Bacteria clustering by polymers induces
the expression of quorum-sensing-controlled
phenotypes
Leong T. Lui1
, Xuan Xue2
, Cheng Sui2
, Alan Brown2
, David I. Pritchard2
, Nigel Halliday3
, Klaus Winzer3
,
Steven M. Howdle4
, Francisco Fernandez-Trillo2†
*, Natalio Krasnogor5
* and Cameron Alexander2
*
Bacteria deploy a range of chemistries to regulate their behaviour and respond to their environment. Quorum sensing is
one method by which bacteria use chemical reactions to modulate pre-infection behaviour such as surface attachment.
Polymers that can interfere with bacterial adhesion or the chemical reactions used for quorum sensing are therefore a
potential means to control bacterial population responses. Here, we report how polymeric ‘bacteria sequestrants’, designed
to bind to bacteria through electrostatic interactions and therefore inhibit bacterial adhesion to surfaces, induce the
expression of quorum-sensing-controlled phenotypes as a consequence of cell clustering. A combination of polymer and
analytical chemistry, biological assays and computational modelling has been used to characterize the feedback between
bacteria clustering and quorum sensing signalling. We have also derived design principles and chemical strategies for
controlling bacterial behaviour at the population level.
N
on-lethal means of targeting bacteria1,2
, such as the stimulation
of host immune systems3,4
, interference with cell
adhesion5,6
or bacterial communication7,8
, are emerging as
attractive means to avoid resistance against antimicrobial therapies.
In recent years, polymeric antimicrobials have increasingly been a
focus of attention due to their ability to present multiple functionalities
for detecting, binding and inactivating pathogens9–11
. There
are now examples of polymers that can prevent cell growth in
multi-drug-resistant strains11
, or that can sequester specific
bacteria12–14, toxins15,16 and/or cell-signal molecules17–19.
Particularly promising are materials that can prevent bacteria
binding to hosts5,6
, a prerequisite for most infections and particularly
those related to invasive pathogens19
. Two main strategies
have been exploited, the first utilizing antifouling surfaces to directly
inhibit bacterial adhesion20–22
, and the second displaying multiple
ligands that bind competitively to the surface of the bacteria, thus
inhibiting their attachment to host surface ligands12–14
. Depending
on the material design, one of the consequences of the second
approach is the aggregation of bacteria into clusters, a microenvironment
where diffusion of nutrients and signals can be
significantly affected.
A number of publications have now described the significant
effects of local concentration and spatial confinement, as well as
molecule and bacteria diffusion, on bacterial cell–cell communication
networks23–28
. Bacterial communication, also known as
quorum sensing29,30
, is an important regulator of bacterial behaviour,
including swarming, aggregation, production of exoenzyme
and toxins, as well as processes preceding infection such as
surface colonization and biofilm formation31–34. Quorum sensing
signalling in bacteria often involves complex feedback mechanisms
and is regulated by gene circuits and multiple interconnected
control mechanisms29,35
. This feedback between cell clustering
and quorum sensing signalling has stimulated intense debate
regarding the nature of quorum sensing and whether it is always a
population density response rather than a function of cell clustering
and signal diffusion36,37.
We recently reported preliminary data indicating that certain
polymers can modulate the luminescence of Vibrio harveyi, a
marine pathogen that responds to the quorum-sensing signal AI-2
by producing light. These materials were designed to cluster bacteria
while simultaneously reducing the concentration of AI-2, a component
of the quorum-sensing circuit of several bacteria38
. Unlike
conventional polymers that are able to bind only to the quorumsensing
signals, and therefore inhibit light production in a dosedependent
way, some of those polymers were able to induce luminescence
in V. harveyi under specific experimental conditions,
suggesting interdependence between bacteria clustering and the
quorum-sensing response39
.
We report here how a polymeric ‘bacteria sequestrant’, which
induces bacterial aggregation through electrostatic interactions
and has no functionalities to interfere with the quorum-sensing
signals, is able to induce quorum-sensing-related responses in a
range of bacteria. These include not only the model microorganism
V. harveyi, but also the human pathogens Escherichia coli and
Pseudomonas aeruginosa. We used synthetic and analytical chemistry,
biological assays and computational modelling to demonstrate
that quorum-sensing-associated behaviour occurs as a direct consequence
of bacteria clustering. Furthermore, the responses of
V. harveyi as a model organism were simulated and compared
against a representative ‘quorum quencher’, which should
only bind to quorum-sensing signals, and a ‘dual-action’ polymer,
with the ability to bind both the surface of bacteria and the
1
School of Computer Science, The University of Nottingham, University Park, Nottingham NG7 2RD, UK, 2
School of Pharmacy, The University of
Nottingham, University Park, Nottingham NG7 2RD, UK, 3
School of Molecular Medical Sciences, Centre for Biomolecular Sciences, The University of
Nottingham, University Park, Nottingham NG7 2RD, UK, 4
School of Chemistry, The University of Nottingham, University Park, Nottingham NG7 2RD, UK,
5
School of Computing Science, Newcastle University, Newcastle NE1 7RU, UK, †
Present address: School of Chemistry, University of Birmingham,
Birmingham B15 2TT, UK. *e-mail: f.fernandez-trillo@bham.ac.uk; natalio.krasnogor@ncl.ac.uk; cameron.alexander@nottingham.ac.uk
ARTICLES
PUBLISHED ONLINE: 10 NOVEMBER 2013 | DOI: 10.1038/NCHEM.1793
NATURE CHEMISTRY | VOL 5 | DECEMBER 2013 | www.nature.com/naturechemistry1058
© 2013 Macmillan Publishers Limited. All rights reserved.
signal molecules. The results give important insight into the
unexpected consequences of feedback between bacteria clustering
and quorum-sensing signalling. Furthermore, the data suggest
entirely new chemical design principles not only for novel
anti-adhesive materials, but also for inducing consequences
of quorum-sensing responses that are beneficial, such as
antibiotic production40,41
.
Results
The starting hypothesis was that polymeric materials with the ability
to aggregate bacteria into clusters would be able to induce the
expression of quorum-sensing-controlled phenotypes (Fig. 1a)39
.
We thus derived a model that predicted, from a phenomenological
point of view, induction of a feedback loop into quorum-sensing signalling
by bacteria clustering, interrelating polymer (P) concentration,
bacterial (B) aggregation and quorum-sensing signals (S).
Three classes of polymers were therefore defined in order to
predict all the potential interactions between polymers, bacteria
and signals: (1) bacteria sequestrants, which should only bind to
bacteria, inducing cell clustering? (2) quorum quenchers, which
would only be able to bind the signals? and (3) dual-action polymers,
with the ability to bind both signals and bacteria. The predicted
clustering and quorum-sensing responses were validated
against experimental data using V. harveyi and its AI-2 network
(Fig. 1b) as a model.
Poly(N-[3-(dimethylamino)propyl] methacrylamide) (P1), a cationic
polymer that should bind to the surface of bacteria through
electrostatic interactions, was synthesized as a representative bacteria
sequestrant. Controlled radical polymerizations (RAFT) were
used to tune the molar mass and the materials were characterized
by NMR and gel permeation chromatography (GPC). The behaviour
of bacteria in the presence of P1 was determined and compared
to model polymers of the other classes. Because AI-2 in V. harveyi is
a borate ester, and its concentration in solution can be reduced by
competitive binding to the boric acid precursor with polymeric
diols (Fig. 1b), commercially available poly(vinyl alcohol) (P2),
and poly(N-dopamine methacrylamide-co-N-[3-(dimethylamino)propyl]
methacrylamide) (P3)39 were chosen as representative
quorum quenchers and dual-action polymers, respectively (Fig. 1c).
The viability of V. harveyi in the presence of these polymers was
assessed by monitoring cell growth during luminescence experiments.
For the relevant duration of the experiment (0–8 h),
before solvent evaporation in the well plates becomes significant,
no differences were observed in the optical density of the cultures
in the absence and presence of increasing amounts of each
polymer (Supplementary Figs 9b–14b). In addition, the viability
of V. harveyi in the presence of P1, a polymer, which might be
expected to exhibit toxicity due to a higher content of tertiary
amines42,43
, was also investigated using nuclear staining and fluorescent
microscopy. When compared against cultures in the
O
HO
HO
OHHO
O
HO
HO
O
B
O
HO OH
OH
OH
O
B
O OH
OH
O
O
OH
OH
HN O
OH
OH
OHN
H
N
n m
OHN
H
N
n
OH
n
OH
P3
Model dual action
P2
Model quorum
quencher
P1
Model bacteria
sequestrant
+ B(OH)3
'inactive'
'active'
DPD
AI-2
1
2
3
= Bacterial surface receptor
Cell clustering
a
b c
QS activation
Anti-adhesive polymer
= Bacterial QS signal receptor = QS signal
Figure 1 | Quorum sensing induction in the AI-2 network. a, Schematic representation of quorum sensing (QS) activation by bacteria sequestrants that
promote bacteria clustering. Step 1: polymer binds to the surface of the bacteria via multivalent interactions. Step 2: bacteria are crosslinked as polymer
interacts with different bacteria. Step 3: signal diffusion is limited, maintaining a high concentration within the cell cluster. b, Key components of the AI-2
network, including 4,5-dihydroxy-2,3-pentanedione (DPD) and the active species formed in the presence of B(OH)3 in the media, and the mechanism of AI-2
quenching by competitive binding with diols. c, Structure of the polymers used in this work: poly(N-[3-(dimethylamino)propyl] methacrylamide) (P1),
poly(vinyl alcohol) (P2) and poly(N-dopamine methacrylamide-co-N-[3-(dimethylamino)propyl] methacrylamide) (P3).
NATURE CHEMISTRY DOI: 10.1038/NCHEM.1793 ARTICLES
NATURE CHEMISTRY | VOL 5 | DECEMBER 2013 | www.nature.com/naturechemistry 1059
© 2013 Macmillan Publishers Limited. All rights reserved.
absence of polymer (positive control) and cultures in the presence
of methanol (negative control), the ratio between viable (green)
and non-viable (red) bacteria in the presence of P1 was similar to
that of the positive control (untreated bacteria) and significantly
different from the negative control (Supplementary Fig. 27). This
indicates that P1 was not altering quorum sensing through a
direct toxic response.
The ability of the polymers to cluster bacteria and their effect
over quorum-sensing networks was investigated in cultures of two
strains of V. harveyi (MM32 and BB170). Initial experiments were
carried out with V. harveyi MM32, which responds to exogenous
AI-2 but does not produce the quorum-sensing signal precursor
4,5-dihydroxy-2,3-pentanedione (DPD). Concurrent with the
experimental assays, polymer–bacteria interactions were simulated
based on a simple affinity model. Cell aggregation experiments
(Fig. 2a) showed a good match with the computationally predicted
results (Fig. 2b), with P1 inducing rapid bacterial clustering, P2 producing
no apparent difference compared to bacterial suspensions
alone, and P3 forming aggregates with bacteria at a similar rate to
P1. Computationally predicted results were simulated ten times
using different randomizations in order to obtain statistically
reliable results. Effects were consistent within the ten simulations.
Initial conditions (cell positioning, random seeds, affinities and
polymer concentrations) for simulations with different polymers
were identical (Fig. 2b, Supplementary Figs 28–30).
We then considered the effects of clustering on the quorumsensing
response as reported by luminescence. Taking into consideration
feasible diffusion rates and affinities for the interactions
between bacteria, signals and polymers, we predicted changes in
luminescence with the addition of bacteria sequestrants (P1),
quorum quenchers (P2) and dual-action polymers (P3). As is
apparent from Fig. 2c, when compared to a control in the absence
1.0
KPB = 5, KPS = 0 a.u. KPB = 5, KPS = 10 a.u.KPB = 0, KPS = 0.1 a.u.
FoldchangeFoldchange
1.0
Foldchange
1.0
Foldchange
Time (a.u.) Time (a.u.) Time (a.u.)
Concentration (0–70,000 a.u.) increases with the arrow direction
0
0.5
0.7
1.0
2.0
3.0
30
a
c
P1 P2 P3
b
20
10
Clusterdiameter(µm)
0
1.5
Foldchange
0.5
0.7
1.0
2.0
1.5
Foldchange
0.5
0.7
1.0
2.0
1.5
2 4 6 8 10
Time (h)
0 10 20 30 40 50 60
Time (min)
0 2 4 6 8 10
Time (h)
Time (a.u.)
0 2 4 6 8 10
Time (h)
Control P1 P2 P3
P1 0.25 mg ml–1 P2 0.5 mg ml–1 P3 0.25 mg ml–1
Control 0.5
1.0 1.5 mg ml–1
Control
0.125 0.25
0.025 0.05
0.5 mg ml–1
Control 0.1250.05
0.25 0.5 mg ml–1
P1
P2
P3
Control
Figure 2 | Effect of polymers on V. harveyi MM32 behaviour. a,b, Aggregation of bacteria in the presence of polymers, measured in a Coulter Counter and
by optical microscopy (a), were in good agreement with those predicted by the computational stochastic simulation, observed from simultaneous
screenshots of simulated bacteria cultures in the absence and presence of polymers (b). Mean value and polydispersity index are reported. c, Quorumsensing
signalling in the presence of polymers, measured by luminescence (top row), was in good agreement with that predicted by the computational
model (bottom row). Bacteria sequestrants (P1) enhanced luminescence (fold change ≥ 1) throughout the duration of the experiment (left). Quorum
quenchers (P2) reduced luminescence (fold change ≤ 1) during the same timeframe (middle). For dual-action polymers (P3), both induction and quenching
were observed (right). Mean value and standard deviation are reported. Two-way Anova analysis of experimental results indicates that significant differences
in fold change are observed as polymer concentration increases, for the relevant duration of the experiment (0–8 h). See Supplementary Figs 2, 7 and 9–11
for further details.
ARTICLES NATURE CHEMISTRY DOI: 10.1038/NCHEM.1793
NATURE CHEMISTRY | VOL 5 | DECEMBER 2013 | www.nature.com/naturechemistry1060
© 2013 Macmillan Publishers Limited. All rights reserved.
of polymers, P1 induced an increase in light production in
V. harveyi MM32 cultures throughout the duration of the experiment
(fold change in luminescence ≥ 1), despite not being targeted
to quorum sensing and lacking the functionalities to interfere with
the signals. P2 was able to reduce luminescence during the same
time (fold change ≤ 1). As expected, P3 showed a dual mode of behaviour,
with the ability either to enhance or reduce light production
depending on the specific polymer concentration and
time (bacteria density). For P1, the absolute change in luminescence
was higher at later stages of the experiments, when cell numbers
were higher (Supplementary Fig. 8), but the relative difference
(fold change) in luminescence was higher at earlier stages. Relative
variations in luminescence at early stages of bacterial growth can
appear exaggerated in cases of low initial values of luminescence,
because the timescale for aggregation is considerably smaller than
that for light production. Therefore, the effects of cell clustering
were most apparent at an early time in the experiment, and
became less pronounced as bacterial growth matched the
density and viscosity within clusters. Thus, at early time periods,
it was expected that slow diffusion of signals from the cell clusters
enabled bacteria to sense a higher concentration of quorumsensing
signals more rapidly. Indeed, during the key timescales of
the experiment (that is, 0–8 h, after which cell numbers increased
markedly), the effect of polymer on quorum-sensingcontrolled
luminescence matched well that predicted by the
theoretical model.
To evaluate further the feedback between the ability of polymers
to induce aggregation and quorum-sensing-controlled light production,
several simulations were performed in which the affinities towards
bacteria (KPB) and signals (KPS) of P3 were systematically varied.
Affinities were investigated over a three order of magnitude range,
and different combinations of KPB and KPS were simulated
(see Supplementary Figs 31–39 for further details). As can be seen
in Fig. 3, the potential of a polymer to inhibit or enhance light
production was highly dependent on polymer concentration,
experiment time and the polymer affinities for signals and bacteria.
Variations in any of these parameters were predicted to lead to, and
indeed showed, marked changes in quorum-sensing signalling,
manifest in the light production. For instance, polymers with a
high affinity for bacteria enhanced light production, regardless of
their concentration and the time of the analysis, even if they
showed a high affinity towards the signals.
The effects bacterial density and growth rate have on the activity
of dual-action polymers were also investigated. Polymer affinities for
bacteria and signals were fixed and simulations with different initial
densities of bacteria B0 or different growth rates, as expressed by
bacteria doubling time T, were performed (see Supplementary
Figs 40–42 for further details). When the initial density of bacteria
B0 was reduced by an order of magnitude (Supplementary Fig. 41),
P3 was able to induce quorum-sensing signalling throughout the
duration of the simulation and regardless of the concentration of
polymers, as opposed to the dual action exhibited when the starting
number of bacteria was higher (Supplementary Fig. 40). Similarly,
the ability of P3 to induce or inhibit light production was significantly
affected in the presence of bacteria growing at different
rates (compare Supplementary Figs 40 and 42).
Foldchange
Foldchange
Quenching Control Enhancement
Log KPS (a.u.) Log KPS (a.u.) Log KPS (a.u.)
LogKPB(a.u.)
LogKPB(a.u.)
LogKPB(a.u.)
KPB
(a.u.)
KPB
(a.u.)
KPB
(a.u.)[P] (a.u.)
[P] (a.u.)
[P] (a.u.)
1.0
Early time Mid time Late time
1.0
Foldchange
1.0
Figure 3 | Effect of relative binding affinities on light production. The effect of polymer affinity for signals (KPS) and bacteria (KPB) on light production was
predicted by the model. In the presence of weak polymer–bacteria interactions, dual-action polymers (P3) quench light production, regardless of the polymer
concentration [P] and time at which light production is evaluated (top). As KPB increases, the overall outcome of polymer interference changes, and
enhancement of light production is expected at higher polymer concentrations. In addition, overall light production depends on the relative intensities
(KPB and KPS) of both affinities as well as the time at which light production is evaluated (bottom). To obtain dual-action polymers that consistently quench
light production, polymers with low KPB are required. Initial conditions (cell positioning, random seeds) for the simulations are identical. Times of 5,000,
15,000 and 30,000 a.u. were selected as representative early, mid and late times, respectively, for the simulations. Top: KPS ¼ 0.1 a.u. was selected as a
representative value. See Supplementary Figs 31–39 for further details.
NATURE CHEMISTRY DOI: 10.1038/NCHEM.1793 ARTICLES
NATURE CHEMISTRY | VOL 5 | DECEMBER 2013 | www.nature.com/naturechemistry 1061
© 2013 Macmillan Publishers Limited. All rights reserved.
The interaction of polymers and bacteria was also investigated
using the V. harveyi BB170 strain, which is capable of producing
DPD. In this case, light production in the absence of polymers
shows two phases (Fig. 4a). In the first phase, luminescence
decreased as the bacteria responded to the lower concentrations of
DPD in the sample media as compared to the pre-culture
Decay
phase
Enhancement
phase
8.0a
b
P1 P2 P3
6.0
4.0
2.0
Logofluminescence
Minimum luminescence (Lummin)
Lummin
Lummin Lummin
0.0
0.1
1
10
1.0
3.0 3 2
3
4
5
6
0 0 0.01 0.05 0.1 0.3 1.012.5 25 50
[P] (µg ml–1) [P] (mg ml–1)
125 250 0 12.5 25 50
[P] (µg ml–1)
125 250500
3.5
4.0 4
5.0
5
4.5
Recoverytime(h)
Recoverytime(h)
Recoverytime(h)
KPB = 5, KPS = 0 a.u. KPB = 5,
KPS = 10 a.u.KPB = 0, KPS = 0.1 a.u.
100
Control 0.0125 0.025
0.125
0.05
0.25 0.5 mg ml–1
Foldchange
0.01
0.1
1
10
Foldchange
0.01
0.1
1
10
Foldchange
Foldchange
1.0 1.0
Foldchange
Foldchange
0 2 4 6 8 10 12 14
Time (h)
0 2 4 6 8 10
Time (h)
0 2 4 6 8 10
Time (h)
0 2 4 6 8 10
Time (h)
Time (a.u.) Time (a.u.) Time (a.u.)
Concentration (0–70,000 a.u.) increases with the arrow direction
Initial light production
Control 0.01 0.05
0.30.1 1.5 mg ml–1
Control 0.0125 0.025
0.1250.05 0.25 mg ml–1
Figure 4 | Effect of polymers on V. harveyi BB170 luminescence. a, Luminescence in V. harveyi BB170 shows two phases: a decay phase where luminescence
is reduced and an enhancement phase, after the minimum concentration of AI-2 for luminescence induction is reached. b, Polymer interference with quorumsensing
signalling, as measured by luminescence (top row), was in good agreement with that modelled (middle row). Bacteria sequestrants (P1) enhanced
luminescence (fold change ≥ 1) (left); quorum quenchers (P2) reduced light production (fold change ≤ 1) (middle); and for dual-action polymers (P3) both
induction and quenching of luminescence were observed (right). Additionally, differences in the time necessary to recover the initial intensity of luminescence
could be observed (bottom row). P1 induced earlier light production and P2 delayed the recovery time. In this case, P3 behaved as a quorum quencher and
delayed the onset of luminescence. This behaviour was well predicted by the model (bottom row, insets). Mean value and standard deviation are reported.
Two-way Anova analysis of experimental results indicates that significant differences in fold change are observed, as polymer concentration increases, for the
relevant duration of the experiment (0–8 h). See Supplementary Figs 12–17 for further details.
ARTICLES NATURE CHEMISTRY DOI: 10.1038/NCHEM.1793
NATURE CHEMISTRY | VOL 5 | DECEMBER 2013 | www.nature.com/naturechemistry1062
© 2013 Macmillan Publishers Limited. All rights reserved.
medium, before their production of endogenous DPD. When a
threshold of DPD concentration was achieved, a new phase
was attained wherein light production increased as a function of
DPD concentration.
Despite the differences in light production profile for both strains,
the effect that all three classes of polymers had on BB170 quorumsensing
signalling (Fig. 4b, top and middle), as measured by luminescence,
was very similar to that described for MM32 (Fig. 2c, top).
Throughout the duration of the experiments, bacteria sequestrants
(P1) were able to induce light production, quorum quenchers (P2)
reduced the overall production of light, and dual-action polymers
(P3) showed both induction and quenching of luminescence. The
effect was weaker during the decay phase (time lower than the time
required for Lummin), as the concentration of AI-2 will be well
below the detection threshold for most of this period.
The effect of polymers on BB170 quorum-sensing signalling was
also reflected in the duration of the decay phase and the time taken
for bacteria to sense a concentration of AI-2 above the threshold.
During this decay phase, light production was reduced because luciferase
production was switched off while bacteria re-adapted to the low
concentration of AI-2 after dilution. As the population of bacteria
increased, the amount of AI-2 in solution increased accordingly, so
that quorum-sensing signalling could recover. This effect was easily
monitored from the light production plots by measuring the time
taken by the bacterial suspension to recover a significant level of light
intensity, for instance, the initial value of light production (Fig. 4a).
Interestingly, by confining cells into clusters, bacteria sequestrants
(P1) were able to induce an earlier activation of quorumsensing
signalling as a consequence of the local higher concentration
of AI-2. Conversely, quorum quenchers (P2) were able to delay the
time needed to do so, as the concentration of AI-2 in solution was
reduced. For dual-action polymers (P3), a combination of both
effects was expected. In the reported example, P3 showed an
overall quenching effect, increasing the time needed for the recovery
of the initial luminescence (Fig. 4b, bottom).
To investigate the generality of this effect and the potential for
directing quorum-sensing-controlled phenotypes in relevant human
pathogens, further experiments were performed with E. coli and
P. aeruginosa. Escherichia coli lux-based acylhomoserine lactone
(HSL) biosensors, JM109 pSB1075 (ref. 44), and JM109 pSB536
(ref. 45), which produce light in response to N-(3-oxododecanoyl)L-homoserine
lactone (3O-C12-HSL)46 and N-butyryl-L-homoserine
lactone (C4-HSL)47
, respectively, were selected as representative
E. coli strains. Pseudomonas aeruginosa PA01 pqsA CTX-pqsA::lux48
,
which produces light in response to 2-heptyl-3-hydroxy-4(1H)-quinolone,
usually termed the Pseudomonas quinolone signal (PQS)49, was
selected as a representative P. aeruginosa strain.
Aggregation of these strains in the presence of P1 was very fast,
and a dose-dependent increase in optical density of the cultures
could be observed as soon as P1 was added (Supplementary Figs
19–26, time 0). In addition, the growth of both E. coli strains in the
presence of P1 was significantly compromised, notably in the case
of pS536 reporter. This lack of growth had an impact on luminescence
production for these strains. As P1 concentration increased, the production
of light decreased, in agreement with the decreased viability of
E. coli in the presence of P1. With higher polymer concentrations,
recovery of light production was observed as a consequence of light
induction being triggered by clustering (Supplementary Figs 20 and
22). P. aeruginosa showed better viability in the presence of P1, and
induction of light production in the presence of bacteria sequestrant
P1 was clearly observed, particularly at higher polymer concentrations
(Supplementary Figs 23–26).
Discussion
The initial finding that polymers intended to suppress quorum
sensing could in fact enhance cell signalling (as reported by light
production)39
was unexpected and suggested that bacteria confinement
into clusters could be responsible for quorum-sensing induction,
thus producing an effect opposite to that desired.
Spatial confinement is inherent to quorum sensing as cell density
helps regulate bacteria behaviour. Significantly, recent publi-
cations23,24,28
have suggested that a confined individual bacterium
can show quorum-sensing-type behaviour. This behaviour has
been computationally anticipated50 in studies, suggesting that programmable
compartmentalization is a Turing-complete mechanism,
and can thus be a potentially useful tool for the control of
population responses in synthetic biology. In addition, as noted in
the case of Pseudomonas species, quorum-sensing signal gradients
have a context-dependent action, with limited effects on biofilm
growth in liquid culture but pronounced and significant effects on
confined cell communities attached to surfaces, that is, when bacteria
and signals are in close proximity27
.
Therefore, to understand how quorum sensing could be activated
by bacteria sequestrants, we derived a phenomenological synthetic
biology model to simulate and predict quorum-sensing-controlled
luminescence as a function of binding affinities towards bacteria
and signals, in two different mutants of V. harveyi. We used the
MM32 strain, which responds to, but cannot produce, its
quorum-sensing signal (DPD, the AI-2 precursor), and the BB170
strain, which is capable of synthesizing DPD and thus introduces
natural variability and nonlinearity into the system.
We synthesized a model bacteria sequestrant, P1, intended to
bind to the surface of bacteria through electrostatic interactions.
The ability of P1 to aggregate bacteria into clusters was confirmed
by measuring cluster size and via optical microscopy. In addition,
P1 induced light production throughout the incubation assay,
despite not being targeted to quorum sensing and having no specific
functionalities to interfere with the signals. Taking into account the
ability of the polymers to cluster bacteria, the predicted relative affinity
of the monomer units within the polymers for the bacteria39
,
the behaviour of V. harveyi in the presence of P1 was well predicted
by the computational model.
In contrast, the behaviour of V. harveyi in the presence of P2, a
model quorum quencher, was markedly different to that in the presence
of P1. In addition, a combination of the two responses could be
observed when using P3. We termed P3 a dual-action polymer as it
incorporated both the cationic groups that bind to the surface of
bacteria, inducing cluster formation, and diols capable of binding
the boronic acid needed to activate AI-2. The numbers of
monomer units, that is, the components in each repeating section
of the polymers able to ‘bind’ signals or cells, were broadly similar
across P1 to P3 (degrees of polymerization, 100–400), and no significant
differences in response were observed when P1 with different
molecular weight was used (Supplementary Figs 6 and 18).
Nevertheless, from a phenomenological point of view, there was
good correlation between simulated and experimental quorumsensing
responses in the presence of these polymers.
The goal of the model was not only to understand the feedback
between aggregation and light production, but also to derive design
principles for quorum-sensing control. We therefore performed a
series of simulations where the relative affinities of P3 towards bacteria
and signal were systematically varied. As can be seen in Fig. 3,
to design efficient polymeric materials that can cluster bacteria,
dual-action polymers have to be considered where the balance
between the affinity towards the bacteria (KPB) and the affinity
towards the signal (KPB) prevents induction of quorum-sensingcontrolled
phenotypes (green to blue colour in the graphs). In a
similar way, the model predicted, and experiments showed, that
quorum quenchers designed to reduce the expression of quorumsensing-controlled
behaviour should also exhibit a very low affinity
for bacterial surfaces in order to retain their intended effects on bacterial
populations. Most notably, the models and experiments
NATURE CHEMISTRY DOI: 10.1038/NCHEM.1793 ARTICLES
NATURE CHEMISTRY | VOL 5 | DECEMBER 2013 | www.nature.com/naturechemistry 1063
© 2013 Macmillan Publishers Limited. All rights reserved.
showed that small changes in initial bacterial density and growth
rates could tip the balance to strongly opposing effects, so either
luminescence enhancement or quenching could be seen for the
same polymer under very similar conditions. This variability in conditions
is likely to be most apparent in therapeutic applications of
polymers, where the numbers and growth for pathogens will
differ significantly across patients, or the degree of infection.
The model developed herein was designed to be ‘agnostic’ to the
nature of the bacteria, as well as the type of response triggered by
quorum sensing. In principle, therefore, any bacteria behaviour
under quorum-sensing control, such as the production of exoenzymes
and toxins, or biofilm formation31,32,34
, could be triggered if
cell clustering is induced by bacteria sequestrants.
We therefore investigated whether the reported enhancement of
light production by a bacteria sequestrant could also be detected
using different bacteria and different signalling molecules.
Experiments under the same conditions optimized for V. harveyi
were performed using E. coli and P. aeruginosa luminescence reporters
for HSLs and PQS, respectively. These signals are significantly
different to AI-2 in terms of their chemical functionality, and the
quorum-sensing response of the microorganisms is not synchronized
in the way that V. harveyi responds to AI-2. Despite these
differences, and the lower viability of E. coli strains in the presence
of P1, the ability of this polymer to induce light production as a consequence
of aggregation was also observed, establishing the generality
both of the quorum sensing/polymer/bacteria feedback model
and the mechanism of activity of the dual-action polymers.
Conclusions
In conclusion, we have shown how polymeric bacteria sequestrants,
with high affinity for bacterial surfaces, have the ability to interfere
with non-targeted signalling pathways such as quorum sensing in a
range of prokaryotes. We have defined a theoretical and practical
framework for understanding bacterial responses to quorumsensing
interference in the presence of polymers with the ability
to bind bacteria and/or signalling molecules, which should aid the
development of novel non-antibiotic anti-infectives.
Given that many bacteria attach to host surfaces before colonization
and invasion, our data suggest that materials designed to interfere
with infection pathways should be designed so that they do not
promote unwanted effects in cell signalling and quorum sensing.
Significantly, the results show that materials that promote bacteria
clustering induce unexpected responses in quorum-sensing-controlled
phenotypes, and that these responses can be better modulated
through control of the affinity towards both bacteria and
signals. As a corollary, the combined model/experiment approach
enables experimental data to be obtained regarding spatial effects
on quorum sensing, which can be interrogated through computational
models, which in turn can feed back into materials
design. This combined chemistry/computation approach should
enhance our understanding of quorum sensing in complex environments.
In turn, the ability to utilize specific chemical design principles
to control cell behaviour should facilitate the development
of antimicrobials that avoid selection pressure and inform synthetic
biology strategies wherein quorum sensing is used to induce the production
of valuable metabolites.
Methods
Aggregation assay. A single colony of V. harveyi grown on Luria Bertani (LB) agar
plates was used to inoculate 2 ml LB medium containing chloramphenicol
(10 mg ml21
), and kanamycin (50 mg ml21
) in the case of BB170. The bacteria were
grown with aeration at 30 8C overnight. Boron depleted assay broth (AB) medium
was then inoculated with this preculture to give a bacterial suspension with an
optical density (600 nm; OD600) of 1.0. Aliquots of this culture were then mixed
with known volumes of stock solutions of polymers in Dulbecco’s phosphate buffer
saline (DPBS). The values of polymer concentration reported for the aggregation
experiments correspond to the polymer concentrations in these suspensions. To
measure cluster size, these bacterial suspensions were added to a Coulter Counter
flow cell filled with H2O (,14 ml) to obtain an obscuration of 8–12%. Cluster size
was then measured at different time intervals. For optical microscopy analysis,
aliquots (10 ml) of the bacterial suspensions, in the absence and presence of
polymers, were collected after 60 min, mounted on a glass slide with a coverslip
on top, and examined with an optical microscope. See Supplementary Figs 1–8 for
further details.
Microbiological assays. A single colony of V. harveyi grown on LB agar plates was
used to inoculate 2 ml LB medium containing chloramphenicol (10 mg ml21
), and
kanamycin (50 mg ml21
) in the case of BB170. The bacteria were grown with
aeration at 30 8C overnight. Boron depleted AB medium was then inoculated with
this preculture (5,000:1). For MM32, boric acid was added to a final concentration of
400 mM, and DPD was added to a final concentration of 22 mM. For BB170,
boric acid was added to a final concentration of 22 mM. A volume of 180 ml of
the inoculated medium was placed in each of the wells of a 96-well plate and
combined with 20 ml of the samples to be analysed. Each compound was tested over
at least three different concentrations. Light production and OD600 were recorded at
30 8C every 30 min for at least 10 h in a 96-well plate, after which time solvent
evaporation became a significant issue. The experiments were carried out in
triplicate and the plotted curves were derived from the mean value. The normalized
luminescence was calculated by dividing the light output by the optical density at
each time point.
Simulation methods. Because of the spatial and time scales of the system, a
mesoscopic lattice-based model and an agent-based approach were used. Analysis of
the results was carried out at the phenomenological level, that is, by capturing the
characteristic effect of the three types of polymers. Modelled parameters were
subsequently refined against measurable overall effects (that is, bacterial binding and
luminescence production). The starting boundary conditions for the model were set
so that any deviations from control experiments were caused by the polymers
manipulating the immediate extracellular environment of the bacteria. Three types
of object were considered in the model: bacteria (B), polymers (P) and signal
molecules (S). The size of each B was fixed to occupy a square of 2 × 2 arbitrary
lattice spaces. The sizes of S and P were considered to be negligibly small. One unit
lattice space could thus contain a quarter of B and unlimited numbers of S and P.
Two types of change were considered to take place in the system: chemical binding
and diffusion of the objects. The reactions and diffusion of the different species
were modelled using a Gillespie algorithm.
Three types of binding reaction were delineated in the model: binding between S
and B, binding between B and P, and binding between P and S. Each interaction was
considered to be reversible. In addition, there were delimiting conditions for the
model: (1) B has separate binding sites for S and P; (2) each quadrant of B has a
number of S binding sites, denoted as BS, and a number of P binding sites, BP.
Similarly, each P has PS binding sites for S and two B binding sites, PB. Each S could
thus only bind to one binding site (either B or P) and would not be available for
other reactions once bound, until the reverse reaction occurred and the molecules
and binding sites again became free. The same constraint was applied for P–B
binding. We considered that S could only bind with P or B in the same lattice, and
that bound S would move with P or B without further activation of the quorumsensing
network. An individual P was considered to bind with B inside the same
lattice space. For S, once a single P was bound to a single B, they were considered to
be fixed in a binding interaction over the timescale of the experiment. P bound to B
were able to bind with other B in the neighbouring lattice. In such a case, the B–P–B
complex moved as a single unit until associative interactions were lost. This
mechanism represents key multiple binding interactions that could lead to bacterial
clustering and aggregation. The values of binding affinities and diffusion rates
are dependent on the concentrations of the different objects in the local environment
(see Supplementary Sections 28–42 for further details).
Received 17 March 2013; accepted 4 October 2013;
published online 10 November 2013
References
1. Clatworthy, A. E., Pierson, E. & Hung, D. T. Targeting virulence: a new paradigm
for antimicrobial therapy. Nature Chem. Biol. 3, 541–548 (2007).
2. Nathan, C. Fresh approaches to anti-infective therapies. Sci. Transl. Med. 4,
140sr2 (2012).
3. Hancock, R. E. W., Nijnik, A. & Philpott, D. J. Modulating immunity as a
therapy for bacterial infections. Nature Rev. Microbiol. 10, 243–254 (2012).
4. Wei, P. et al. Bacterial virulence proteins as tools to rewire kinase pathways
in yeast and immune cells. Nature 488, 384–388 (2012).
5. Klemm, P., Vejborg, R. M. & Hancock, V. Prevention of bacterial adhesion.
Appl. Microb. Biotechnol. 88, 451–459 (2010).
6. Bricarello, D. A., Patel, M. A. & Parikh, A. N. Inhibiting host–pathogen
interactions using membrane-based nanostructures. Trends Biotechnol.
30, 323–330 (2012).
7. Kaufmann, G. F., Park, J. & Janda, K. D. Bacterial quorum sensing: a new target
for anti-infective immunotherapy. Exp. Opin. Biol. Ther. 8, 719–724 (2008).
ARTICLES NATURE CHEMISTRY DOI: 10.1038/NCHEM.1793
NATURE CHEMISTRY | VOL 5 | DECEMBER 2013 | www.nature.com/naturechemistry1064
© 2013 Macmillan Publishers Limited. All rights reserved.
8. Park, J. et al. Infection control by antibody disruption of bacterial quorum
sensing signaling. Chem. Biol. 14, 1119–1127 (2007).
9. Hancock, R. E. W. & Sahl, H.-G. Antimicrobial and host–defense peptides
as new anti-infective therapeutic strategies. Nature Biotechnol. 24,
1551–1557 (2006).
10. Nederberg, F. et al. Biodegradable nanostructures with selective lysis of microbial
membranes. Nature Chem. 3, 409–414 (2011).
11. Liu, L. et al. Self-assembled cationic peptide nanoparticles as an efficient
antimicrobial agent. Nature Nanotech. 4, 457–463 (2009).
12. Alexander, C., Pasparakis, G. & Cockayne, A. Control of bacterial
aggregation by thermoresponsive glycopolymers. J. Am. Chem.Soc. 129,
11014–11015 (2007).
13. Shepherd, J. et al. Hyperbranched poly(NIPAM) polymers modified with
antibiotics for the reduction of bacterial burden in infected human tissue
engineered skin. Biomaterials 32, 258–267 (2011).
14. Lee, D-W., Kim, T., Park, I-S., Huang, Z. & Lee, M. Multivalent nanofibers of
a controlled length: regulation of bacterial cell agglutination. J. Am. Chem. Soc
134, 14722–14725 (2012).
15. Rai, P. et al. Statistical pattern matching facilitates the design of polyvalent
inhibitors of anthrax and cholera toxins. Nature Biotechnol. 24, 582–586 (2006).
16. Piletska, E. V. et al. Attenuation of Vibrio fischeri quorum sensing using
rationally designed polymers. Biomacromolecules 11, 975–980 (2010).
17. Bush, K. et al. Tackling antibiotic resistance. Nature Rev. Microbiol. 9,
894–896 (2011).
18. Garner, A. L. et al. A multivalent probe for AI-2 quorum-sensing receptors.
J. Am. Chem. Soc. 133, 15934–15937 (2011).
19. Cossart, P. & Sansonetti, P. J. Bacterial invasion: the paradigms of enteroinvasive
pathogens. Science 304, 242–248 (2004).
20. Hook, A. L. et al. Combinatorial discovery of polymers resistant to bacterial
attachment. Nature Biotechnol. 30, 868–875 (2012).
21. Smith, R. S. et al. Vascular catheters with a nonleaching poly-sulfobetaine
surface modification reduce thrombus formation and microbial attachment.
Sci. Transl. Med. 4, 153ra132 (2012).
22. Epstein, A. K., Wong, T.-S., Belisle, R. A., Boggs, E. M. & Aizenberg, J.
Liquid-infused structured surfaces with exceptional anti-biofouling
performance. Proc. Natl Acad. Sci. USA 109, 13182–13187 (2012).
23. Kastrup, C. J. et al. Spatial localization of bacteria controls coagulation of human
blood by ‘quorum acting’. Nature Chem. Biol 4, 742–750 (2008).
24. Carnes, E. C. et al. Confinement-induced quorum sensing of individual
Staphylococcus aureus bacteria. Nature Chem. Biol. 6, 41–45 (2010).
25. Alberghini, S. et al. Consequences of relative cellular positioning on quorum
sensing and bacterial cell-to-cell communication. FEMS Microbiol. Lett.
292, 149–161 (2009).
26. Goryachev, A. B. Understanding bacterial cell–cell communication with
computational modeling. Chem. Rev. 111, 238–250 (2011).
27. Flickinger, S. T. et al. Quorum sensing between Pseudomonas aeruginosa
biofilms accelerates cell growth. J. Am. Chem. Soc. 133, 5966–5975 (2011).
28. Boedicker, J. Q., Vincent, M. E. & Ismagilov, R. F. Microfluidic confinement
of single cells of bacteria in small volumes initiates high-density behavior of
quorum sensing and growth and reveals its variability. Angew. Chem. Int. Ed.
48, 5908–5911 (2009).
29. Lowery, C. A., Dickerson, T. J. & Janda, K. D. Interspecies and interkingdom
communication mediated by bacterial quorum sensing. Chem. Soc. Rev.
37, 1337–1346 (2008).
30. Atkinson, S. & Williams, P. Quorum sensing and social networking in the
microbial world. J. R. Soc. Interface 6, 959–978 (2009).
31. Miller, M. B. & Bassler, B. L. Quorum sensing in bacteria. Annu. Rev. Microbiol.
55, 165–199 (2001).
32. Camilli, A. & Bassler, B. L. Bacterial small-molecule signaling pathways. Science
311, 1113–1116 (2006).
33. Williams, P. Quorum sensing, communication and cross-kingdom signaling
in the bacterial world. Microbiology 153, 3923–3938 (2007).
34. Williams, P., Winzer, K., Chan, W. C. & Camara, M. Look who’s talking:
communication and quorum sensing in the bacterial world. Phil. Trans. R. Soc. B
362, 1119–1134 (2007).
35. Antunes, L. C. M. & Ferreira, R. B. R. Intercellular communication in bacteria.
Crit. Rev. Microbiol. 35, 69–80 (2009).
36. Fuqua, C. & Greenberg, E. P. Listening in on bacteria: acyl-homoserine lactone
signaling. Nature Rev. Mol. Cell Biol. 3, 685–695 (2002).
37. Redfield, R. J. Is quorum sensing a side effect of diffusion sensing? Trends
Microbiol. 10, 365–370 (2002).
38. Federle, M. J. & Bassler, B. L. Interspecies communication in bacteria. J. Clin.
Invest. 112, 1291–1299 (2003).
39. Xue, X. et al. Synthetic polymers for simultaneous bacterial sequestration and
quorum sense interference. Angew. Chem. Int. Ed. 50, 9852–9856 (2011).
40. Barnard, A. M. L. et al. Quorum sensing, virulence and secondary metabolite
production in plant soft-rotting bacteria. Phil. Trans. R. Soc. B 362,
1165–1183 (2007).
41. Fineran, P. C., Slater, H., Everson, L., Hughes, K. & Salmond, G. P. C.
Biosynthesis of tripyrrole and beta-lactam secondary metabolites in Serratia:
integration of quorum sensing with multiple new regulatory components in the
control of prodigiosin and carbapenem antibiotic production. Mol. Microbiol.
56, 1495–1517 (2005).
42. Kenawy, E.-R., Worley, S. D. & Broughton, R. The chemistry and applications
of antimicrobial polymers: a state-of-the-art review. Biomacromolecules 8,
1359–1384 (2007).
43. Engler, A. C. et al. Emerging trends in macromolecular antimicrobials to
fight multi-drug-resistant infections. Nano Today 7, 201–222 (2012).
44. Winson, M. K. et al. Construction and analysis of luxCDABE-based plasmid
sensors for investigating N-acyl homoserine lactone-mediated quorum
sensing. FEMS Microbiol. Lett. 163, 185–192 (1998).
45. Swift, S. et al. Quorum sensing in Aeromonas hydrophila and Aeromonas
salmonicida: identification of the LuxRI homologs AhyRI and AsaRI and
their cognate N-acylhomoserine lactone signal molecules. J. Bacteriol.
179, 5271–5281 (1997).
46. Pearson, J. P. et al. Structure of the autoinducer required for expression
of Pseudomonas aeruginosa virulence genes. Proc. Natl Acad. Sci. USA 91,
197–201 (1994).
47. Winson, M. K. et al. Multiple N-acyl-L-homoserine lactone signal molecules
regulate production of virulence determinants and secondary metabolites in
Pseudomonas aeruginosa. Proc. Natl Acad. Sci. USA 92, 9427–9431 (1995).
48. Diggle, S. P. et al. The Pseudomonas aeruginosa 4-quinolone signal molecules
HHQ and PQS play multifunctional roles in quorum sensing and iron
entrapment. Chem. Biol. 14, 87–96 (2007).
49. Pesci, E. C. et al. Quinolone signaling in the cell-to-cell communication
system of Pseudomonas aeruginosa. Proc. Natl Acad. Sci. USA 96,
11229–11234 (1999).
50. Krasnogor, N. et al. An appealing computational mechanism drawn
from bacterial quorum sensing. Bull. Eur. Assoc. Theor. Comput. Sci. 85,
135–148 (2005).
Acknowledgements
The authors acknowledge the UK Engineering and Physical Sciences Research Council and
Biotechnology and Biological Sciences Research Council (grants EP/G042462/1,
EP/H005625/1, EP/D022347/1, EP/D021847/1, EP/I031642/1, EP/J004111/1 and
BB/F01855X/1) and the University of Nottingham for funding. The authors thank
B. Bassler (Department of Molecular Biology, Princeton University) for the gift of
V. harveyi and P. Williams and S. Diggle (University of Nottingham) for the gift of E. coli
and P. aeruginosa strains. The authors also thank colleagues at SynBioNT (www.synbiont.
net) for helpful discussions, and in particular B. Davis and L. Cronin for their advice and
suggestions in the wider SynBioNT collaboration.
Author contributions
All authors conceived and designed the experiments, analysed the data and discussed the
results. N.K. and C.A. secured funding, X.X., C.S., A.B., L.L. and F.F-T performed the
experiments and L.L., F.F-T., N.K. and C.A. co-wrote the manuscript.
Additional information
Supplementary information is available in the online version of the paper. Reprints and
permissions information is available online at www.nature.com/reprints. Correspondence and
requests for materials should be addressed to F.F.T., N.K. and C.A.
Competing financial interests
The authors declare no competing financial interests.
NATURE CHEMISTRY DOI: 10.1038/NCHEM.1793 ARTICLES
NATURE CHEMISTRY | VOL 5 | DECEMBER 2013 | www.nature.com/naturechemistry 1065
© 2013 Macmillan Publishers Limited. All rights reserved.