Salmonella typhimurium intercepts Escherichia coli

signaling to enhance antibiotic tolerance

Nicole M. Vegaa,b,c, Kyle R. Allisona,b,c, Amanda N. Samuelsa,b,c, Mark S. Klempnerd, and James J. Collinsa,b,c,e,f,1
a

Howard Hughes Medical Institute, bDepartment of Biomedical Engineering, and cCenter of Synthetic Biology, Boston University, Boston, MA 02215;
MassBiologics, University of Massachusetts Medical School, Boston, MA 02126; eBoston University School of Medicine, Boston, MA 02118; and fWyss Institute
for Biologically Inspired Engineering, Harvard University, Boston, MA 02118
d

Edited* by Hans Leo Kornberg, Boston University, Boston, MA, and approved July 22, 2013 (received for review April 29, 2013)

Bacterial communication plays an important role in many populationbased phenotypes and interspecies interactions, including those in
host environments. These interspecies interactions may prove critical
to some infectious diseases, and it follows that communication
between pathogenic bacteria and commensal bacteria is a subject
of growing interest. Recent studies have shown that Escherichia
coli uses the signaling molecule indole to increase antibiotic tolerance throughout its population. Here, we show that the intestinal
pathogen Salmonella typhimurium increases its antibiotic tolerance in response to indole, even though S. typhimurium does not
natively produce indole. Increased antibiotic tolerance can be induced in S. typhimurium by both exogenous indole added to clonal
S. typhimurium populations and indole produced by E. coli in mixedmicrobial communities. Our data show that indole-induced tolerance
in S. typhimurium is mediated primarily by the oxidative stress response and, to a lesser extent, by the phage shock response, which
were previously shown to mediate indole-induced tolerance in
E. coli. Further, we nd that indole signaling by E. coli induces
S. typhimurium antibiotic tolerance in a Caenorhabditis elegans
model for gastrointestinal infection. These results suggest that the
intestinal pathogen S. typhimurium can intercept indole signaling
from the commensal bacterium E. coli to enhance its antibiotic
tolerance in the host intestine.

ather than acting autonomously, bacterial cells communicate

with one another to coordinate their efforts and relay vital
information. Interspecies and intraspecies bacterial communication has been implicated in many community-dependent behaviors including virulence (1), biolm formation (2), and antibiotic
tolerance (3). Communication may therefore allow control of heterogeneity, which is important in determining tness of microbial
populations (4). Recently, we reported that bacterial communication through indole signaling induces persister formation in
Escherichia coli (3). Persistence is an antibiotic-tolerant phenotype
in which a dormant subpopulation of cells (persisters) survives
antibiotic treatment without having genetically encoded resistance
factors (5, 6). In E. coli, we found that indole signaling induced
oxidative stress response and phage shock response pathways,
thereby increasing the persister frequency within the population.
This work suggested that bacteria can use intraspecies signaling
to modify the antibiotic tolerance of their population in response
to environmental conditions.
Indole signaling is used by bacteria in the distal intestine of
humans and other mammals (7). In this environment, alkaline and
low nutrient conditions induce expression of the indole-producing
tryptophanase (tnaA) enzyme in commensal E. coli and related
bacteria (8). Indole concentrations in the mammalian intestine
(300 M to 1 mM) (9, 10) can induce antibiotic tolerance in E. coli
without adversely affecting growth (11). As the mammalian intestine contains a richly mixed microbial population (12), signaling
molecules such as indole might be detected and used by both
commensal and pathogenic bacteria. Although there is increasing
interest in the roles that commensal bacteria play in mammalian
health (13), the mechanisms by which commensal bacteria interact
with invading pathogens are not yet well understood.

1442014425 | PNAS | August 27, 2013 | vol. 110 | no. 35

We hypothesized that pathogenic bacteria could use communication signals produced by commensal bacteria to sense and
adjust their physiological state to the host environment. As indole induces antibiotic tolerance in E. coli, we hypothesized that
it might also increase tolerance in related pathogens. Salmonella
typhimurium is one such pathogen which, although it does not
produce indole (14), has been shown to respond to signaling
molecules produced by other bacteria (15). S. typhimurium is
a common gastrointestinal pathogen and a major epidemiological threat, as it is a causative agent of gastroenteritis and sepsis.
This pathogen can survive macrophage engulfment and persist
within phagocytes, resulting in an asymptomatic but infectious
carrier state (16) where antibiotic tolerance is a signicant problem
(17). We therefore sought to determine if indole signaling by
E. coli might be exploited by S. typhimurium, leading to increased
tolerance of the pathogen in a host intestinal environment.
Here we show that indole signaling can indeed increase the
antibiotic tolerance of S. typhimurium. This tolerance can be
induced by exogenous indole in Salmonella-only cultures or by
indole produced by E. coli in a mixed-microbial population. Our
data suggest that this tolerance is mediated, in part, by oxidative
stress and phage shock response systems. Further, we nd, using
a Caernohabditis elegans infection model (18), that indole induces antibiotic tolerance of S. typhimurium in a mixed-microbial,
intestinal environment.
Results
We rst sought to test whether indole induced antibiotic tolerance in S. typhimurium (strain LT2). We treated exponentialphase S. typhimurium grown in tryptophan-free medium (M9CG
Signicance
Bacterial communication plays an important role in many
population-based phenotypes and interspecies interactions,
including those in host environments. Social interaction within
bacterial communities, and particularly communication between
pathogenic and commensal bacteria, is a subject of growing interest with relevance to ecology and human health. In this study,
we show a case of interspecies communication where the intestinal pathogen Salmonella typhimurium increases its antibiotic tolerance in response to the bacterial signaling molecule
indole, even though S. typhimurium does not natively produce
indole. These results suggest that this intestinal pathogen can
benet from indole signaling produced by E. coli and other
commensal bacteria.
Author contributions: N.M.V., K.R.A., A.N.S., M.S.K., and J.J.C. designed research; N.M.V.
and A.N.S. performed research; N.M.V. contributed new reagents/analytic tools; N.M.V.
analyzed data; N.M.V., K.R.A., and J.J.C. wrote the paper.
The authors declare no conict of interest.
*This Direct Submission article had a prearranged editor.
Freely available online through the PNAS open access option.
1

To whom correspondence should be addressed. E-mail: jcollins@bu.edu.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.

1073/pnas.1308085110/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1308085110

0.5
0.4

Carbenicillin

Ciprofloxacin

0.3
0.2
0.1
0.0

//

1.4

St % Survival

1.2
1.0

25

50

75 100 125
Indole (M)

250

//

500

Wild-Type Coculture
tnaA Coculture

0.8
0.6

**

0.4

**

0.2

0.0

St % Survival

0.8

0.5
1
Tryptophan (mM)

0 ng/mL aTc
50 mg/mL aTc

0.6
0.4

*
*

0.2
0.0

tnaA ptnaA

tnaA pGFP

Fig. 1. E. coli indole signaling induces antibiotic tolerance in Salmonella

typhimurium. (A) Exponential-phase cultures of S. typhimurium (St, n 3)
were incubated with indole for 1 h before treatment with carbenicillin
(100 g/mL) or ciprooxacin (0.5 g/mL). (B) Coculture of S. typhimurium with
indole-producing cultures of E. coli (n 6). T tests were performed vs. the
0 mM tryptophan condition. (C) Expression of plasmid-borne tnaA in E. coli
for induction of S. typhimurium tolerance in coculture (n 3). Cultures were
grown in M9CG + 2 mM tryptophan with or without anhydrotetracycline
(aTc) for induction of plasmid-borne genes. T tests were performed vs. the
uninduced condition (0 ng/mL aTc). Assays in B and C were performed using
ciprooxacin (0.5 g/mL). Error bars represent mean SD of biological replicates, and stars indicate signicance level of two-sided two-sample t tests
assuming unequal variance (*P 0.05; **P 0.01).

Vega et al.

in a mixed-microbial environment would also induce S. typhimurium

tolerance. To test this, we mixed an exponential-phase culture of
S. typhimurium with a stationary-phase culture of E. coli [E. coli
K-12 wild-type strain (EMG2) Pro, Table S1] grown in the presence of tryptophan to enable indole production. After a 1-h
incubation, the mixed culture was treated with ciprooxacin, as
uoroquinolone antibiotics are known to retain bactericidal activity in dense, starved cultures (20) such as those used in these
assays. Antibiotic-treated cultures were plated on selective media
to quantify the induced antibiotic tolerance in S. typhimurium
(Materials and Methods and SI Materials and Methods). As
expected, baseline tolerance of S. typhimurium was higher under
dense coculture conditions than in exponential-phase culture
(Fig. 1A) (21). We found that coculture with wild-type E. coli
increased the tolerance of S. typhimurium to ciprooxacin
when cultures were grown in the presence of tryptophan (Fig.
1B and Fig. S1). Interestingly, the relationship between tolerance
and tryptophan concentration was not strictly monotonic, with the
benet of tryptophan peaking at 1 mM. By measuring the E. coliproduced indole in cultures, we determined that 1 mM tryptophan corresponded to 125300 M of produced indole (Fig.
S2), consistent with tolerance levels observed when indole was
exogenously added (Fig. 1A). The conferred advantage for S.
typhimurium of the mixed-microbe environment required indole
production, as coculture with E. coli unable to produce indole
(tnaA strain) did not induce tolerance (Fig. 1B). Plasmid-based
expression of tnaA, but not a gfp control, restored the coculture
tolerance phenotype to the tnaA knockout strain (Fig. 1C), further
demonstrating the specic effect of indole on induced antibiotic
tolerance in S. typhimurium. These results indicate that E. coli indole signaling can induce antibiotic tolerance in S. typhimurium.
We previously determined that indole-induced tolerance in
E. coli was mediated by the oxidative stress response (oxyR) and
phage shock response (psp) (3). As these processes are largely
conserved in S. typhimurium, with some differences in effectors
and regulation (22, 23), we reasoned that indole-induced tolerance
might function through a similar mechanism in S. typhimurium. To
test this hypothesis, we performed qPCR on mRNA transcripts
from several genes in these pathways from S. typhimurium cultures
treated for 30 min with 50 M and 125 M indole, the concentrations that induced high carbenicillin and high ciprooxacin tolerance, respectively. Our qPCR results (Fig. 2A) indicated that,
as in E. coli, expression of the OxyR and phage shock regulons is
induced in S. typhimurium by indole. We found that 125 M indole
strongly induced the expression of genes in the OxyR regulon [dps
and hydroxyperoxidase I (katG)] and an effector of the phage
shock pathway (pspE), although 50 M indole produced weak
induction of OxyR regulon genes and did not have a detectable
effect on pspE (Fig. 2A). These results demonstrate that, as in
E. coli, indole can stimulate in S. typhimurium expression of genes
in the oxidative stress and phage shock pathways. Previous reports
have indicated that higher levels of indole induced expression of
genes involved in virulence and drug efux pumps in S. typhimurium (24). We did not observe induction of these genes (presented here as Other, Fig. 2A) when cultures were treated with
indole concentrations that induce increased antibiotic tolerance.
Interestingly, indole treatment produced a small but consistent
decrease in transcript abundance of a transcriptional regulator
associated with multiple antibiotic resistance (ramA); this gene
product has been shown to control expression of multidrug efux
pumps in S. typhimurium (24), suggesting that indole treatment
(at the levels considered here) may actually decrease efux in
this system.
We next used genetic knockouts to determine whether the
OxyR and phage shock pathways are involved in indole-induced
antibiotic tolerance in S. typhimurium. The oxyR and pspBC
mutants were constructed to allow inactivation of the OxyR and
phage shock responses, respectively (SI Materials and Methods). We
PNAS | August 27, 2013 | vol. 110 | no. 35 | 14421

MICROBIOLOGY

A
St % Survival

[M9 + 0.2% Casamino acids + 0.4% (wt/vol) glucose]; Materials

and Methods and SI Materials and Methods) with a range of
indole concentrations to determine the effect on tolerance to
carbenicillin and ciprooxacin, antibiotics that are used in clinical
treatment of enteric Salmonella infections (19) (Fig. 1A). We
found that exogenous indole increased tolerance to both antibiotics by over threefold, demonstrating that indole enhances
antibiotic tolerance despite the inability of S. typhimurium to
produce this signal (14). Interestingly, the protective range of
indole was different for the two antibiotics. Carbenicillin tolerance
peaked in the presence of 50 M indole and declined at higher
indole concentrations, whereas ciprooxacin tolerance was enhanced by higher concentrations of indole, peaking at 125 M
indole. The reason for this difference is unclear, but it may suggest
that the protective processes induced by indole play differing
roles against different antibiotics.
As addition of exogenous indole increased tolerance of S.
typhimurium cultures, we postulated that indole produced by E. coli

125 M indole

4
3
2

**

**
*

0
-1
-2

ahpC

dps

katG

OxyR

B 0.12
0.10
% Survival

50 M indole

**

trxC

pspA

pspE invF

Phage
Shock

0 M indole
125 M indole

napF ramA
Other

**

0.08
0.06
0.04
0.02
0.00

Effect
Size (d)

oxyR

pspBC

2.89

0.96

narV

Fig. 2. Response of OxyR and phage shock pathways to indole in S. typhimurium. (A) Transcriptional response of S. typhimurium to indole treatment
as determined by qPCR. S. typhimurium cultures were treated with 0, 50
(pink bars) or 125 (red bars) M indole (SI Materials and Methods). Results
are shown as fold-change expression (Ct) of indole-treated vs. untreated
cultures. Error bars represent mean SD of three to four biological replicates. (B) Exponential-phase cultures of S. typhimurium narV:CmR (control),
oxyR:CmR, and pspBC:CmR were incubated with indole (0 or 125 M) for
1 h before treatment with ciprooxacin (0.5 g/mL). Error bars represent
mean SD of 813 biological replicates. Stars indicate signicance level of
two-sided two-sample t tests assuming unequal variance (*P 0.05; **P
0.01). Cohens effect size (d) using pooled SD was calculated for the difference
between indole-treated and untreated cultures compared with the difference
observed in the narV control; d > 0.8 suggests a large effect.

found that indole-induced tolerance was reduced in the pspBC

mutant and entirely eliminated in the oxyR mutant relative to
the control (Fig. 2B). These results suggest that both the OxyR
and phage shock responses are involved in indole-induced antibiotic tolerance in S. typhimurium, as was previously observed in
E. coli. Interestingly, although the mechanism of indole-induced
antibiotic tolerance appears to be conserved between species, the
role of the phage shock response is attenuated in S. typhimurium
compared with E. coli.
Given the high levels of induction in some aspects of the oxidative stress response (Fig. 2A), we hypothesized that the OxyR
regulon might play a role in inducing antibiotic tolerance in
S. typhimurium. To test this hypothesis, we treated S. typhimurium
cultures with a range of H2O2 concentrations 1 h before treating
cultures with antibiotic (Fig. 3). We found that low levels of oxidative stress (Fig. S3) could increase antibiotic tolerance of S.
typhimurium to the same degree induced by indole, suggesting that
indole may primarily act through the OxyR regulon to induce
tolerance in S. typhimurium.
The protective range of H2O2 was different for carbenicillin
and ciprooxacin, with better protection against carbenicillin at
lower H2O2 concentrations and better protection against ciprooxacin at higher H2O2 concentrations. Interestingly, this antibiotic-specic concentration dependence for H2O2 treatment (Fig.
3) resembles the concentration dependence observed during indole
treatment (Fig. 1A). The qualitative similarities in the antibiotic
tolerance curves suggest that the OxyR regulon plays a functional
role in indole-induced antibiotic tolerance in S. typhimurium. These
14422 | www.pnas.org/cgi/doi/10.1073/pnas.1308085110

data may additionally suggest that contribution of reactive oxygen

species (ROS) to antibiotic lethality differs between carbenicillin
and ciprooxacin, adding further nuance to a growing understanding of antibiotic-induced cell death (2528).
Having demonstrated that S. typhimurium responds to E. coliproduced indole by inducing an oxidative stress response, thereby
increasing antibiotic tolerance, we sought to determine whether
this interspecies signaling occurred in an intestinal environment.
We used a C. elegans model for S. typhimurium infection (18) to
explore in vivo the relation between E. coli-produced indole and
S. typhimurium tolerance. This model provides a tractable approach for investigating intestinal mixed-microbial interactions
and has been used for determining in vivo antibiotic efcacy
against pathogenic bacteria (29). Although clearly less complex
than the mammalian intestine, the S. typhimurium intestine maintains many critical factors that pertain to microbes, including
mixed, semiimmobilized bacterial populations with spatial distributions, innate antimicrobial defenses, and the requirement
of pathogen-epithelium adhesion (30).
We used uorescent microscopy to verify that E. coli and
S. typhimurium established a mixed-microbial community in the
C. elegans intestine (18). Synchronized cultures of adult C. elegans
were fed on E. coli (mCherry) in tryptophan-free S-medium with
or without 125 M indole then incubated with S. typhimurium
(GFP) to allow infection; indole was added exogenously to allow
precise control of indole concentrations (Fig. 4 A and B; SI
Materials and Methods). Worms generally showed observable uorescence in both channels, indicating the simultaneous presence of
E. coli and S. typhimurium in the intestine. Interestingly, indoletreated cultures (125 M) showed low-intensity S. typhimurium
infection with small adhered colonies in nearly all worms sampled
(Fig. 4B), whereas untreated cultures contained a mix of heavily
infected and uninfected worms (Fig. 4A). However, by measuring
cfu/worm, we found that treatment with 125 M indole did not
affect the average pathogen count per worm (Fig. 4C), suggesting
that indole altered heterogeneity of the S. typhimurium infection
in C. elegans.
We next sought to determine if indole affected S. typhimurium
antibiotic tolerance in the C. elegans model. Synchronized adult
cultures of C. elegans in S-medium 125 M indole were fed on
E. coli (EMG2 tnaA Pro; SI Materials and Methods). Infection
with S. typhimurium [chloramphenicol resistance (CmR); SI
Materials and Methods] was then facilitated as described above,

1.0

Ciprofloxacin

Carbenicillin

120

Growth

100

0.8

80
0.6
60
0.4
40
0.2

%Growth

St % Survival

Fold-Change Expression

20

0.0
0 15 30 60

90 120 150

//

300

//

0
600

H2O2 (M)
Fig. 3. Hydrogen peroxide induces tolerance in S. typhimurium. Exponential-phase cultures of S. typhimurium LT2 in M9CG were incubated with H2O2
(0600 M) for 1 h before treatment with antibiotics. Black line indicates
percent growth of LT2 cultures during incubation with hydrogen peroxide,
where growth in the absence of hydrogen peroxide was used as reference
(100%). Percent survival was determined after 4-h treatment with ciprooxacin (0.5 g/mL, dashed red line) or carbenicillin (100 g/mL, dotted blue
line). Error bars represent mean SD of at least six biological replicates.

Vega et al.

0 M Indole

125 M Indole

4.0

St %Survival

St log(cfu/Worm)

3.0
2.0
1.0

20

*
15
10
5

0.0
0 M

125 M
Indole

coculture, enhance the antibiotic tolerance of S. typhimurium. Indole-induced tolerance was observed consistently in S. typhimurium
despite naturally occurring variation in baseline tolerance measurements (31). Physiological responses to indole have been observed in non-indole-producing bacteria (32), suggesting that indole
can function as an interspecies signal (2, 3335). The data presented
here implicate indole signaling by commensal bacteria and exploitation of this signal by pathogenic bacteria as a potential factor in
establishing antibiotic tolerance of pathogens.
We also showed that indole signaling by E. coli enhances
S. typhimurium tolerance in a C. elegans host intestinal model.
Despite the relative simplicity of the C. elegans model, it has
been suggested as a useful model for S. typhimurium infection of
mammals based on the observation that strains attenuated in
virulence in mammals were also attenuated in C. elegans (29).

0 M

125 M
Indole

Fig. 4. Indole increases antibiotic tolerance of S. typhimurium in the C. elegans

intestine. All experiments were performed in S-Medium (SI Materials and
Methods) at 25 C to ensure sterility and survival of C. elegans AU37. Images
were selected to represent the variation observable in each experimental condition. (A and B) Visualization of infection in the C. elegans intestine. C. elegans
fed on E. coli EMG2 (mCherry) and infected with S. typhimurium (GFP) with (A)
no indole or (B) 125 M indole added to media were imaged 36 h after initial
infection. Differential interference contrast (DIC) images are shown overlaid with
red and green uorescent channels, and uorescent channels are shown in
isolation (SI Materials and Methods). (C) Average intensity of infection by
S. typhimurium (St). (D) Survival of S. typhimurium in the C. elegans intestine
after treatment of worm cultures with ciprooxacin (2 g/mL). All error
bars indicate mean SD of four biological replicates. Stars indicate signicance level of one-sided two-sample t tests assuming unequal variance,
comparing indole-treated and untreated cultures (*P 0.05; **P 0.01).

Wild-type, 0 mM Trp

Wild-type, 0.5 mM Trp

tnaA, 0 mM Trp

Discussion
Here, we report that physiologically relevant concentrations of
indole, whether added exogenously or produced by E. coli in
Vega et al.

tnaA, 0.5 mM Trp

0 mM Trp

0.5 mM Trp

4
3
2
1
0

Wild-type

tnaA

30
25

0 mM Trp
0.5 mM Trp

20
MICROBIOLOGY

St % Survival

E
St log(cfu/worm)

and infected cultures were treated with ciprooxacin to determine antibiotic tolerance. We observed that, although ciprooxacin was effective in killing S. typhimurium in the C. elegans
intestine, worms incubated with indole carried S. typhimurium with
higher antibiotic tolerance. This nding suggests that indole increased antibiotic tolerance of S. typhimurium in the C. elegans
host intestine (Fig. 4D and Fig. S4).
Having observed that indole could increase S. typhimurium
tolerance in a host-commensal-pathogen model, we sought to
test if indole produced by E. coli in this model was sufcient for
increasing tolerance. We investigated this by incubating synchronized cultures of adult C. elegans in modied S-medium
(pH7) with or without additional tryptophan using E. coli wildtype or tnaA strains as the C. elegans food source (SI Materials
and Methods). We found that addition of tryptophan to worm
media encouraged the formation of small punctate intestinal
colonies of S. typhimurium when worms were fed on wild-type E.
coli but not when fed on the tnaA E. coli strain (Fig. 5 AD).
Average pathogen count per worm was unchanged across all
experimental conditions (Fig. 5E). We found that adding tryptophan to C. elegans culture media increased antibiotic tolerance
of S. typhimurium when wild-type E. coli was present, but not in
the presence of the tnaA E. coli strain (Fig. 5F). These results
suggest that S. typhimurium can intercept E. coli indole signaling
in the C. elegans intestine, allowing the pathogen to increase its
tolerance to antibiotics.

15
10
5
0

Wild-type

tnaA

Fig. 5. Indole produced by E. coli in a mixed-microbe population induces

tolerance of S. typhimurium in the C. elegans intestine. All experiments were
performed in modied S-Medium (SI Materials and Methods). Images were
selected to represent the variation observable in each experimental condition. (AD) Visualization of infection in the C. elegans intestine. C. elegans
fed on (A and B) E. coli EMG2 (mCherry) or (C and D) tnaA (mCherry) and
infected with S. typhimurium (GFP) with (A and C) no tryptophan or (B and
D) 0.5 mM tryptophan added to media were imaged 36 h after initial
infection. DIC images are shown overlaid with red and green uorescent
channels, and uorescent channels are shown in isolation. Images were
autoscaled using software default settings (SI Materials and Methods).
(E ) Average intensity of infection by S typhimurium (St). (F ) Survival of
S. typhimurium in the C. elegans intestine after treatment of worm cultures
with ciprooxacin (2 g/mL). Error bars indicate mean SD of four biological
replicates. Stars indicate signicance level of one-sided two-sample t tests
assuming unequal variance, comparing tryptophan-treated and untreated
cultures (*P 0.05; **P 0.01).

PNAS | August 27, 2013 | vol. 110 | no. 35 | 14423

The C. elegans infection model was used here to create a spatially

nonhomogenous mixed-bacterial culture with adhesion of bacteria
to an epithelium. Our results with this system indicate that pathogenic S. typhimurium are able to eavesdrop on commensal
bacterial communication by E. coli to enhance their antibiotic
tolerance within the host intestine (Fig. 6).
We found that indole-induced tolerance in S. typhimurium is
mediated, in part, by the oxidative stress response and the phage
shock response, as was seen previously in E. coli (3). Antioxidant
capability has previously been linked to antibiotic tolerance in
bacteria (36, 37), and in this case, induction of the oxidative
stress response appears to be largely responsible for the indoleinduced antibiotic-tolerant phenotype. We observed increased
transcript levels for katG, which acts to protect the cell from
damage under conditions of increased oxidative stress (38, 39),
but not alkyl hydroperoxide reductase C (ahpC), which has been
implicated in scavenging of low-level, endogenously produced
H2O2 (40), suggesting that indole response prepares these bacteria
for survival in stressful conditions.
We found some differences in indole-induced changes in
transcript abundance between E. coli and S. typhimurium. For
example, while both species show an indole-responsive increase
in pspE transcript, indole-treated E. coli show increased levels
of pspA transcript, whereas indole-treated S. typhimurium do not
(3). Although stress response genes are largely conserved and
highly homologous between E. coli and S. typhimurium, there are
important functional and regulatory differences between these
species (22, 23, 41), and it is therefore expected that indole-based
induction of individual genes within the relevant stress response
pathways will vary somewhat between species. However, it is notable that the same pathways are observed to respond to indole in
both species, suggesting that the mechanism of indole-induced
tolerance is conserved. It remains to be tested whether this crossspecies protection extends to other bacterial pathogens, including
Gram-positives.
Indole signaling may affect stress responses that are important
for survival in a host and establishment of a chronic infection.
Previous work has shown that survival in macrophages is critical
for Salmonella virulence (42, 43) and that peroxide catalase activity plays an essential role in survival (39). OxyR-mediated
processes may protect Salmonella intracellularly by inducing
tolerance to the oxidative bursts that immune cells use to kill
bacteria. The importance of oxidative stress in establishment and
recurrence of Salmonella infection is demonstrated by chronic
granulomatous disease, a hereditary immunodeciency in which
macrophage cannot generate the oxidative burst; patients with
this disease are highly susceptible to bacteremic infection by
nontyphoidal Salmonella and to recurrences of these infections
(44). We found that indole signaling strongly induced expression
of OxyR regulon genes in S. typhimurium, suggesting that the ability
to detect commensal indole signaling may provide the pathogen
with a method of altering its physiology to tolerate the stresses

Host
Indole
Susceptible

E. coli

S. typhimurium

Tolerant

Fig. 6. Bacterial communication, signal interception, and antibiotic tolerance in the host environment. Within the C. elegans host intestine, pathogenic S. typhimurium encounters indole-producing E. coli (7) and is able to
intercept indole signaling to enhance its antibiotic tolerance.

14424 | www.pnas.org/cgi/doi/10.1073/pnas.1308085110

of the immune system and antibiotics and thereby establish a

persistent infection.
More broadly, our results suggest that persistence of pathogens in the host environment may be induced by the interception
of nonnatively produced bacterial-signaling molecules. There is
a growing body of work exploring the complex relationship between the host, the innate microbiota, and gastrointestinal pathogens (4549), and it is increasingly evident that the interplay
between the established microbiota and introduced organisms
in the intestine may be critical in determining the progress and
resolution of infections and the aftermath of disease. Here, we
observe a case where a bacterial pathogen has lost the capacity
for production of a signal but has retained the ability to respond
to signaling produced by commensals in the shared intestinal
environment. Further, we show that interception of and response
to nonnative signaling produces an antibiotic-tolerant phenotype
in a bacterial pathogen, suggesting that interactions between
the commensal microbiota and invading pathogens may in some
cases improve stress tolerance in pathogens, thereby increasing
recalcitrance of bacterial infections.
Materials and Methods
Bacterial Strains and Strain Construction. All experiments were performed using
laboratory strains of E. coli and Salmonella enterica serovar Typhimurium.
Ancestral wild-type E. coli K-12 EMG2 +fertility factor plasmid (F+) obtained
from Yale E. coli Genetic Stock Center (ECGC 4401) was the reference wildtype E. coli strain used in all experiments, and S. typhimurium LT2 [American
Type Culture Collection (ATCC) 700720] was the reference strain of nontyphoidal Salmonella. C. elegans strains were provided by the C. elegans
Genetic Stock Center, which is funded by National Institutes of Health Ofce
of Research Infrastructure Programs (P40 OD010440). Bacterial strains and
primers used in this study are presented in Table S1. Details of strain construction are presented in SI Materials and Methods.
Antibiotics and Chemicals. The following concentrations of antibiotics were
used in this study: 100 g/mL carbenicillin, 1060 g/mL kanamycin, 0.52 g/mL
ciprooxacin, and 15 g/mL ooxacin. Strains containing kanamycinresistance plasmids were grown with 3060 g/mL kanamycin for selection,
and strains containing spectinomycin-resistance cassettes were grown with
50 g/mL spectinomycin. For induction of plasmid-borne genes, 2550 ng/mL
anhydrotetracycline (aTc) was added after 24 h of growth. Otherwise,
antibiotic treatments were 10 minimum inhibitory concentration (MIC) to
ensure killing of sensitive cells.
Growth and Tolerance Assays. All experiments were performed in accordance
with standard protocols unless otherwise stated. Briey, bacterial cultures
were grown in light-insulated shakers at 37 C with shaking at 300 rpm
(14 mL-Falcon tubes, Fisher Scientic) or 900 rpm (96-well, clear, at-bottom
culture plates, Fisher Scientic; with Breathe-Easy adhesive gas-permeable
membrane, USA Scientic). Cultures were grown in tryptophan-free rich
media (M9 + 0.2% casamino acids + 0.4% glucose, M9CG, pH >7.2) or in
complete rich media (Luria-Bertani, LB).
Antibiotic tolerance was assessed by incubating cultures for at least 4 h
with antibiotic to allow full killing of sensitive cells. Serial dilution and plating
were used to determine cfu/mL before and after treatment (SI Materials
and Methods).
For coculture experiments, E. coli K-12 EMG2 was grown to stationary
phase in M9CG + 02 mM tryptophan to allow indole production. The tnaA
knockout was used to prevent indole production. Cultures of E. coli tnaA
pZA21-tnaA or pZA21-GFP were grown in M9CG + 2 mM tryptophan, with or
without 50 ng/mL anhydrotetracycline (aTc) for induction of plasmid-borne
genes. Twenty percent of culture volume was replaced with exponentialphase culture of S. typhimurium in M9CG, and cultures were incubated 1 h
before treatment with ciprooxacin.
Indole Quantication. Indole quantication was performed via HPLC. Details
of sample preparation and chromatography are presented in SI Materials
and Methods.
Quantitative PCR. RNA was collected from indole-treated and untreated
cultures of S. typhimurium LT2. Cultures were inoculated 1:500 from overnight LB cultures into 1 mL of M9CG in 14-mL Falcon tubes and incubated

Vega et al.

Induction of OxyR. Cultures were inoculated 1:200 (E. coli ) or 1:500

(S. typhimurium) from overnight LB cultures into M9CG (1 mL in 14-mL Falcon
tubes or 150 L in 96-well plates) and allowed to grow to exponential (3.5 h)
or stationary phase (24 h) at 37 C. Cultures were incubated 1 h with hydrogen
peroxide (0600 M) before treatment with antibiotics. Serial dilution plating
was performed before and after 1 h of hydrogen peroxide incubation and
after 4 h of ooxacin treatment to determine survival at each stage.

(sek) mutant strain AU37 [glp-4 (bn2) I; sek-1 (km4) X] of C. elegans (Caenorhabditis Genetics Center) was used as a model organism for Salmonella
pathogenesis. E. coli OP50 was used as a food source in maintenance cultures.
E. coli EMG2 and tnaA Pro were used as food sources during experiments, and
S. typhimurium nitrate reductase 2 mutant (narV:CmR) was used as a pathogen.
For uorescent microscopy, worms were fed on E. coli EMG2 pZS4-mCherry or
E. coli tnaA pZS4-mCherry, and S. typhimurium pZA21-GFP was used as the
infectious agent. Details of C. elegans culture and experimental conditions are
given in SI Materials and Methods.

C. elegans Intestinal Model for S. typhimurium Infection. The temperaturesensitive germ line proliferation (glp) mitogen-activated protein kinase kinase

ACKNOWLEDGMENTS. We thank Ari E. Friedland for C. elegans experimental advice and Daniel J. Dwyer and D. Ewen Cameron for helpful suggestions
on the manuscript. This work was supported by the US National Science
Foundation, the National Institutes of Health Directors Pioneer Award Program, and the Howard Hughes Medical Institute.

1. Rutherford ST, Bassler BL (2012) Bacterial quorum sensing: Its role in virulence and
possibilities for its control. Cold Spring Harb Perspect Med 2(11):a012427.
2. Martino PD, Fursy R, Bret L, Sundararaju B, Phillips RS (2003) Indole can act as an
extracellular signal to regulate biolm formation of Escherichia coli and other indoleproducing bacteria. Can J Microbiol 49(7):443449.
3. Vega NM, Allison KR, Khalil AS, Collins JJ (2012) Signaling-mediated bacterial persister
formation. Nat Chem Biol 8(5):431433.
4. Nevozhay D, Adams RM, Van Itallie E, Bennett MR, Balzsi G (2012) Mapping the
environmental tness landscape of a synthetic gene circuit. PLOS Comput Biol 8(4):
e1002480.
5. Rotem E, et al. (2010) Regulation of phenotypic variability by a threshold-based mechanism underlies bacterial persistence. Proc Natl Acad Sci USA 107(28):1254112546.
6. Allison KR, Brynildsen MP, Collins JJ (2011) Metabolite-enabled eradication of bacterial persisters by aminoglycosides. Nature 473(7346):216220.
7. Botsford JL, Demoss RD (1972) Escherichia coli tryptophanase in the enteric environment. J Bacteriol 109(1):7480.
8. Han TH, Lee JH, Cho MH, Wood TK, Lee J (2011) Environmental factors affecting indole production in Escherichia coli. Res Microbiol 162(2):108116.
9. Karlin DA, Mastromarino AJ, Jones RD, Stroehlein JR, Lorentz O (1985) Fecal skatole
and indole and breath methane and hydrogen in patients with large bowel polyps or
cancer. J Cancer Res Clin Oncol 109(2):135141.
10. Zuccato E, et al. (1993) Role of bile acids and metabolic activity of colonic bacteria in
increased risk of colon cancer after cholecystectomy. Dig Dis Sci 38(3):514519.
11. Li G, Young KD (2013) Indole production by the tryptophanase TnaA in Escherichia coli is
determined by the amount of exogenous tryptophan. Microbiology 159(Pt 2):402410.
12. Hughes DT, et al. (2010) Chemical sensing in mammalian host-bacterial commensal
associations. Proc Natl Acad Sci USA 107(21):98319836.
13. Blaser M, Bork P, Fraser C, Knight R, Wang J (2013) The microbiome explored: Recent
insights and future challenges. Nat Rev Microbiol 11(3):213217.
14. Bergey DH, Breed RS (1957) Bergeys Manual of Determinative Bacteriology (Williams
& Wilkins, American Society for Microbiology, Baltimore), 7th Ed.
15. Sperandio V (2010) SdiA bridges chemical signaling between Salmonella enterica
serovar Typhimurium and Yersinia enterocolitica in mice. J Bacteriol 192(1):2122.
16. Ruby T, McLaughlin L, Gopinath S, Monack D (2012) Salmonellas long-term relationship with its host. FEMS Microbiol Rev 36(3):600615.
17. Sirinavin S, et al. (2003) Noroxacin and azithromycin for treatment of nontyphoidal
salmonella carriers. Clin Infect Dis 37(5):685691.
18. Aballay A, Yorgey P, Ausubel FM (2000) Salmonella typhimurium proliferates and
establishes a persistent infection in the intestine of Caenorhabditis elegans. Curr Biol
10(23):15391542.
19. Chiu C-H, Lin T-Y, Ou JT (1999) In vitro evaluation of intracellular activity of antibiotics
against non-typhoid Salmonella. Int J Antimicrob Agents 12(1):4752.
20. Eng RH, Padberg FT, Smith SM, Tan EN, Cherubin CE (1991) Bactericidal effects of
antibiotics on slowly growing and nongrowing bacteria. Antimicrob Agents Chemother 35(9):18241828.
21. Levin BR, Rozen DE (2006) Non-inherited antibiotic resistance. Nat Rev Microbiol 4(7):
556562.
22. Farr SB, Kogoma T (1991) Oxidative stress responses in Escherichia coli and Salmonella
typhimurium. Microbiol Rev 55(4):561585.
23. Joly N, et al. (2010) Managing membrane stress: The phage shock protein (Psp) response,
from molecular mechanisms to physiology. FEMS Microbiol Rev 34(5):797827.
24. Nikaido E, Shirosaka I, Yamaguchi A, Nishino K (2011) Regulation of the AcrAB
multidrug efux pump in Salmonella enterica serovar Typhimurium in response to
indole and paraquat. Microbiology 157(Pt 3):648655.
25. Dwyer DJ, Kohanski MA, Collins JJ (2009) Role of reactive oxygen species in antibiotic
action and resistance. Curr Opin Microbiol 12(5):482489.
26. Dwyer DJ, Kohanski MA, Hayete B, Collins JJ (2007) Gyrase inhibitors induce an oxidative damage cellular death pathway in Escherichia coli. Mol Syst Biol 3:91.

27. Kohanski MA, Dwyer DJ, Hayete B, Lawrence CA, Collins JJ (2007) A common mechanism of cellular death induced by bactericidal antibiotics. Cell 130(5):797810.
28. Brynildsen MP, Winkler JA, Spina CS, MacDonald IC, Collins JJ (2013) Potentiating
antibacterial activity by predictably enhancing endogenous microbial ROS production. Nat Biotechnol 31(2):160165.
29. Labrousse A, Chauvet S, Couillault C, Kurz CL, Ewbank JJ (2000) Caenorhabditis elegans is a model host for Salmonella typhimurium. Curr Biol 10(23):15431545.
30. Alegado RA, Tan M-W (2008) Resistance to antimicrobial peptides contributes to
persistence of Salmonella typhimurium in the C. elegans intestine. Cell Microbiol
10(6):12591273.
31. Jers A, Kaldalu N, Tenson T (2010) The frequency of persisters in Escherichia coli
reects the kinetics of awakening from dormancy. J Bacteriol 192(13):33793384.
32. Lee JH, Lee J (2010) Indole as an intercellular signal in microbial communities. FEMS
Microbiol Rev 34(4):426444.
33. Hamilton S, et al. (2009) The transcriptional programme of Salmonella enterica serovar Typhimurium reveals a key role for tryptophan metabolism in biolms. BMC
Genomics 10:599.
34. Lee J, Attila C, Cirillo SLG, Cirillo JD, Wood TK (2009) Indole and 7-hydroxyindole
diminish Pseudomonas aeruginosa virulence. Microb Biotechnol 2(1):7590.
35. Chu W, et al. (2012) Indole production promotes Escherichia coli mixed-culture
growth with Pseudomonas aeruginosa by inhibiting quorum signaling. Appl Environ
Microbiol 78(2):411419.
36. Grant SS, Kaufmann BB, Chand NS, Haseley N, Hung DT (2012) Eradication of bacterial
persisters with antibiotic-generated hydroxyl radicals. Proc Natl Acad Sci USA 109(30):
1214712152.
37. Wakamoto Y, et al. (2013) Dynamic persistence of antibiotic-stressed mycobacteria.
Science 339(6115):9195.
38. Gonzlez-Flecha B, Demple B (1997) Homeostatic regulation of intracellular hydrogen
peroxide concentration in aerobically growing Escherichia coli. J Bacteriol 179(2):
382388.
39. Hbrard M, Viala JPM, Mresse S, Barras F, Aussel L (2009) Redundant hydrogen
peroxide scavengers contribute to Salmonella virulence and oxidative stress resistance. J Bacteriol 191(14):46054614.
40. Seaver LC, Imlay JA (2001) Alkyl hydroperoxide reductase is the primary scavenger of
endogenous hydrogen peroxide in Escherichia coli. J Bacteriol 183(24):71737181.
41. Storz G, et al. (1989) An alkyl hydroperoxide reductase induced by oxidative stress in
Salmonella typhimurium and Escherichia coli: Genetic characterization and cloning of
ahp. J Bacteriol 171(4):20492055.
42. Monack DM, Bouley DM, Falkow S (2004) Salmonella typhimurium persists within
macrophages in the mesenteric lymph nodes of chronically infected Nramp1+/+ mice
and can be reactivated by IFNgamma neutralization. J Exp Med 199(2):231241.
43. Fields PI, Swanson RV, Haidaris CG, Heffron F (1986) Mutants of Salmonella typhimurium that cannot survive within the macrophage are avirulent. Proc Natl Acad Sci
USA 83(14):51895193.
44. Gordon MA (2008) Salmonella infections in immunocompromised adults. J Infect
56(6):413422.
45. Thiennimitr P, Winter SE, Bumler AJ (2012) Salmonella, the host and its microbiota.
Curr Opin Microbiol 15(1):108114.
46. Ahmer BM, Gunn JS (2011) Interaction of Salmonella spp. with the intestinal microbiota. Front Microbiol 2:101.
47. Stecher B, et al. (2010) Like will to like: Abundances of closely related species can
predict susceptibility to intestinal colonization by pathogenic and commensal bacteria. PLoS Pathog 6(1):e1000711.
48. Parker CT, Sperandio V (2009) Cell-to-cell signalling during pathogenesis. Cell Microbiol 11(3):363369.
49. Srikanth CV, McCormick BA (2008) Interactions of the intestinal epithelium with the
pathogen and the indigenous microbiota: A three-way crosstalk. Interdiscip Perspect
Infect Dis 2008:626827.

Vega et al.

PNAS | August 27, 2013 | vol. 110 | no. 35 | 14425

MICROBIOLOGY

3.5 h before treatment with indole. After 30-min incubation with indole (0,
50, or 125 M), cultures were stabilized with RNAprotect Bacteria Reagent
(Qiagen) according to the manufacturers protocol. Details of RNA extraction, cDNA synthesis, and qPCR are presented in SI Materials and Methods.

Supporting Information
Vega et al. 10.1073/pnas.1308085110
SI Materials and Methods
Bacterial Strains and Strain Construction. All experiments were

performed using laboratory strains of Escherichia coli and Salmonella enterica serovar Typhimurium. Ancestral wild-type E. coli
K-12 EMG2 +fertility factor plasmid (F+) obtained from Yale
E. coli Genetic Stock Center (ECGC 4401) was the reference
wild-type E. coli strain used in all experiments, and S. typhimurium LT2 [American Type Culture Collection (ATCC) 700720]
was the reference strain of nontyphoidal Salmonella (14). Single-gene knockout mutants were constructed from an E. coli
KanR knockout library (5) via lambda Red transduction (6) into
wild type. All Salmonella knockouts and any E. coli mutants not
available in the knockout library were constructed using the
Datsenko-Wanner PCR products method (6). PCR primers for
the Datsenko-Wanner method were designed using Oligocalc (7)
and obtained from Integrated DNA Technologies. Strains and
primers used in this study are presented in Table S1.
Rescue plasmids were constructed using the pZ system (8).
Purication of restriction digest products and PCR products was
performed using commercially available kits (Qiagen). Cloning
primers contained an 8-bp GC-rich leader sequence, a restriction
site for plasmid insertion, and a 26-bp homology sequence
overlying the START or STOP codon of the gene of interest
(Table S1). Genes for complementation were cloned from E. coli
EMG2 using hot-start Phusion PCR (Sigma-Aldrich). PCR products and the parent plasmid pZA21-GFP were digested using KpnI/
BamIII and ligated using T4 DNA ligase (Sigma-Aldrich), replacing
the GFP cassette with the gene of interest.
E. coli strains were transfected with the spectinomycinresistant Pro cassette to enable activation from the PLtet0-1 promoter. Strains were then made electrocompetent by washing
twice in ice-cold sterile water and twice in ice-cold sterile 10%
glycerol. Electrocompetent strains were incubated 20 min with
plasmid in cold 10% (vol/vol) glycerol before electroporation at
2,400 V. Plasmid-transformed cells were inoculated in super
optimal broth with catabolite repression (SOC) media and incubated 1 h at 37 C to allow expression of plasmid-based resistance genes before addition of the appropriate antibiotic for
selection.
Antibiotics and Chemicals. The following concentrations of antibiotics were used in this study: 100 g/mL ampicillin or carbenicillin, 1060 g/mL kanamycin, 0.52 g/mL ciprooxacin, and
15 g/mL ooxacin. Strains containing kanamycin-resistance
plasmids were grown with 3060 g/mL kanamycin for selection,
and strains containing spectinomycin-resistance cassettes were
grown with 50 g/mL spectinomycin. Otherwise, antibiotic treatments were 10 minimum inhibitory concentration (MIC) to
ensure killing of sensitive cells (9). All antibiotics were purchased
from Sigma-Aldrich. L-tryptophan, indole [99%, Food Chemicals
Codex (FCC)], and hydrogen peroxide [American Chemical
Society (ACS) reagent, stock strength 30%] were purchased from
Sigma-Aldrich.
In all experiments, cultures were grown in media containing
antibiotics for plasmid selection (35 g/mL kanamycin), and
2550 ng/mL anhydrotetracycline (aTc) was added after 24 h
growth for induction of plasmid-borne genes.
Growth and Tolerance Assays. All experiments were performed
in accordance with standard protocols unless otherwise stated.
Briey, bacterial cultures were grown in light-insulated shakers
at 37 C with shaking at 300 rpm (14-mL Falcon tubes, Fisher
Vega et al. www.pnas.org/cgi/content/short/1308085110

Scientic) or 900 rpm (96-well, clear, at-bottom culture plates,

Fisher Scientic; with Breathe-Easy adhesive gas-permeable membrane, USA Scientic). Cultures were grown in tryptophan-free
minimal medium [M9 + 0.2% Casamino acids + 0.4% (wt/vol)
glucose, M9CG, pH >7.2] or in rich media (LB).
MIC was determined via growth assay in 96-well plates, using
log2 dilution series of antibiotics (10).
Antibiotic tolerance was assessed by incubating cultures for
at least 3 h with antibiotic to allow full killing of sensitive cells
(11, 12). For stationary-phase cultures, ciprooxacin treatment
was used for killing of sensitive cells, as quinolones have bactericidal activity against stationary-phase bacteria (11, 13, 14). Ten
microliters of culture medium was removed before addition of
antibiotic and after antibiotic challenge for serial dilution in 1
PBS, and serial dilutions were spot-plated onto LB agar (10-L
spots). Serial dilution plates were allowed to grow overnight at
37 C, and colony counts were used to calculate cfu/mL. The cfu
measurements before and after antibiotic treatment were compared to determine survival.
The S. typhimurium oxidative stress response (oxyR) mutant,
despite lacking a homolog to the E. coli occulation gene (u)
(15), strongly self-aggregates, making dilution plating difcult. In
tolerance assays where this strain was used (Fig. 2), all samples
were collected by spinning 2 min at 7,600 g in a tabletop
centrifuge, washed twice in 10 volumes sterile 1 PBS, and resuspended in one volume sterile 1 phosphosaline (pH 4) + 0.5%
Tween-20 by pipetting and vortexing. All strains used in these
assays were treated identically to ensure consistency. Serial dilution
was performed in resuspension buffer, and dilutions were immediately plated onto LB agar. As we were unable to clear the
antibiotic-resistance cassette from oxyR and phage shock response (pspBC) strains using pCP20 (6), tolerance assays using
these knockouts were performed with S. typhimurium LT2 nitrate
reductase 2 mutant bearing a chloramphenicol resistance cassette
(narV:CmR) (described in the next section) as the reference
strain.
Coculture of E. coli and S. typhimurium. E. coli Pro was inoculated
1:200 from overnight LB cultures into M9CG containing 02 mM
tryptophan, then dispensed in 150-L aliquots into a at-bottomed
clear 96-well culture plate, covered with Breathe-Easy gaspermeable membrane, and incubated at 37 C as previously
described for 16 h (stationary phase). S. typhimurium was inoculated 1:500 from overnight LB cultures into M9CG, aliquoted into a 96-well plate, and grown under identical conditions
for 3.5 h (exponential phase) before coculture. Twenty percent
of stationary-phase E. coli cultures was removed and replaced
with an equal volume of exponential-phase S. typhimurium culture, a fresh gas-permeable membrane was applied, and cocultures were returned to incubation for 1 h before treatment
with 2 g/mL ciprooxacin. Cultures were incubated 6 h with
ciprooxacin before dilution plating onto selective media (25 g/
mL spectinomycin for E. coli or 25 g/mL chloramphenicol for
S. typhimurium).
The S. typhimurium LT2 narV:CmR knockout was generated
to allow separation of bacterial strains. The narV gene was selected for deletion because the activity encoded by this gene
(nitrate reductase) is functionally redundant in S. typhimurium
LT2 (16).
Growth of S. typhimurium on Conditioned Media. Conditioned
media was prepared by inoculating 1 mL M9CG + 1 mM tryp1 of 7

tophan 1:200 from overnight cultures of LT2, EMG2 wild type,

and EMG2 tnaA and growing to stationary phase. S. typhimuriumconditioned media was used as the control condition. Cultures
were pelleted in a tabletop centrifuge (2 min at 6,000 g), and
supernatant was sterilized by syringe ltration with a 0.22-m
lter. Sterility of conditioned media was conrmed by incubating
ltered media overnight at 37 C under standard culture conditions. S. typhimurium was inoculated 1:200 from overnight
culture into 150 L M9CG in 96-well plate and covered with
a gas-permeable membrane before incubation. After 1 h, these
cultures were mixed with conditioned media (30% exponentialphase culture in 70% conditioned media, nal volume 150 L);
cultures were incubated 3.5 h further before treatment with ciprooxacin (0.5 g/mL) to assess tolerance. All cultures were
incubated at 37 C with shaking.
Indole Quantication. Indole quantication was performed via
HPLC using a protocol based on (17), (18), and (19). Cell-free
medium for quantication of extracellular indole was obtained
by centrifugation of whole culture (6,000 g, tabletop centrifuge) and purication of the resulting supernatant on Spin-X
lter columns. HPLC was performed using a Waters C18 (5 m,
4.6 150) column with H20 + 0.1% (vol/vol) formic acid/acetonitrile (1:1) as the isocratic mobile phase. Under these conditions, in M9 + 0.2% casamino acids + 0.4% glucose, indole
eluted at 4.8 min as a single peak at 271 nm. A series of indole
standards (12.5600 M) in sterile growth medium was used to
estimate indole concentration as a linear function of peak height.
Quantitative PCR. RNA was collected from indole-treated and
untreated cultures of S. typhimurium LT2. Cultures were inoculated 1:500 from overnight LB cultures into 1 mL of M9CG
in 14-mL Falcon tubes and incubated 3.5 h before treatment with
indole. After 30-min incubation with indole (0, 50, or 125 M),
cultures were stabilized with RNAprotect Bacteria Reagent
(Qiagen) according to the manufacturers protocol, and the resulting cell pellets were stored overnight at 80 C. RNA extraction was performed using RNeasy Mini Kit (Qiagen), and
DNA contamination was eliminated using TURBO DNA-Free
(Ambion) according to the manufacturers protocols. Sample
concentration was estimated using the ND-1000 NanoDrop
spectrophotometer. Standard PCR using Taq polymerase and
qPCR primers (see below for primer information) was used to
test for DNA contamination in RNA samples after TURBO
DNA-Free digestion (4% RNA suspension by volume, 3035
cycles). RNA was stored at 80 C.
cDNA for qPCR was synthesized from RNA using the SuperScript III First Strand Synthesis kit (Invitrogen) and stored
at 20 C. Guanylate kinase (gmk), DNA gyrase subunit B
(gyrB), and 16S ribosomal RNA (rrsH) were used as reference
genes for S. typhimurium. Quantitative PCR primers for each
transcript of interest and the reference transcripts were designed based on the NCBI E. coli K-12 MG1655 genomic sequence (Refseq NC_000913) or the NCBI S. typhimurium LT2
genomic sequence (Refseq NC_003197) using Primer3Plus
software (20) (Table S1). Primers were designed under the
following constraints: amplicon size was 130160 bp, the calculated primer melting temperature was 55 C, GC content
was 4555%, and probabilities of primer-dimer/hairpin formations were minimized. Primer specicity was conrmed via
standard Taq polymerase PCR on genomic DNA. qPCR using
LightCycler 480 SYBR Green I Master Kit (Roche Applied
Science) were prepared manually according to the manufacturers instructions. qPCR were performed with a total volume of 20 L, containing 0.5 M of forward primers and 0.5
M of reverse primers and 10 L 2 480 SYBR Green Master
Mix. Reactions were carried out in white LightCycler 480 96-well
plates (Roche). One negative control (replacing cDNA with
Vega et al. www.pnas.org/cgi/content/short/1308085110

PCR H2O) was performed for each primer set in all qPCR runs.
PCR parameters were denaturation (95 C for 10 min) and 30
35 cycles of three-segment amplication (95 C for 10 s, 5052 C
for 10 s, and 72 C for 10 s). The thermal-cycling program was
concluded with a dissociation curve (65 C ramped to 95 C, 10 s
at each 1 C interval) to detect nonspecic amplication or
primer-dimer formation. Ct values were obtained using standard
methods (21). Alkyl hydroperoxide reductase (ahpF) was excluded from nal analyses because induction of this gene could
not be detected after treatment with 60 M hydrogen peroxide,
possibly due to error in the primers used here (Fig. S3).
Induction of OxyR. Hydrogen peroxide treatment was performed
as follows. Cultures were inoculated 1:200 (E. coli) or 1:500
(S. typhimurium) from overnight LB cultures into M9CG (1 mL
in 14-mL Falcon tubes or 150 L in 96-well plates) and allowed
to grow to exponential phase (3.5 h) or stationary phase (24 h) at
37 C. Cultures were incubated 1 h with the indicated concentration of hydrogen peroxide before treatment with antibiotics.
Serial dilution plating was performed before and after 1 h of
hydrogen peroxide incubation and after 3 h of ooxacin treatment to determine survival at each stage.
C. elegans Intestinal Model for S. typhimurium Infection. The temperature-sensitive germ line proliferation (glp) mitogen-activated
protein kinase kinase (sek) mutant strain AU37 [glp-4 (bn2) I;
sek-1 (km4) X] of C. elegans (Caenorhabditis Genetics Center)
was used as a model organism for Salmonella pathogenesis in the
host intestine. C. elegans cultures were synchronized from 1014
100-cm agar plates using a standard protocol (22). Synchronized
L1 larvae were washed to remove dauer pheromone and grown
34 d in S-medium (22) + concentrated E. coli OP50 at 25 C
with shaking at 200 rpm to allow the larvae to reach adulthood.
Synchronized cultures of adult C. elegans were then sucrosewashed to remove OP50 and other debris (23) and resuspended
in 1 mL S-medium with E. coli EMG2 tnaA Pro as a food
source (50 concentrated from overnight stationary-phase
cultures in LB, 100 L food per mL worm culture). Indole (0 or
125 M) was added with bacteria, and worm cultures were
incubated 24 h at 25 C to acclimatize worms. S. typhimurium
narV:CmR was then grown to late exponential/early stationary phase in LB, then spun down and resuspended at 50
concentration; 25 L of concentrated S. typhimurium cultures
were introduced to 1 mL worm culture, and cultures were incubated for 12 h at 25 C to allow establishment of the initial
infection, after which cultures were washed 4 in M9 worm
buffer (22) to remove external pathogens and incubated 24 h
further on E. coli (100 L 50 food per mL worm culture as
previously described) in S-medium indole to allow Salmonella
infection to progress. Cultures were then treated with 2 g/mL
ciprooxacin; higher concentrations were used here than in in
vitro experiments because the worm represents a barrier to diffusion of the drug.
At 0 and 24 h antibiotic treatment, samples of C. elegans
cultures were washed 6 in M9 worm buffer to remove external
bacteria, and internal bacteria were harvested by mechanical
disruption of worms in M9 worm buffer + 0.5% (vol/vol) Triton
X-100 + 5% (wt/vol) SDS. Internal bacteria were pelleted by
centrifugation (2 min at 13,500 g in a tabletop centrifuge),
washed 3 in M9 worm buffer to remove Triton X-100 and SDS,
and dilution-plated on selective media to determine cfu/mL.
Subsamples (25100 L) of undisrupted worm culture were
paralyzed with 10% sodium azide, dropped on LB agar, and
counted under a dissection microscope to determine worms/mL;
this number was used to calculate average cfu/worm.
These experiments were repeated using E. coli EMG2 wild
type or tnaA pZS4-mCherry as a food source and S. typhimurium
pZA21-GFP as the infectious agent. After 2448 h development
2 of 7

of Salmonella infection, samples of worm culture were washed and

mounted on 2% agar slides for uorescent microscopy. Images
were collected at 10 magnication [Plan apochromat (Apo) VC
100] using an Eclipse Ti epiuorescence inverted microscope
(Nikon Instruments, Inc.) equipped with the Perfect Focus
System, a ProScan II three-dimensional (XYZ)-motorized stage
(Prior Scientic, Inc.), and a CoolSNAP HQ2 charge-coupled
device (CCD) camera (Photometrics) and controlled by the Nikon
Instruments software (NIS)-Elements Advanced Research software (Nikon Instruments, Inc.). Images were acquired in the
differential interference contrast (DIC) conguration and in the
GFP (FITC) and mCherry (TRITC) uorescent channels, using a
motorized uorescence excitation and emission lter turret with
a uorescent channel lter set (Chroma Technology Corp.), a
Lambda SC SmartShutter (Sutter Instrument), and an Intensilight C-Mercury Fiber Illuminator, motorized (HGFIE) mercury
uorescent lamp (Nikon Instruments, Inc.). Images were autoscaled after acquisition to reduce background in the uorescence channels. Quantitation of relative bacterial load per
worm from the uorescence microscopy channels was deemed
intractable due to the ubiquitous presence of saturated pixels
and the inadequacy of single 2D epiuorescence images to

convey infection intensity across the z axis in the nematode

intestine.
To determine whether indole production by E. coli could affect
Salmonella persistence in the worm model, synchronized cultures
of adult C. elegans were incubated in modied S-medium (pH
7, adjusted with 10N NaOH) with 01 mM tryptophan using
E. coli EMG2 Pro or tnaA Pro as a food source. When prepared
according to standard protocols, fresh S-medium has pH 5.5,
and indole production in E. coli is inefcient under acidic conditions (24, 25). Cultures were incubated as previously described
for 24 h to allow production of indole before infection. After
24 h, S. typhimurium narV:CmR (50 concentrated culture
prepared as previously described) was introduced, and infection
was permitted to develop for 48 h; cultures were not washed to
remove external pathogens during this step, as this would remove
any E. coli-produced indole from the medium. Antibiotic treatment, worm disruption, and cfu/worm counts were performed as
described above. T tests conducted for these data were one-sided,
as data from batch culture experiments indicated that the presence
of indole could be expected to increase survival of Salmonella,
and this expectation could be tested against a null hypothesis of
no difference.

1. Bachmann BJ (1972) Pedigrees of some mutant strains of Escherichia coli K-12.

Bacteriol Rev 36(4):525557.
2. Hayashi K, et al. (2006) Highly accurate genome sequences of Escherichia coli K-12
strains MG1655 and W3110. Mol Syst Biol 2:2006.0007.
3. Swords WE, Cannon BM, Benjamin WH, Jr. (1997) Avirulence of LT2 strains of
Salmonella typhimurium results from a defective rpoS gene. Infect Immun 65(6):
24512453.
4. Wilmes-Riesenberg MR, Foster JW, Curtiss R, 3rd (1997) An altered rpoS allele contributes
to the avirulence of Salmonella typhimurium LT2. Infect Immun 65(1):203210.
5. Baba T, et al. (2006) Construction of Escherichia coli K-12 in-frame, single-gene
knockout mutants: The Keio collection. Mol Syst Biol 2:2006.0008.
6. Datsenko KA, Wanner BL (2000) One-step inactivation of chromosomal genes in
Escherichia coli K-12 using PCR products. Proc Natl Acad Sci USA 97(12):66406645.
7. Kibbe WA (2007) OligoCalc: An online oligonucleotide properties calculator. Nucleic
Acids Res 35(Suppl 2):W43W46.
8. Lutz R, Bujard H (1997) Independent and tight regulation of transcriptional units in
Escherichia coli via the LacR/O, the TetR/O and AraC/I1-I2 regulatory elements. Nucleic
Acids Res 25(6):12031210.
9. Wiuff C, Andersson DI (2007) Antibiotic treatment in vitro of phenotypically tolerant
bacterial populations. J Antimicrob Chemother 59(2):254263.
10. Andrews JM (2001) Determination of minimum inhibitory concentrations. J Antimicrob
Chemother 48(Suppl 1):516.
11. Keren I, Kaldalu N, Spoering A, Wang YP, Lewis K (2004) Persister cells and tolerance
to antimicrobials. FEMS Microbiol Lett 230(1):1318.
12. Keren I, Shah D, Spoering A, Kaldalu N, Lewis K (2004) Specialized persister cells and
the mechanism of multidrug tolerance in Escherichia coli. J Bacteriol 186(24):81728180.
13. Ashby MJ, Neale JE, Knott SJ, Critchley IA (1994) Effect of antibiotics on nongrowing planktonic cells and biolms of Escherichia coli. J Antimicrob Chemother
33(3):443452.

14. Eng RH, Padberg FT, Smith SM, Tan EN, Cherubin CE (1991) Bactericidal effects
of antibiotics on slowly growing and nongrowing bacteria. Antimicrob Agents
Chemother 35(9):18241828.
15. Sufya N, Allison DG, Gilbert P (2003) Clonal variation in maximum specic growth rate
and susceptibility towards antimicrobials. J Appl Microbiol 95(6):12611267.
16. McClelland M, et al. (2001) Complete genome sequence of Salmonella enterica
serovar Typhimurium LT2. Nature 413(6858):852856.
17. Lee HH, Molla MN, Cantor CR, Collins JJ (2010) Bacterial charity work leads to
population-wide resistance. Nature 467(7311):8285.
18. Kobayashi A, Hirakawa H, Hirata T, Nishino K, Yamaguchi A (2006) Growth phasedependent expression of drug exporters in Escherichia coli and its contribution to
drug tolerance. J Bacteriol 188(16):56935703.
19. Hirakawa H, Kodama T, Takumi-Kobayashi A, Honda T, Yamaguchi A (2009) Secreted
indole serves as a signal for expression of type III secretion system translocators in
enterohaemorrhagic Escherichia coli O157: H7. Microbiology 155:541550.
20. Untergasser A, et al. (2007) Primer3Plus, an enhanced web interface to Primer3.
Nucleic Acids Res 35(Web Server issue, suppl 2):W71-4.
21. Livak KJ, Schmittgen TD (2001) Analysis of relative gene expression data using realtime quantitative PCR and the 2(- C(T)) Method. Methods 25(4):402408.
22. Stiernagle T (2006) Maintenance of C. elegans. WormBook, ed the C. elegans Research
Community (WormBook), 10.1895/wormbook.1.101.1. Available at www.wormbook.org.
23. Portman DS (2006) Proling C. elegans gene expression with DNA microarrays. WormBook, ed the C. elegans Research Community (WormBook), 10.1895/wormbook.1.104.1.
Available at www.wormbook.org.
24. Blankenhorn D, Phillips J, Slonczewski JL (1999) Acid- and base-induced proteins
during aerobic and anaerobic growth of Escherichia coli revealed by two-dimensional
gel electrophoresis. J Bacteriol 181(7):22092216.
25. Han TH, Lee JH, Cho MH, Wood TK, Lee J (2011) Environmental factors affecting
indole production in Escherichia coli. Res Microbiol 162(2):108116.

Vega et al. www.pnas.org/cgi/content/short/1308085110

3 of 7

St %Survival

0.06
0.05
0.04
0.03
0.02
0.01
0

LT2 CM

Effect
Size (d)

tnaA CM

EMG2 CM

0.69

11.33

Fig. S1. E. coli-produced indole in conditioned media induces antibiotic tolerance in Salmonella typhimurium. Conditioned media was prepared by inoculating M9CG + 1 mM tryptophan with S. typhimurium LT2, EMG2 wild type, and EMG2 tnaA and growing to stationary phase; S. typhimurium was grown
to exponential phase in 30% fresh/70% conditioned media and treated with ciprooxacin (0.5 g/mL). S. typhimurium-conditioned media was used as the
control condition, and Cohens effect sizes (d) were calculated with respect to this condition. Bars represent mean SD of seven biological replicates. Exposure
to tnaA-conditioned media did not have a large effect on Salmonella tolerance compared with exposure to S. typhimurium-conditioned medium (d = 0.69),
whereas exposure to media conditioned by wild-type E. coli did have a large effect on tolerance relative to the control (d = 11.33), indicating that most if not
all of the increase in tolerance upon exposure to E. coli is due to the effects of indole signaling.

b 700
Indole Standard (M)

a 500
Indole (M)

400
300
200
100
0
0
-100

0.5

600
500
400
300
200
100
0

-100

Tryptophan (mM)

1000

2000

3000

4000

Peak Height (271 nm)

Fig. S2. Indole production as determined by HPLC. (A) Indole concentrations in conditioned E. coli medium. Wild-type E. coli was grown to stationary phase
(16 h) in M9CG + indicated concentrations of tryptophan. (B) Linear regression against a series of indole concentration standards in M9CG was used to calculate
indole concentration in conditioned media (calculated regression line: m = 1.64, b = 0.127).

Fold Change Expression (-Ct)

-1

ahpC

ahpF

dps

katG

trxC

Fig. S3. qPCR of OxyR regulon genes after treatment of S. typhimurium with hydrogen peroxide. Exponential-phase (3.5 h) cultures of S. typhimurium in
M9CG were treated with 60 M H2O2 for 15 min. Bars represent mean SD of three biological replicates.

Vega et al. www.pnas.org/cgi/content/short/1308085110

4 of 7

125 M Indole

-Cipro

-Cipro

+Cipro

+Cipro

EMG2 + Trp

Fig. S4. Ciprooxacin treatment does not produce visually obvious changes in bacterial content of the C. elegans intestine after 24 h. All experiments were
performed in S-Medium (SI Materials and Methods) at 25 C to ensure sterility and survival of C. elegans AU37. Images were taken from ciprooxacin-treated
and untreated cultures 24 h after the addition of antibiotic and were selected to represent the variation observable in each experimental condition. DIC images
are shown overlaid with red and green uorescent channels. The lack of a clear difference in uorescence between ciprooxacin-treated and untreated
cultures may be the result of the high stability of the uorescence proteins and the fact that ciprooxacin does not cause lysis or signicant permeabilization of
bacteria cells. For these reasons, the uorescence of cells killed by ciprooxacin should resemble the uorescence of untreated cells. (A) C. elegans fed on E. coli
(mCherry) and infected with S. typhimurium (GFP) with 125 M indole added to media. (B) C. elegans fed on E. coli EMG2 (mCherry) and infected with
S. typhimurium (GFP) with 0.5 mM tryptophan added to media.

Vega et al. www.pnas.org/cgi/content/short/1308085110

5 of 7

Table S1. Bacterial strains, plasmids, and primers used in this study
Strains and primers

Genotype or sequence

E. coli K12 strains

EMG2
Wild type
EMG2 tnaA
K12 EMG2 tnaA
EMG2 Pro
K12 EMG2 Pro, SpR, lacR, tetR
EMG2 tnaA Pro
K12 EMG2 tnaA Pro, SpR, lacR, tetR
Salmonella typhimurium strains
S. typhimurium LT2
Wild type
LT2 narV
S. typhimurium narV CmR
LT2 oxyR
S. typhimurium oxyR CmR
LT2 pspBC
S. typhimurium pspBC CmR
Knockout conrmation primers
tnaA Forward
CCAGAGCCAAACCGATTAGA
tnaA Reverse
ACCATAACACCCCAAAATGC
narV Forward
CTGGGAAGAAGAACAGGTGAAG
narV Reverse
GAAGGCGTCAACATGGTAAATC
mCherry F internal
TGTACGGTTCCAAGGCCTAC
mCherry R internal
CCCATGGTCTTCTTCTGCAT
SpR F internal
GGTCACCGTAACCAGCAAAT
SpR R internal
GCAGTCGCCCTAAAACAAAG
CmR F internal
GGGCGAAGAAGTTGTCCATA
CmR R internal
TCCGGCCTTTATTCACATTC
KmR F internal
AATATCACGGGTAGCCAACG
KmR R internal
TGCTCCTGCCGAGAAAGTAT
oxyR detect F LT2
TAAGTGCGGTCGCGTACAAGG
oxyR detect R LT2
CCGCGGTATGGATGTGAAAGTG
pspBC detect F LT2
TCCTGTTGAGCCTGAAGAGAAG
pspBC detect R LT2
CGACATCGTGAACGCCAATATC
Primers for phage-mediated PCR product knockouts
LT2 narV Forward
TGAACGGGAAGAGCAGGAAAATAGTCATGTGTAGGCTGGAGCTGCTTCG
LT2 narV Forward extend
TACTCAAACGGCGCGCTCCAGACATGAACCAGACGGGTGAACGGGAAGAGCAGG
LT2 narV Reverse
TGCGCTATGACTACGGACAATATACCTGCGCGAATATCCTCCTTAGTTCCTATTCC
LT2 narV Reverse extend
TGCGCCACGGTATTCTTCCTCGGCAGTTGGCTGCGCTATGACTACGGAC
LT2 oxyR Upstream F
GGATCCACTAGTTGGCGTCATCATGCACCTG
LT2 oxyR Upstream R
GCCTACACGCATGCTGCGTAGCCGTTATGAGCAACTG
LT2 oxyR AbR F
GCTACGCAGCATGCGTGTAGGCTGGAGCTGCTTCG
LT2 oxyR AbR R
CGGGCGGTGTCCTCGAGCATATGAATATCCTCCTTAGTTCC
LT2 oxyR Downstream F
CATATGCTCGAGGACACCGCCCGTTCAGAGGG
LT2 oxyR Downstream R
GCGGTGGCGGCCGCATACCTGGCTTGACTGCCTGCATATGG
LT2 oxyR Insert F
AAACGCCCACGTTGTCAG
LT2 oxyR Insert R
CTATGCTAATGGCCGTACCG
LT2 pspBC Upstream F
GGATCCACTAGTCCATATTCATCGCGTGCGTATAC
LT2 pspBC Upstream R
GCCTACACGCATGCCTGTAAGTGATATATCGGGAAGG
LT2 pspBC AbR F
CTTACAGGCATGCGTGTAGGCTGGAGCTGCTTCG
LT2 pspBC AbR R
CATGAGCTCGAGCATATGAATATCCTCCTTAGTTCCTATTC
LT2 pspBC Downstream F
CATATGCTCGAGCTCATGCGTTCTCCTTACAC
LT2 pspBC Downstream R
GCGGTGGCGGCCGCGTTATCCCGTCGTATTGAAC
LT2 pspBC Insert F
ATTCATCGCGTGCGTATACCC
LT2 pspBC Insert R
ATTGAACAGGCTACGGCTCAG
Rescue plasmid construction
tnaA cloning F
CCGCGCCGTGCGGTACCATGGAAAACTTTAAACATCTCCCTG
tnaA cloning R
CGCGGCGTCGTCTAGAGCTAACATCCTTATAGCCACTCTG
Rescue plasmids
pZA21G
KanR; PltetO-1-GFP
pZA21-tnaA
derivative of pZA21G; tnaA gene cassette replaces GFP (KpnI/HindIII)
qPCR primers
LT2 pspA F
AGATCGAGTGGCAGGAAAAAG
LT2 pspA Reverse
GTGACTTCCTGTTCAAGCGTAG
LT2 pspE F
CAGACAGGAACGATACGGTAAA
LT2 pspE Reverse
ACTGATACCGCCCATATTCATC
LT2 ahpC F
CACAGCAGCTCTGAAACTATCG
LT2 ahpC Reverse
CTTCATCTTCACGCATGTTGTC
LT2 ahpF F
GAACGAAATCACCGAACGTAAC
LT2 ahpF Reverse
CCAGTATCCACTTTAGCGACAA

Vega et al. www.pnas.org/cgi/content/short/1308085110

6 of 7

Table S1. Cont.

Strains and primers
LT2
LT2
LT2
LT2
LT2
LT2
LT2
LT2
LT2
LT2
LT2
LT2
LT2
LT2
LT2
LT2

trxC F
trxC Reverse
dps F
dps Reverse
katG F
katG Reverse
napF F
napF Reverse
invF F
invF Reverse
ramA F
ramA Reverse
gmk F
gmk Reverse
gyrB F
gyrB Reverse

Genotype or sequence
ATCTTCCAGTGGTGATCGACTT
CTTTGACGAAACGGACTTTACC
CACTGACCGATCATCTGGATAC
CGGATAGCTTTTCAGTGGAGTT
GATCCGGAGTTCGAGAAGATTT
GCTTTTGGTCCCATATCTCTGT
GGATCTGGTTTTTACGCTCAC
GATAACGTGGGACGAAAGGTAA
GAATGCTGGGAGAAGACTATGG
ACGCCAGTTTCGTAATTCACTC
CACGATTGTCGAGTGGATTG
CCAGACTCTCCCCTTTGTACTG
CACGGTGAGCACTATTTCTTTG
AGTGCCGTAGTAATTGCCAAAC
AGCGATGGATCAGTACCAGATT
TGGCGTTATATTCAGAGACCAG

AbR, antibiotic resistance; F/R, forward/reverse; inv, invasion protein; KmR, kanamycin resistance; lacR, lactose repressor; mtr,
tryptophan/indole:H+ symporter; nap, nitrate reductase; ramA, resistance antibiotic multiple; SpR, spectinomycin resistance; tetR,
tetracycline repressor; trx, thioredoxin.

Vega et al. www.pnas.org/cgi/content/short/1308085110

7 of 7