available under a
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
The copyright holder for this preprint
this version posted November 24, 2021.
;
https://doi.org/10.1101/2021.11.23.469799
doi:
bioRxiv preprint
3
antibiotic use has been well docu
mented
,
comparatively
little is known about the
ecological
effects
55
of microbially
-
produced antibiotics in natural environments
(Aminov, 2009; Sengupta
et al.
, 2013)
.
56
R
ecent
studies have suggested that natural antibiotics
may
serve a variety of
functions
for their
57
producers
beyond
the
suppression of competi
n
g microbes
(Davies, 2006; Davies
et al.
, 2006)
,
58
including
nutrient acquisition
(Wang and Newman, 2008; McRose and Newman, 2021)
,
59
conserv
ation of
energy
in the absence of oxygen
(Glasser
et al
.
, 2014; Glasser
et al.
, 2017)
,
and
cel
l
-
60
cell signaling
(Dietrich
et al.
, 2006; Linares
et al.
, 2006)
.
At the same time
,
toxicity
to
one or more
61
microorganisms
is
by definition
a feature
of
these molecule
s
,
but
t
he
extent to which this
trait
shapes
62
their influence on microbial commun
ities
is
unclear
(Demain and Fang, 2000; Davies and Ryan,
63
2012; Bérdy, 2012)
.
In addition, for
many if not
most natural antibiotics,
the
determinants
and
64
prevalence
of susceptibility or resistance
to their toxicity
remain
unknown
or poorly characterized
65
(Handelsman and Stabb, 1996)
.
T
hese gaps in knowledge
hinder our a
bility to understand and predict
66
the impacts of
these
metabolites
on microbial communities of interest
.
67
One
environmental context in which
natural
antibiotics are thought to be particularly
68
a
bundant
and ecologically relevant
is the rhizosphere
—
the narrow
plane of soil immediately
69
adjacent to plant roots
(Haas and Défago, 2005; Mavrodi
et al.
, 2012; Tyc
et al.
, 2017)
.
N
atural
70
antibiotics
such as phenazines and 2,4
-
diacetylphloroglucinol
have been directly detected in
the
71
rhizospheres of multiple crops
, including
wheat
, potato, and sugar beet
(Thomashow
et al.
, 1990;
72
Bergsma
-
Vlami
et al.
, 2005; Mavrodi
et al.
, 2012)
, and p
henazines have been shown to increase the
73
fitness of their producers
when competing
with other
microbes in
the rhizosphere
(Mazzola
et al.
,
74
1992; Yu
et al.
, 2018)
.
P
roduction
of these
molecules
has
also
been demonstrated to underpin
the
75
ability of certain bacteria to control
fungal
crop disea
ses
(Thomashow and Weller, 1988;
76
Thomashow
et al.
, 1990; Haas and Keel, 2003; Mazurier
et al.
, 2009; Yu
et al.
, 2018)
,
furt
her
77
indicating that
natural antibiotics
can act as agents of microbe
-
microbe antagonism in
the
78
rhizosphere
.
As a result of
this
activity, phenazine
-
producing
Pseudomonas
strains have received
79
attention as potential biocontrol agents that
could serve as a m
ore sustainable alternative to traditional
80
synthetic pesticides in agriculture
(Handelsman and Stabb, 1996; Haas and Keel, 2003)
.
However,
81
several challenges remain with respect to the
practical
application of
these strains,
including
82
inconsistent efficacy under field conditions
(Haas
and Keel, 2003; Haas and Défago, 2005)
,
limited
83
understanding of the
mechanisms
and evolutionary dynamics
of resistance to phenazines
(Mazzola
84
.
CC-BY-NC-ND 4.0 International license
available under a
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
The copyright holder for this preprint
this version posted November 24, 2021.
;
https://doi.org/10.1101/2021.11.23.469799
doi:
bioRxiv preprint
4
et al.
, 1995; Handelsman and Stabb, 1996)
,
and
the possibility of
off
-
target effects
(Haas and Keel,
85
2003)
.
86
Importantly,
with regard to the latter concern
,
p
henazines are toxic not only to fungi, but also
87
to some bacteria
(Baron and Rowe, 1981; Turner and Messenger, 1986; Costa
et a
l.
, 2015)
.
Yet
88
while
the
utility of
phenazine
-
producing pseudomonads
for
biocontrol of fungal crop diseases
has
89
been extensively investigated
,
the
potential
impact
of these strains
on
non
-
target
bacterial residents
90
of the rhizosphere
is less well understo
od
.
O
ne study
found that
inoculation of
Pseudomonas
91
biocontrol strains shifted the rhizosphere community of maize seedlings
,
pushing the ratio of
Gram
-
92
positive
to Gram
-
negative
bacteria
in favor of the latter
; however, this
analysis was
based on
93
profiling
colony growth rates and
whole
-
cell fatty acids from pooled cultured isolates
,
greatly
94
limiting the taxonomic resolution
and making it difficult to rule out whether the Gram
-
negative
95
biocontrol strains might themselves have directly contributed to the shift
(Kozdrój
et al.
, 2004
)
.
On
96
the other hand, at least three studies found no
notable
or consistent
effect
s
of
introduced
97
Pseudomonas
species
on the
rhizosphere
bacterial communities of wheat or potato
(Gagliardi
et al.
,
98
2001; Bankhead
et al.
, 2004; Roquigny
et al.
, 2018)
, albeit
the studies in wheat employed methods
99
with limited discriminatory power
(
namely, carbon
source
utilization pr
ofiling and terminal
100
restriction fragment length polymorphism
)
. Given these mixed results and the lack of fine
-
grained
101
spatial or taxonomic resolution in most studies on this topic, whether phenazine
-
producing bacteria
102
actively
shape the surrounding rhizos
phere
bacterial
community through antibiosis
, potentially
at the
103
micro
n or millimeter
scale
,
remains an open question.
104
In addition
to
the lack of clarity regarding
the
effects of phenazines on
rhizosphere
bacterial
105
communities, t
he
taxonomic
and physiolog
ical
correlates of
phenazine resistance
in bacteria
remain
106
incompletely understood.
The toxicity of phenazines
is
generally
ascribed
to the generation of
107
reactive oxygen species
(ROS)
and interference with respiration
(Hassan and Fridovich, 1980; Baron
108
and Rowe, 1981; Voggu
et al.
, 2006; Perry and Newman, 2019)
. Previous studies have
suggested
109
that efflux pump expression, cell permeability, oxidative stress responses, and the composition of
110
the respiratory electron transport chain can affect bacterial susceptibility to phenazines
(Voggu
et
111
al.
, 2006; Khare and Tavazoie, 2015; Noto
et al.
, 2017; Wolloscheck
et al.
, 2018; Perry and
112
Newman, 2019; Meirelles
et al.
, 2021)
.
In addition, a comparison of 14 bacterial strains
indicated
113
that Gram
-
negative bacteria as a grou
p may be more resistant to phenazine toxicity than Gram
-
114
.
CC-BY-NC-ND 4.0 International license
available under a
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
The copyright holder for this preprint
this version posted November 24, 2021.
;
https://doi.org/10.1101/2021.11.23.469799
doi:
bioRxiv preprint
5
positive bacteria
(Baron and Rowe, 1981)
.
However, all of these studi
es
focused on
a specific
115
phenazine, pyocyanin, that is
particularly toxic
(Meirelles and Newman, 2018)
and best known for
116
its role as a virulence f
actor during infections
of humans and animals
(Lau
et al.
, 2004; Liu and
117
Nizet, 2009)
. Whether the same observations hold true for more agriculturally
-
relevant phenazines
118
such as phenazine
-
1
-
carboxylic acid (P
CA)
(Thomashow and Weller, 1988; Thomashow
et al.
,
119
1990; Mavrodi
et al.
, 2012; Dar
et al.
, 2020)
is unknown
.
120
In this study, we set out to lay a foundation for addressing
unresolved
question
s
about the
121
ecologi
cal
impact
of phenazine toxicity
in the rhizosphere
by determining
the prevalence of
122
phenazine resistance among
bacteria
isolated from
a wheat field in the Inland Pacific Northwest
,
a
123
region
where phenazine production and the biocontrol potential of indige
nous
Pseudomonas
species
124
have been studied for decades
(Thomashow and Weller, 1988; Thomashow
et al.
, 1
990; Mazzola
et
125
al.
, 1992; Mavrodi
et al.
, 2012)
.
W
e
designed
a culture
-
based
assay to
measure sensitivity to
PCA
,
126
which is
the best
-
studied and
one of the
most abundant phenazine
s
in this environment
(Thomashow
127
et al.
, 1990; Mavrodi
et al.
, 2012; Dar
et al.
, 2020)
. We
also
performed full
-
length 16S rRNA gene
128
sequencing of
our
isolates in order to
assess the relationship between taxonomy and
PCA
resistance
.
129
Finally, to a
ssess potential physiological correlates of PCA resistance, we measured PCA reduction
130
rates for a subset of phenotypically
-
diverse strains and investigated the effects of broad
-
spectrum
131
efflux pump inhibitors on growth in the presence of PCA
.
132
Results
133
Taxo
nom
ic diversity
of culturable bacteria from dryland wheat rhizosphere
s
and bulk soil
134
A total of
166
strains of bacteria were isolated from 12 soil samples collected from a wheat
135
field at Washington State University’s Lind Dryland Research Station
in early
August 2019, shortly
136
after the wheat harvest
. The samples comprised 4 replicates each of wheat rhizosphere
(“
Wheat
”)
,
137
bulk soil collected between planted rows
(“Between”)
, and bulk soil from a virgin patch of
138
uncultivated soil adjacent to the field
(“Virgi
n”
)
.
Full
-
length 16S rRNA gene sequencing revealed
139
that the isolates represented
2
4
genera across 4 phyla: Actinobacteria, Bacteroidetes, Firmicutes, and
140
Proteobacteria.
The vast majority of isolates from the bulk soil samples (both Between and Virgin)
141
wer
e Actinobacteria or Firmicutes
. I
n
Wheat
samples,
on the other hand,
the
combined
proportions
142
.
CC-BY-NC-ND 4.0 International license
available under a
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
The copyright holder for this preprint
this version posted November 24, 2021.
;
https://doi.org/10.1101/2021.11.23.469799
doi:
bioRxiv preprint
6
of these
two
phyla were lower,
accounting
for 18
-
50%
and
10
-
25% of isolates
respectively
,
while
143
Proteobacteria accounted for approximately 25
-
60% of isolates depe
nding on the replicate, and
1
-
3
144
strains of
Bacteroidetes were also detected
in each
replicate
(5
-
12% of isolates)
(
Fig. 1
)
.
145
Taxonomic and spatial distribution of
PCA resistance phenotypes
146
We screened our isolates for resistance to PCA at both circumneutr
al and acidic pH (7.3 and
147
5.1, respectively), because
the
toxicity of PCA to diverse organisms is known to vary depending on
148
pH
(Brisbane
et al.
, 1987; Cezairliyan
et al.
, 2
013)
, and because the bulk soil and rhizosphere pH of
149
wheat fields in the Inland Pacific Northwest can vary by >3 units (pH 4.3
-
8.0) depending on
150
geographic location and fertilizer treatment status
(Smiley and Cook, 1973)
. Th
e pH
-
dependency of
151
PCA toxicity
has been attributed
to the
fact that the
deprotonated form
of PCA
is negatively charged
152
(
Fig.
S1
)
;
the negative
charge on bacterial cell
walls and the outer membrane of Gram
-
negative
153
bacteria (or the negative membrane potential of eukaryotic cells)
likely
hinder
s
uptake of this species
.
154
T
he protonated form
of PCA
, on the other hand,
is neutral and
presumably can
passively diffuse
155
across cel
l membranes given the small size and hydrophobic nature of the molecule
(Price
-
Whelan
156
et al.
, 2006; Cezairliyan
et al.
, 2013)
.
The pKa of PCA is 4.24 at 25 °C; thus, at
pH 7, only 0.17%
157
of PCA in solution
is
protonated compared to 14.8% at pH 5
,
typically
leading to
greater toxicity at
158
the lower pH
(Brisbane
et al.
, 1987)
.
We chose 100
μ
M as the
working
concentration of PCA as this
159
is likely to be in a physiologically relevant range
based on concentrations
measured
both in pu
re
160
cultures and in the field
. In broth cultures of biocontrol strains of
Pseudomonas
, PCA accumulates
161
to concentrations ranging from dozens to hundreds of micromolar
(Séveno
et al.
, 2001; Tagele
et al.
,
162
2019)
. In
natural
wheat
rhizospheres
, PCA
has
been detected at
nanomolar
concentrations
(Mavrodi
163
et al.
, 2012)
, but these bulk measurements almost certainly underestimate local concentrations at the
164
micron scale
given that
bacteria colonize the rhizosphere in a
patchy manner
(Thomashow
et al.
,
165
1990)
.
Notably,
PCA can accumulate in biofilms to concentrations 360
-
fold greater on a per volume
166
basis compared to broth cultures
(Séveno
et al.
, 2001)
, suggesting that local concentrations of PCA
167
in the rhizosphere, where biocontrol strains form
robust
biofilms
(LeTourneau
et al.
, 2018)
, may be
168
orders of magnitude higher than the
reported
bulk
values
.
169
The screen was performed
using 24
-
well plates and 0.1x
tryptic soy agar (TSA)
that was
170
either left unadjusted (pH 7.3
-
7.5
) or adjusted to pH 5.1.
Each isolate wa
s spotted onto agar in
171
.
CC-BY-NC-ND 4.0 International license
available under a
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
The copyright holder for this preprint
this version posted November 24, 2021.
;
https://doi.org/10.1101/2021.11.23.469799
doi:
bioRxiv preprint
7
separate wells
to prevent crosstalk and antagonism between the strains
, and
image analysis
was used
172
to derive quantitative information about the growth of each strain
over the course of 7 days (with
173
one timepoint per day)
.
Because se
veral
strains among our
166
isolates appeared to be duplicates
of
174
each other based on 16S sequence and colony morphology
, w
e restricted
the screen
to
1
08
strains
175
that
we judged likely to be unique (
Table S1
A
); wh
e
re
multiple strains appeared identical, we
chose
176
a
representative strain
.
We also included
30 strains obtained from public culture collections
that
177
represent
ed species found among our
isolates (
Table S1
B
)
, in order to investigate the extent to which
178
PCA resistance phenotypes are consistent
within s
pecies
across different strains
isolated from
179
different geographical locations
.
A
ccurate quantification of growth
was
not possible
in this screen
180
for
certain
strains that formed transparent colonies
,
produced
dark pigments
, or had a tendency to
181
turn mucoid
and spread
(
Fig. S2
)
.
Nevertheless
,
this
assay
enabled us to derive
detailed
profiles of
182
PCA sensitivity and resistance
for the vast majority of our strains
.
183
To
compare PCA susceptibility across strains
, we
first focused
on a single
-
timepoint
184
snapshot of
each strain’s phenotype
taken
at the
equivalent
of
early stationary phase (i.e.
around the
185
time that
the spots on non
-
PCA control plates reached
their
maximal density).
In accordance with
186
previous reports of the pH dependency of PCA toxicity, we found that
more of our strains were
187
sensitive to PCA at pH 5.1 than at pH 7.3, and strains that were mildly inhibited by PCA at pH 7.3
188
were typically inhibited
more strongly
at pH 5.1 (
Fig.
2A
-
B
).
At pH 5.1, strains of
Actinobacteria
189
and Firmicutes were strongly inh
ibited by PCA, while most Proteobacteria were relatively resistant
.
190
M
embers of Bacteroidetes
exhibited variable responses, ranging from mild to severe growth
191
inhibition (
Fig
. 2A
-
B
)
.
At pH 7.3, there was considerably more phenotypic variation
among
the
192
Acti
nobacteria
(
Fig.
2A
-
B
)
, particularly within the
Streptomyces
genus
.
In some cases, there was
193
even
noticeable
variation within the same
putative
species of
Streptomyces
(
Fig.
S
3
).
In addition,
194
two strains of
Microbacterium
(W2I7 and W4I20)
, as well as one s
train of
Arthrobacter
(W3I6),
195
grew
slightly
better in the presence of PCA at pH 7.3 compared to the control condition
, perhaps
196
suggesting that they can use PCA as a carbon source or otherwise benefit from its presence at a pH
197
where toxicity is limited
(
Fig
.
2B
)
.
On the other hand, i
n contrast to the phenotypic variation seen
198
among Actinobacteria,
Firmicutes remained
almost
universally
sensitive to PCA
at pH 7.3
(albeit
199
somewhat less so than at pH 5.1),
while all members of Bacteroidetes were resistant to PC
A at pH
200
7.3, in contrast to their sensitivity to PCA at pH 5.1 (
Fig.
2A
-
B
).
Proteobacteria
remained
generally
201
.
CC-BY-NC-ND 4.0 International license
available under a
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
The copyright holder for this preprint
this version posted November 24, 2021.
;
https://doi.org/10.1101/2021.11.23.469799
doi:
bioRxiv preprint
8
resistant
(
Fig.
2A
-
B
)
, with the most notable exception being
a strain of
Sphingomonas
faeni
,
isolate
202
W4I17
,
that
was
inhibited by PCA
at both
pH 7
.3
and
pH 5.1
(
Fig.
2A
-
B
)
.
203
Interestingly, while most strains yielded qualitatively consistent results across
independent
204
biological replicates that were separated by months
(
Fig.
S
4
-
S
7
)
, a handful of strains displayed
205
variable
growth and
lag times at pH 5
.1
either
in the absence of PCA (mostly among Actinobacteria
206
and Firmicutes), or in the presence of PCA (a few strains among the Bacteroidetes). Growth at pH
207
7.3 was generally more consistent, with the exception being two strains of
Neob
acillus niacini
208
(V4
I10 and V4I25) that initially failed to grow in the presence of PCA, but grew with minimal to no
209
inhibition in subsequent replicates. The reasons for these discrepancies are unclear, but may be
210
related to
variations in
how long the strains had been in stat
ionary phase prior to inoculating the
211
experimental cultures,
which was difficult to control precisely due to differing growth rates across
212
the large number of strains.
T
he trace nutrient content may
also
have varied
between different
213
batches of media
, as
t
he first replicate was performed
using
a different lot of
tryptic soy broth
214
compared to subsequent replicates.
Thus,
the growth and PCA sensitivity of
some
strains
may be
215
influenced by environmental factors beyond pH that remain to be elucidated.
216
We
next
e
xamined whether there was any evidence of correlation between PCA resistance
217
phenotypes and
the
type of soil each strain was isolated from
(Between, Virgin, or Wheat)
.
A
218
previous study based on samples
taken
from the same
wheat
field found that the relativ
e abundance
219
of phenazine producers was higher in the wheat rhizosphere compared to adjacent bulk soil
(Dar
et
220
al.
, 2020)
, a finding that we also confirmed using shotgun metagenomic sequencing of our soil
221
samples (
Fig.
2C
)
.
We
therefore
hypothesized that
if
PCA
-
mediated antibiosis had shaped
the
222
bacterial community composition
of
this
field
,
the prevalence of PCA resistance would be higher
223
among isolates from the rhizosphere samples. Indeed, t
he Wheat samples clearly had the highest
224
proportion of PCA
-
resistant isolates
, regardless of whether resistance was assessed at pH 5.1 (
Fig.
225
2D
) or pH 7.
3 (
Fig.
2E
)
.
By contrast, a
ll
strains
from Between or Virgin samples were highly
226
sensitive to PCA
at pH 5.
1, though
at pH 7.3, the resistance phenotypes of isolates from either type
227
of bulk soil spanned the full range from highly sensitive
to completely re
sistant.
However, these
228
findings are conflated with the fact that all of our Bacteroidetes strains, and all but one strain of
229
Proteobacteria, were exclusively isolated from the Wheat samples, and that the members of these
230
two phyla were generally relativel
y resistant to PCA
, especially at pH 7.3
.
Further experiments
will
231
.
CC-BY-NC-ND 4.0 International license
available under a
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
The copyright holder for this preprint
this version posted November 24, 2021.
;
https://doi.org/10.1101/2021.11.23.469799
doi:
bioRxiv preprint
9
be
necessary
in order to distinguish whether
the enrichment of PCA
-
resistant phenotypes in the
232
rhizosphere samples is a
consequence
of PCA
-
mediated antibiosis or
merely
a
n indirect
reflecti
on
233
of other factors
that favor
the growth of Proteobacteria and Bacteroidetes in the rhizosphere.
234
Finally,
visualizing
the growth of each strain over
the full 7
-
day
time
course
revealed
more
235
nuanced variations among the PCA resistance phenotypes
at
both
pH
5.1 and pH 7.3
(
Figs.
3 and
236
S
4
-
S
7
)
.
For example,
while
some
strains of Bacteroidetes displayed increased lag
and/or slower
237
growth rates
in the presence of PCA at pH 5.1, the growth on PCA
often
eventually caught up to the
238
PCA
-
free
control
s
.
A
t pH 7.3,
the
effect of PCA on
Perib
acillus
species was a combination of
239
increased lag
and
lower final
cell density,
but apparently not decreased maximal growth rate,
as
240
compared to no growth at all on PCA at pH 5.1
.
This was distinct from the slower growth rates seen
241
for certain strains of
Bacillus
,
Neobacillus
, and
Paenibacillus
at pH 7.3
.
A
few strains of
the latter
242
three genera
appeared
to be still
unable to grow on PCA at pH 7.3
based on the values
derived from
243
image analysis
(
Fig.
S
5
)
.
However,
examination of the
plate images by eye revealed that this
244
sometimes
reflected
a
limit
ation in the
sensitivity of our imaging and quantification method
s
to low
245
levels of growth
,
rather than a true lack of growth (
Fig.
S2
)
.
246
PCA reduction rates in relation to PCA susceptibilit
y
247
We next sought to determine whether there are physiological correlates of PCA susceptibility
248
or resistance across taxonomically
-
diverse bacteria.
Given that
phenazines
a
re
redox
-
active
249
molecule
s
whose toxicity is thought to be related to ROS generation
as a consequence of redox
250
cycling in the cell
(Hassan and Fridovich, 1980; Mavrodi
et al.
, 2006)
, we hypothesized that
251
susceptibility to PCA would be correlated with higher redox
-
cycling rates. PCA oxidation occurs
252
rapidly and abiotically in the presence of ox
ygen
(Wang and Newman, 2008)
, suggesting
that the
253
rate of
PCA reduction would be the primary driver of
differences in redox
-
cycling rates across strains
254
under oxic conditions
.
Therefore, to test our hypothesis, we measured the rate of PCA reduction
255
under anoxia as a proxy for the true redox
-
cycling rate
, using a
subset of strains that covered a range
256
of taxonomic groups and PCA susceptibility phenotypes
.
257
For each tested
strain, the PCA reduction rate was typically higher at pH 5.1 than at pH 7.3
258
(Fig. 4A)
, matching the expectation that PCA uptake would tend to be
faster at
a lower
pH
and the
259
.
CC-BY-NC-ND 4.0 International license
available under a
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
The copyright holder for this preprint
this version posted November 24, 2021.
;
https://doi.org/10.1101/2021.11.23.469799
doi:
bioRxiv preprint
10
observation that most susceptible strains were more sensitive to PCA at pH 5.1
.
The exceptions were
:
260
1)
strains belonging to
the Firmicutes, which had lower reduction rates at pH 5.1 in our assay despite
261
being more susceptible
to PCA under this condition
, and 2)
Sphingomonas faeni
W4I17, which was
262
in fact more susceptible to PCA at pH 7.3 than at pH 5.1
(
Fig.
2A
-
B
)
, matching the pH dependency
263
of its rate of PCA reduction. Notably,
some of
the tested Firmicutes
often
struggled t
o grow at pH
264
5.1 even in the absence of PCA
(Fig.
S
5
)
.
This observation might account for the inhibitory effect
265
of acidic pH on PCA reduction in these strains, given that a functional metabolism
would be
required
266
for the generation of cellular reductants,
such as NADH, that can indirectly or directly reduce PCA.
267
F
or
the
strains of Firmicutes that were able to grow at acidic pH
(namely B1I1 and B1I6)
, the reason
268
for the
ir
lower
PCA
reduction rate at pH 5.1 remains unclear.
269
We also examined whether there was
any correlation between PCA reduction rates
and
270
resistance levels
across different strains at each
pH.
To do so, we plotted each strain’s PCA reduction
271
rate against the
early stationary phase snapshot
ratio of growth on PCA versus no PCA
(
Fig.
4B
).
272
We als
o calculated
Spearman’s correlation coefficient (
r
s
)
.
Interes
tingly, there was a statistically
273
significant negative correlation between reduction rate and resistance at pH 7.3 (
r
s
=
-
0.
70
,
p
<
274
0.001
), but not at pH 5.1 (
r
s
=
-
0.
16
,
p
= 0.
4485
).
However, th
ere were exceptions to the trend even
275
at pH 7.3; for example, while PCA severely
and consistently
inhibited the growth of the four fastest
276
PCA
-
reducing strains, the fifth
-
fastest strain, W2I1 (
Pantoea agglomerans
)
,
appeared to be
277
completely resistant to PC
A
in one biological replicate, though its growth was mildly inhibited by
278
PCA in another replicate (
Fig. S
7
)
.
In addition, strains with reduction rates in the
middle
range of
279
0.2
-
0.4 nmol/OD
600
/min spanned the full range of resistance phenotypes, from a gro
wth ratio of 0.1
2
280
(
Perib
acillus simplex
B2I2) to 0.9
6
(
Phyllobacterium ifrigiyense
W4I11). T
aken together, these data
281
suggest that a low reduction rate may be helpful in mitigating PCA toxicity but is neither necessary
282
nor sufficient for resistance.
283
Effec
ts of efflux pump inhibitors on PCA susceptibility
284
G
iven that PCA reduction rate alone was insufficient to account for
the range of sensitivities
285
to PCA
,
we asked if efflux
pumps might serve as an important contributor to PCA resistance in some
286
strains. To
address this question, we used two efflux pump inhibitors
(EPIs)
in combination: 1)
287
reserpine, which is thought to target efflux pumps in the major facilitator superfamily (MFS)
(Kumar
288
.
CC-BY-NC-ND 4.0 International license
available under a
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
The copyright holder for this preprint
this version posted November 24, 2021.
;
https://doi.org/10.1101/2021.11.23.469799
doi:
bioRxiv preprint
11
et al.
, 2013)
, and 2) phenylalanine
-
arginine
β
-
naphthylamide
(
PA
β
N), which is thought to target
289
resistance
-
nodulation
-
division (RND) efflux pumps
(Jamshidi
et al.
, 2017)
,
although it may also
290
permeabilize the outer membrane of Gram
-
negative bacteria
(Lamers
et al.
, 2013; Schuster
et al.
,
291
2019)
.
We tested the
se
EPIs on
a subset of
the strains that were used in the PCA reduction a
ssay
;
we
292
excluded
the
Bacillus
strains due to their tendency to settle on the bottom of the wells during growth
293
in the plate reader, which confounded the OD
600
readings. Preliminary experiments
revealed
that
294
different strains were differentially sensitive
to the EPIs themselves
, with growth inhibition evident
295
even in PCA
-
free controls
. We therefore determined for each strain the maximal EPI concentration
296
that, when the two EPIs were used separately, did not markedly reduce growth of
PCA
-
free
controls
297
(up t
o a maximum dose of
20
μ
g/mL
for reserpine and
50
μ
g/mL
for PA
β
N)
(
Table
S2
)
.
298
Interestingly, for many strains, the combination of both EPIs remained highly toxic even after this
299
optimization step, particularly at pH 7.3, possibly indicating synergistic tox
icity or redundant roles
300
of different efflux pumps in preventing intracellular accumulation of toxic metabolic intermediates
301
(
Fig.
S9
); the toxicity appeared to be pH
-
dependent, suggesting that the protonation state of the EPIs
302
might influence their effica
cy
.
In these cases
, it was not possible to assess the impact of efflux pump
303
inhibition on PCA sensitivity. However, for the remaining strains, three types of responses to the
304
EPIs emerged. For one group of strains, t
reatment with both
EPI
s in combination
d
id not affect their
305
resistance to PCA
(Fig. 5A
-
B)
; however, we cannot distinguish whether this is because efflux does
306
not contribute to PCA resistance in these strains, or because the tested EPIs did not effectively target
307
the relevant efflux pumps
. For
th
e second
group of strains, the EPIs
enhanced the toxicity of PCA,
308
as would be expected if efflux is an important mechanism of resistance
(Fig. 5C)
. Finally, for the
309
third group of strains, the EPIs counterintuitively appeared to mitigate the toxicity of PC
A, or vice
310
versa
(Fig. 5D)
. Although
unexpected, the latter phenomenon
indirectly
suggests
that PCA interacts
311
with efflux systems in these strains.
For example,
treatment with
either PCA or the EPIs might
312
upregulate the expression of certain efflux pumps,
leading to decreased toxicity relative to treatment
313
with either type of toxin alone.
Alternatively, given that some pumps in the MFS family are thought
314
to be bidirectional
(Pao
et al.
, 1998)
, it is possible that MFS pump
s
promote uptake
rather than export
315
of PCA in certain strains.
316
Discussion
317
.
CC-BY-NC-ND 4.0 International license
available under a
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
The copyright holder for this preprint
this version posted November 24, 2021.
;
https://doi.org/10.1101/2021.11.23.469799
doi:
bioRxiv preprint
12
In this study, we have characterized profiles of resistance to a
n agriculturally
-
relevant
318
phenazine
across taxonomically diverse bacteria isolated from a wheat field where
phenazine
319
producers are indigenous.
We a
lso examined potential physiological correlates of phenazine
320
resistance
in a subset of these strains.
Our findings
establish a basis
for inferring whether intrinsic
321
resistance is a factor that affects how phenazine production shapes bacterial communities i
n the
322
rhizosphere
. In particular, i
t will now be possible to test data
-
driven predictions regarding which
323
strains, species, or even phyla are most likely to be affected by phenazine
-
mediated antibiosis
, even
324
though the mechanistic underpinnings of phenazin
e susceptibility or resistance are likely
325
multifactorial and appear to differ across species.
326
One of our more notable findings is
that
the previously reported pH dependency of PCA
327
toxicity
varie
s
across
bacterial
taxonomic groups.
For most Actinobacteria
a
nd Firmicutes
,
328
sensitivity to PCA was clearly higher at pH 5.1 than at pH 7.3, indicating greater toxicity at the lower
329
pH as expected
according to the pKa of PCA
. On the other hand
, nearly all tested Proteobacteria
330
were completely resistant to 100
μ
M PCA
regardless of pH, at least down to pH 5.1.
Consequently
,
331
at pH 5.1,
PCA resistance phenotypes largely correlated with phylum
—
and more broadly, a Gram
-
332
positive versus Gram
-
negative divide, which has previously been reported for
the
toxicity
of another
333
phena
zine, pyocyanin
(Baron and Rowe, 1981)
.
The Gram
-
positive versus Gram
-
negative divide
at
334
pH 5.1 is perhaps not surprising
, as
at this pH, a significant proportion
(~14%)
of PCA in solution
is
335
protonated and therefore
would
not
be
repelled by the
negatively charged cell wall.
Under this
336
condition,
the outer membrane of Gram
-
negative species presumably presents an additional barri
er
337
to
the entry of
PCA, helping to limit the intracellular accumulation of the toxin
in the same manner
338
as for numerous other antibiotics
(Nikaido, 1989)
.
Less expected
, however,
was the within
-
phylum
339
and even within
-
species variation in
PCA
resistance phenotypes at pH 7.3
among certain taxonomic
340
groups
.
Interestingly,
se
veral
Streptomyces
strains exhibited at least some PCA
-
dependent growth
341
inhibition at pH 7.3,
even though
many
Streptomyces
species are capable of producing their own
342
toxic
phenazines
(Turner and Messenger, 1986; Dar
et al.
, 2020)
.
Given that the major limit on PCA
343
toxicity at circumneutral pH is thought to be its ability to enter cells, the phenotypic
variability at pH
344
7.3 may
indicate
the presence of transporters or channels capable of phenazine uptake in
some
PCA
-
345
sensitive
Actinobacteria
.
Alternatively, it is possible that
some
PCA
-
sensitive strains
locally
346
acidified the growth medium,
which
was only
weakly buffered (1.4 mM
K
2
HPO
4
)
.
347
.
CC-BY-NC-ND 4.0 International license
available under a
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
The copyright holder for this preprint
this version posted November 24, 2021.
;
https://doi.org/10.1101/2021.11.23.469799
doi:
bioRxiv preprint
13
Importantly
, despite
the general Gram
-
positive versus Gram
-
negative divide
in sensitivity to
348
PCA
at pH 5.1, possessing an outer membrane is evidently not a
leakproof shield against PCA
349
toxicity.
Proteobacteria as a group
were more resistant to PCA at pH 5.1 compared to strains of
350
Bacteroidetes, most of which exhibited increased lag in the presence of PCA, even though both phyla
351
comprise Gram
-
negative bacteria.
Intriguingly, one major difference between
these clades is that
352
Proteobacteria generally
utilize ubiquinone as an electron carrier during aerobic growth
(Collins and
353
Jones, 1981)
, while
the Bactero
idetes genera
screened in this study (
Chitinophaga
,
354
Chryseobacterium
, and
Pedobacter
)
utilize menaquinone
(Lin
et al.
, 2015; Singh
et al.
, 2017; Kong
355
et al.
, 2019)
.
Menaquinones have a lower reduction potential compared to ubiquinones
(White
et al.
,
356
2012)
. Although the standard reduction pot
ential of menaquinone is still higher than that of PCA (
-
357
74 mV compared to
-
177 mV)
(Price
-
Whelan
et al.
, 2006; White
et al.
, 2012)
,
indicating that PCA
358
likely is not reduced by menaquinol,
this difference with ubiquinone nevertheless raises the
359
possibility that PCA may interact differently, and
perhaps more readily, with the aerobic respiratory
360
electron
transport
chain of menaquinone
-
utilizing Bacteroidetes strains compared to Proteobacteria
,
361
thereby generating more
ROS
and/or interfering with the generation of ATP
.
Interestingly, another
362
study h
as shown that
the reduced form of
different phenazine with a low reduction potential, neutral
363
red (
3
-
a
mino
-
7
-
dimethylamino
-
2
-
methylphenazine
), can
directly
transfer electrons to menaquinone,
364
bypassing the proton
-
pumping NADH dehydrogenase complex
that norm
ally transfers electrons
365
from NADH to menaquinone
and thereby
“
short
-
circuiting
” the electron transport chain.
We
366
hypothesize that a
similar phenomenon
may occur
with PCA
in strains that rely on menaquinone
.
In
367
future studies, this
hypothesis
could be test
ed by
1)
performing
in vitro
experiments with purified
368
menaquinone
, ubiquinone,
and reduced PCA to
directly
test whether
quinones
oxidize the latter
and
369
if so, whether the kinetics differ between menaquinone and ubiquinone
, 2)
measuring
ROS
370
production and
steady
-
state ATP pools in selected strains of Bacteroidetes and Proteobacteria both
371
in the presence and absence of PCA, and
3
)
forcing
species
of Proteobacteria (such as
Escherichia
372
coli
) that utilize both ubiquinone and menaquinone in different branches o
f their electron transport
373
chains to rely only on
the
latter (for example, by deleting the genes for ubiquinone biosynthesis),
374
followed by reassessing their sensitivity to PCA
to see if there is any effect
.
375
Beyond the specific strains screened in this stu
dy, the risk assessment of phenazine
-
376
producing biocontrol strains, and the understanding of phenazine biology in general,
would benefit
377
.
CC-BY-NC-ND 4.0 International license
available under a
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
The copyright holder for this preprint
this version posted November 24, 2021.
;
https://doi.org/10.1101/2021.11.23.469799
doi:
bioRxiv preprint
14
greatly from the development of
a platform for the prediction of phenazine resistance phenotypes
378
from genomic or phyloge
netic information. The results of this study already indicate that
, depending
379
on the environmental pH,
phylogenetic information may be of limited utility for prediction of PCA
380
resistance, given the variation in phenotypes for Actinobacteria at circumneutra
l pH.
However, this
381
variability may also hold the key to identifying genome
-
based predictive markers of phenazine
382
resistance.
Given the taxonomic and phenotypic diversity of our strain collection, w
hole
-
genome
383
sequencing of
our isolates
, followed by
compar
ative genomics
, could potentially
reveal
such
384
markers
.
385
In summary, t
his work has laid the groundwork for rectifying a major gap in studies of how
386
introduction
of a phenazine
-
producing biocontrol strain affects rhizosphere bacterial communities.
387
Previous s
tudies have lacked information about the baseline prevalence of resistance in the native
388
communities
(Gagliardi
et al.
, 2001; Bankhead
et al.
, 2004; Kozdrój
et al.
, 200
4; Roquigny
et al.
,
389
2018)
, and in the absence of such information, it is impossible to determine whether a negative result
390
(lack of change in the rhizosphere community) reflects a high prevalence of resistance to PCA that
391
is particular to the studied comm
unity, versus a general lack of toxicity of PCA to most bacteria or
392
fundamental abiotic constraints on the antibacterial activity of PCA in the rhizosphere (e.g. limited
393
diffusion, adsorption to soil particles, etc.). Distinguishing between these scenarios
is key to
394
assessing the risk of unwanted side
effects in rhizosphere communities upon the application of
395
phenazine
-
producing biocontrol strains.
In addition, recent studies have demonstrated that
396
phenazines
produced by the opportunistic pathogen
Pseudomon
as aeruginosa
can promote
bacterial
397
tolerance and resistance to clinical antibiotics
(Schiessl
et al.
, 2019; Zh
u
et al.
, 2019; Meirelles
et
398
al.
, 2021; VanDrisse
et al.
, 2021)
, and that these effects can extend to
other
opportunistic pathogens
399
that are resistant to phenazines
(Meirelles
et al
.
, 2021)
. Thus, understanding the prevalence and
400
genetic determinants of
resistance to phenazines may have implications
not only for agriculture but
401
also for human medicine and beyond, as we continue to uncover new ecological roles for these
402
multifaceted
bacterial metabolites.
403
Methods
404
Isolation of bacteria from wheat rhizosphere and bulk soil samples
405
.
CC-BY-NC-ND 4.0 International license
available under a
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
The copyright holder for this preprint
this version posted November 24, 2021.
;
https://doi.org/10.1101/2021.11.23.469799
doi:
bioRxiv preprint