Indole toxicity involves the inhibition of adenosine triphosphate production and protein folding in Pseudomonas putida


Correspondence: Woojun Park, Department of Environmental Science and Ecological Engineering, Korea University, Seoul 136-713, Korea. Tel.: +82 82 3290 3067; fax: +82 2 953 0737;



High concentrations of indole are known to be toxic to cells due to perturbations in membrane potential. Here, we report for the first time a transcriptome analysis of a soil model bacterium, Pseudomonas putida KT2440, under indole treatment. We demonstrated that 47 genes are differentially expressed, including 11 genes involved in the tricarboxylic acid cycle (TCA cycle) and 12 genes involved in chaperone and protease functions (hslV, hslU, htpG, grpE, dnaK, ibpA, groEL, groES, clpB, lon-1, lon-2, and hflk). Mutant analysis supported the observation that protease genes including hslU are essential for the indole resistance of Pseudomonas strains. Subsequent biochemical analyses have shown that indole increases the NADH/NAD+ ratio and decreases the adenosine triphosphate (ATP) concentration inside cells, due to membrane perturbation and higher expression of TCA cycle genes in the presence of indole. This energy reduction leads to a reduction in cell size and an enhancement of biofilm formation in P. putida. The observed upregulation in many chaperones and proteases led us to speculate that protein folding might be inhibited by indole treatment. Interestingly, our in vitro protein-refolding assay using malate dehydrogenase with purified GroEL/GroES demonstrated that indole interferes with protein folding. Taken together, our data provide new evidence that indole causes toxicity to Pputida by inhibiting cellular energy production and protein folding.


Indole is widespread in nature, because various species of bacteria produce considerable amounts of indole (Lee & Lee, 2010). Indole is known to participate in various biological events such as biofilm formation (Lee et al., 2007), pathogenicity (Chu et al., 2012), plasmid stabilization (Field & Summers, 2012), spore formation (Stamm et al., 2005), acid resistance (Hirakawa et al., 2010), and persister cell formation (Vega et al., 2012). Indole has received great attention owing to the broad spectrum of its effects on bacterial physiology. Recent interesting findings also suggest that indole plays an important role in bacterial persister cell formation under antibiotic treatment (Vega et al., 2012). Interestingly, a tnaA mutation (non-indole-producing) in Escherichia coli decreased persister cell formation (Vega et al., 2012). Cells of E. coli that are lysed by antibiotics protect the majority of neighboring cells of the same kind by producing indole as a defense signal molecule (Lee et al., 2010a). Stationary-phase cells of E. coli under nutrient-rich conditions could also produce indole, which might be linked to stress defense (Vega et al., 2012). It has been speculated that the survival of bacterial population might be enhanced for turning on antibiotic defense systems, such as multidrug efflux pumps and oxidative stress–protective pathway by producing indole (Lee et al., 2010a).

Indole has been reported to function as an intercellular signal in bacterial quorum sensing. There might be cross talk between indole-based and acyl-homoserine lactone (AHL)-based signaling, because a recent study demonstrated that indole-producing E. coli could inhibit quorum sensing–regulated virulence factors of Pseudomonas aeruginosa (Chu et al., 2012). The E. coli LuxR homolog, SdiA, is considered to be related to this indole signaling (Lee et al., 2007). In our previous study, the addition of exogenous indole increased the expression of the ppoR gene, an sdiA homolog, in Pseudomonas putida (Lee et al., 2010b). Furthermore, increased biofilm formation and reduced swimming motility were also observed upon treatment with indole, but these effects were inhibited by the expression of the cviI gene (encoding N-hexanoyl homoserine lactone synthase, HHL) in P. putida. Thus, we speculated that the PpoR-HHL complex inhibits the effects of indole, and that indole may act as a signal via PpoR (Lee et al., 2010b). However, there is still no direct evidence showing that indole can bind to any SdiA homolog, and it is unclear how indole and SdiA act together in controlling many cellular processes. In contrast to all these observations, it has been recently argued that SdiA might not respond to indole in E. coli and in the Salmonella enterica serovar Typhimurium (Sabag-Daigle et al., 2012). Thus, the detailed linkage between indole signaling and bacterial quorum sensing remains unclear.

It has been reported that millimolar concentrations of indole are present in human intestines (Sabag-Daigle et al., 2012). The Mtr transporter might be involved in indole import and export in E. coli (Yanofsky et al., 1991). However, indole has been recently known to be freely diffusible across bacterial membranes (Piñero-Fernandez et al., 2011). Indole has been reported to act as a proton ionophore at high concentrations and inhibit cell division (Chimerel et al., 2012). The electrochemical potential is reduced when indole passes through the cytoplasmic membrane. Thus, high indole concentrations are toxic to bacterial cell growth, apart from the fact that indole functions as a signal. Here, we report the first transcriptome data of P. putida upon indole treatment and directly show that exogenous indole can cause a reduction in adenosine triphosphate (ATP) production by membrane perturbation. Furthermore, we demonstrated that indole inhibits the protein-folding process inside cells, which might cause the higher expression of many proteases in P. putida.

Materials and methods

Bacterial strains, culture conditions, and DNA manipulation

The bacterial strains and plasmids utilized in this study are shown in Table S1. Pseudomonas putida KT2440 and P. aeruginosa PAO1 were grown at 30 and 37 °C in LB and modified M9 media [Na2HPO4·7H2O (6.8 g L−1), KH2PO4 (3 g L−1), NaCl (0.5 g L−1), NH4Cl (1 g L−1), MgSO4 (2 mM), and CaCl2 (0.1 mM)] containing 10 mM glucose with aeration by shaking. When required, antibiotics were added at the following concentrations: kanamycin (50 μg mL−1) and ampicillin (100 μg mL−1). Indole was added to the growth medium at various concentrations (0.25–5 mM). Growth was monitored by measuring the OD600 nm of the cultures using a BioPhotometer (Eppendorf, Hamburg, Germany).

HslU mutant construction

The primers utilized in this study are listed in Table 1. A 589-bp fragment of the internal region of the hslU gene was amplified using the PP_5001 (hlsU-SC)-F and PP_5001 (hlsU-SC)-R primers. The polymerase chain reaction (PCR) product for the hslU mutant was digested with the EcoRI and KpnI restriction enzymes. Each fragment was subsequently inserted into a pVIK112 vector via ligation. The constructed plasmids were then transformed into the E. coli S17-1 λ pair. Conjugation was performed using the biparental filter mating method with the KT2440R strain.

Indole sensitivity test

To conduct sensitivity tests, all cells were collected at the exponential phase (OD600 nm = 0.4) and then washed three times with PBS. Cells were inoculated into PBS at approximately 107 CFU mL−1 and serially diluted. The diluted samples were spotted on LB plates containing different concentrations of indole. Then, cells were incubated at 30 °C for 24 h.

Microarray analysis

The cells were grown to the exponential phase (OD600 nm = 0.4) at 30 °C with aeration. The cells were then treated with 1 mM indole for 10 min. Total RNA was isolated using the RNeasy Mini kit (Qiagen, Valencia, CA) according to the manufacturer's instructions. Following procedure was conducted, as previously described (Lee et al., 2010b) and detailed methods were located in Supporting Information. Genes that showed changes of more than 1.5-fold (upregulated genes) and < 0.67-fold (downregulated genes) in at least two replicates were selected. The microarray data were deposited in the National Center for Biotechnology Information (NCBI) GEO site (under accession number GSE 41254). Microarray data were confirmed by quantitative reverse transcriptase PCR for the 12 heat-shock and protease genes. Detailed experimental procedure described in Supporting Information.

Measurement of the NADH/NAD+ ratio

Nicotinamide adenine dinucleotide (NAD+) and NADH concentrations were measured using the EnzyChrom™ NAD+/NADH Assay kit (BioAssay Systems, Hayward, CA) according to the manufacturer's instructions. Exponentially growing cultures were treated with 1 mM indole, 1 mM tryptophan, and 1 mM indole acetic acid (IAA) for 30 min, and cells were collected for NADH and NAD+ extraction. Approximately 108 cells for each condition were harvested after treatment and washed with cold PBS. The samples were homogenized in either 100 μL NAD extraction buffer for NAD determination, or 100 μL NADH extraction buffer for NADH determination. Then, extracts were heated at 60 °C for 5 min, and 20 μL assay buffer and 100 μL of the corresponding extraction buffer were added to neutralize the extracts. The samples were centrifuged at 14 000 g for 5 min, and the supernatants used for NAD+/NADH assays. The assay is based on a lactate dehydrogenase cycling reaction, in which the formed NADH reduces a formazan (3-(4,5-Dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide; MTT) reagent. The intensity of the reduced product color, measured at 565 nm, and the change in absorbance during the reaction between enzyme and substrates were calculated from the standard curve. A microplate spectrophotometer (PowerWaveXS; Bio-Tek, Winooski, VT) was used for measuring optical density. NAD+ and NADH concentrations were normalized by the amount of protein.

Determination of ATP concentrations

To measure intracellular ATP concentrations, the ENLITEN ATP assay system bioluminescence detection kit (Promega, Madison, WI) was used in accordance with the manufacturer's instructions. To confirm ATP concentrations, exponentially growing cultures were treated with 0.5, 1, 2, and 3 mM indole for 30 min and cells were harvested for ATP extraction using 1% trichloroacetic acid (TCA) buffer. Before carrying out the assay, samples in TCA buffer were diluted fivefold with Tris-acetate buffer to neutralize the extracts. The luminescence was measured using a microtiter plate reader (VICTOR3; Bio-Rad). The ATP concentration was expressed as molar concentration per mg of protein.

Malate dehydrogenase denaturation, refolding, and enzymatic assays

To demonstrate that indole affects protein folding, we used mitochondrial malate dehydrogenase (mMDH) as a substrate in the GroE-assisted protein-folding reaction. The protein-folding assay was performed as described by Diamant et al. (1995, 2001). mMDH (35 μM) was denatured at 25 °C for 2 h in 6 M guanidium HCl containing 40 mM MOPS buffer and 20 mM dithiothreitol and incubated at 47 °C for 30 min to ensure complete denaturation. The refolding reaction was initiated by 100-fold dilution of the guanidium/heat-treated mMDH into 40 mM MOPS buffer, pH 7.5, 200 mM KCl, 4 mM MgCl2 1 mM ATP, and 5 μL of a GroEL/GroES 1 : 1 mixture (25 mg mL−1; Sigma, St. Louis, MO) at 37 °C. During the refolding process, 1 mM indole and ethanol were added. The activity of mMDH was assayed at 25 °C in 40 mM MOPS buffer, pH 7.5, 10 mM dithiothreitol, 0.5 mM oxaloacetate, and 0.28 mM NADH. Native malate dehydrogenase (MDH) was used as an enzyme activity control. The time-dependent oxidation of NADH by MDH was monitored at 340 nm. The refolding activity was expressed as a relative percentage, taking the activity of native MDH after 60 min of refolding to be 100%.

Biofilm formation

Biofilm formation of the KT2440 strain was analyzed as described by Jackson et al. (2002). Bacterial cells were inoculated into LB or modified M9 media with glucose (10 mM) as a carbon source and incubated for 24 h at 30 °C. The cultures were then washed with PBS, and 106 CFU mL−1 of cells were inoculated into LB or M9 with 10 mM glucose. Bacterial cultures were grown in 96-well polystyrene microtiter plates (Costar, Washington, DC) for 24 h (1 day), 48 h (2 days), and 72 h (3 days) at 30 °C in static conditions. Biofilm formation was measured by staining the attached cells with crystal violet. After staining, the attached cells were resuspended in ethanol, and the absorbance was measured at 595 nm. A microplate spectrophotometer (PowerWaveXS; Bio-Tek) was used for measuring the optical density.

Microscopic observation

Indole was added to the cells at the exponential phase (OD600 nm = 0.4), and the cells were then incubated for 2 h at 30 °C. One milliliter of cells was collected and washed twice with PBS, and 2 μg mL−1 of 4′,6-diamidino-2-phenylindole (DAPI) was added to the cell suspension, followed by staining for 10 min by rocking at room temperature. Excess DAPI was removed by washing, and pellet was resuspended in PBS. Then, 5 μL of cells was placed on a glass slide and observed.


The effect of indole toxicity on the growth of P. putida

Indole is produced by bacteria that possess tryptophanase (TnaA), which catalyzes the synthesis of indole from tryptophan. Pseudomonas putida KT2440 does not have the tnaA gene in its genome. To confirm that KT2440 cells cannot produce indole, Kovács reagent was used (Gabriel & Gadebusch, 1956). In the presence of indole-producing bacteria (Vibrio harveyi and E. coli O157:H7 Sakai), a cherry-red layer is formed on top of the medium following the addition of Kovács reagent. We observed that KT2440 cells cannot produce indole (data not shown). The growth of KT2440 cells in LB media with various indole concentrations (0–5 mM) was measured to determine the minimal concentration of indole toxicity. Millimolar concentrations of indole have been used in many previous studies (Lee & Lee, 2010), and they reflect the actual concentrations in animal intestines (Sabag-Daigle et al., 2012). No apparent toxicity of indole was observed at 0.25, 0.5, and 1 mM indole, but the growth rate of KT2440 cells sharply decreased at 2 mM indole (Fig. 1a). The KT2440 cells did not grow in indole concentrations of 3 mM or greater during 24-h incubation. The effect of ethanol as a solvent for indole was evaluated; no difference was observed at 3 mM indole (data not shown). We observed some degradation of indole following long-term (5 days) incubation of KT2440 cells (Fig. S1). However, the cells could not grow well in LB containing 3 mM indole (OD600 nm, < 0.1), which suggested indole inhibition cannot be overcome by longer-term growth. KT2440 cells grown on agar plates with increasing concentrations of indole were also sensitive to indole (Fig. 1b), and surviving colonies on indole-containing agar plates were smaller than in the control. Thus, high concentration of indole exhibits toxic effects on the growth of KT2440 cells in both liquid and solid media.

Figure 1.

The effect of indole toxicity on the growth of Pseudomonas putida KT2440. (a) Growth curve of KT2440 in LB media containing various concentrations of indole (0–5 mM). Cells were grown at 30 °C, and the growth at each condition was monitored by measuring the OD600 nm of cultures. EtOH control indicates treatment with ethanol instead of indole, corresponding to 5 mM indole. (b) Indole sensitivity test. When the cells reached the exponential phase (OD600 nm = 0.5), they were diluted serially, spotted onto individual plates.

Transcriptome analysis of Pseudomonas putida under indole treatment

Microarray analysis of KT2440 cells was conducted to investigate their response to indole. Our duplicate microarray data demonstrated that a total of 47 genes in indole-treated KT2440 cells were either upregulated more than 1.5-fold or downregulated < 0.67-fold, compared with control (Table S2). The highest gene expression among all upregulated genes was found in trpB (PP_0083, tryptophan synthase beta subunit; 3.52-fold upregulated), which is involved the synthesis of tryptophan from indole. Interestingly, 11 genes involved in the TCA cycle (kgdAB, PP_4012, more than twofold; sdhABD, sucCD, gltA, lpdG, and PP_0897, more than 1.5-fold and less than twofold) and 12 heat-shock or protease genes (hslV, hslU, htpG, grpE, dnaK, ibpA, groEL, groES, clpB, lon-1, lon-2, and hflk) were upregulated upon indole treatment. Our microarray data were confirmed by qRT-PCR for the 12 heat-shock and protease genes (Fig. 2a). To uncover the functions of these upregulated genes during indole treatment, P. aeruginosa PAO1 mutants from the Washington University Genome Center were obtained and their sensitivities to indole were checked. The sensitivity test was conducted on the PAO1 mutants on plates containing different concentrations of indole (0, 1, 3, and 5 mM; 0 and 3 mM indole-containing plates are shown in Fig. 2b). All the mutants were unable to grow in the presence of 5 mM indole. Among these mutants, the PA0036 (trpB, tryptophan synthase subunit beta), PA0796 (prpB, carboxyphosphonoenolpyruvate phosphonomutase), PA5053 (hslV, ATP-dependent protease peptidase subunit), and PA5054 (hslU, ATP-dependent protease ATP-binding subunit) strains were very sensitive to 3 mM indole, compared with wild type. PrpB catalyzes the synthesis of a phosphonate (C-P) bond (Hidaka et al., 1990) and has been shown to be required for the catabolism of propionate in Salmonella typhimurium along with prpBCDE operon (Horswill & Escalante-Semerena, 1997). However, very little is known about regulation and expression of the PrpB gene. Our data suggest that these indole-sensitive gene products play an important role in defense against indole toxicity, although other gene functions are unknown.

Figure 2.

(a) Quantitative reverse transcriptase PCR analysis of KT2440 cells treated with 1 mM indole or no indole. Relative expression via qRT-PCR (black bar) of 12 genes involving heat-shock proteins or proteases in KT2440 is consistent with microarray data (gray bar). (b) The sensitivity test was conducted on plates with various indole concentrations (1, 3, and 5 mM) in Pseudomonas aeruginosa PAO1 mutants. All mutant cells were selected based on the Pseudomonas putida microarray results. Serial dilutions of all mutants were spotted and grown at 37 °C.

The absence of HslU promotes indole sensitivity and cell filamentation

To ascertain the effects of the proteases identified from the indole microarray and our sensitivity tests, the hslU mutant of KT2440 was constructed. In E. coli, the HslVU protease is an ATP-dependent protease (Missiakas et al., 1996). HslV is known to have peptidase activity and HslU has ATPase activity. Both HslV and HslU participate together in the degradation of abnormal polypeptides (Missiakas et al., 1996). Disruption of HslU decreased the growth of KT2440 cells in LB media (Fig. 3a). The growth rates (h−1) of wild-type and hslU mutant KT2440 cells were 1.47 ± 0.05 and 0.98 ± 0.03, respectively. Unlike wild-type KT2440 cells, the hslU mutant showed growth retardation upon the addition of 1 mM indole (at time points between 8 and 10 h; Fig. 3a). Furthermore, the growth of hslU mutant was severely inhibited by high concentration of indole (> 2 mM; Fig. S2). When both the wild-type and the hslU mutant cells were incubated on indole-containing agar plates, the hslU mutant was sensitive to indole, compared with the wild type (Fig. 3b). Our data showed that the hslU mutant of KT2440 cells is more sensitive to indole in both liquid and solid media. To confirm that HslU functions as a heat-shock protein, the indole sensitivity test was performed at temperatures higher (35 °C, 37 °C) than the optimum temperature (30 °C). However, no dramatic difference was observed (data not shown). Indole has recently been shown to inhibit E. coli cell division by acting as a proton ionophore (Chant & Summers, 2007; Chimerel et al., 2012; Field & Summers, 2012). We investigated the effect of indole on the cell division of KT2440 cells. DAPI staining for microscopic observation showed that indole results in smaller cell sizes in the wild type (Fig. 3c). The HslVU protease degrades SulA (Wu et al., 1999), which is known to inhibit FtsZ ring formation. FtsZ has GTPase activity that is important for Z-ring formation during cell division (Bi & Lutkenhaus, 1993). Thus, we speculated that the absence of HslU might increase cell filamentation due to the lack of SulA degradation. The hslU mutant cells were elongated approximately twofold compared with wild-type cells (Fig. 3c), although cell size reduction by indole was not changed in the mutant, suggesting that HslU alone does not directly control cell size reduction by indole.

Figure 3.

Influence of indole on growth and cell size in wild-type and hslU mutant KT2440 cells. (a) Growth curve of wild-type KT2440 and the hlsU mutant. To compare the change in growth patterns of wild-type KT2440 and the hslU mutant, 1 mM indole was added at the exponential phase. (b) Indole sensitivity test comparison between wild-type and hslU mutant cells. (c) Microscopic observation (scale bar, 10 μm) of indole-treated wild-type and hslU mutant cells. The average cell size was measured from 30 cells. The error bar indicates standard deviation; scale bar indicates 10 μm.

Indole perturbs energy pools in bacterial cells

The expression of genes involved in the TCA cycle led us to examine the NADH/NAD+ ratio and ATP levels under indole treatment. The NADH/NAD+ ratio plays a critical role in the regulation of cell metabolism (Ebert et al., 2011). We observed higher NADH/NAD+ ratio by increasing the concentration of indole (Fig. 4a). The NADH/NAD+ ratio was 3, 50-fold higher with 2 and 3 mM indole, respectively. However, there was no change upon treatment with the structurally similar compounds tryptophan and indole-3-acetic acid. Furthermore, the NADH/NAD+ ratio also showed no alteration in the presence of other amino acids (1 mM) such as glycine, cysteine, and phenylalanine (data not shown). Indole can interact with lipid membranes (Gaede et al., 2005; Mitchell, 2009). The increase in the NADH/NAD+ ratio might be due to the inhibition of electron transport in the membrane by indole, which lowers ATP production inside the cells.

Figure 4.

Indole affects the energy complexes of Pseudomonas putida KT2440. (a) The effect of indole on the NADH/NAD+ ratio. The NADH/NAD+ ratio was determined by measuring the absorbance at 565 nm, and each concentration was calculated from the standard curve. NAD+ and NADH concentrations were normalized to the amount of protein (mg). (b) Influence of indole on ATP generation. To determine ATP concentrations, exponentially growing cultures were treated with 0 and 3 mM indole for 30 min. The ATP concentration was expressed as molar concentration per mg of protein. All data show the average of four replicates, and one standard deviation is shown.

Internal levels of ATP were reduced in the presence of indole (Fig. 4b). We speculate that increased ATP consumption by many ATP-dependent proteases (Table S2, Fig. 2a) and inhibition of ATP synthesis by reduction in membrane potentials (Fig. 4) occurred in the presence of a high concentration of indole. There was no significant change in ATP level in the presence of lower concentrations of indole (< 2 mM, data not shown).

In our previous study, the addition of indole promoted biofilm formation in KT2440 cells (Lee et al., 2010b). Furthermore, others have shown that indole increases biofilm formation in other pseudomonads including P. aeruginosa and P. fluorescens (Lee et al., 2007). In contrast, biofilm formation and cell adhesion decreased in E. coli upon treatment with indole (Domka et al., 2006; Bansal et al., 2007; Lee & Lee, 2010; Lee et al., 2010b), suggesting that indole acts differently in different bacterial species. Biofilm formation in KT2440 cells increased in the presence of indole in both rich and minimal media (Fig. S3A,B). The changes in NADH/NAD+ levels and ATP concentrations along with the small cell and colony size imply that cells experienced metabolic burdens or stresses. Thus, high indole concentration stresses bacterial cells and affects their adherence to some surfaces.

Indole inhibits protein folding

Indole is a structurally simple and hydrophobic molecule. It has a bicyclic structure, consisting of a benzene ring fused to a nitrogen-containing pyrrole ring. It is generally accepted that proteins are organized with a hydrophobic core and a hydrophilic exterior (Meirovitch & Scheraga, 1981). Our microarray data, which revealed that many protease genes are highly upregulated, led us to speculate that indole can intercalate into proteins during the process of folding, which could interfere with hydrophobic interactions that are needed for folding. Unfolded proteins are known to induce the bacterial heat-shock response (Neidhardt et al., 1984; Goff & Goldberg, 1985; VanBogelen et al., 1987).

To test whether indole had any effect on chaperone-mediated protein folding, a MDH refolding assay was performed with the GroEL/GroES chaperones. The data showed an increase in MDH refolding with time (Fig. 5a). After 60 min of MDH refolding in the presence of the GroEL/GroES chaperones, up to 70% of native MDH activity was regained (Fig. 5b). Although GroEL and GroES supported MDH refolding in the presence of indole, refolding rates were lower than those seen with only chaperones (Fig. 5a). The same volume of ethanol corresponding to that of indole treatment did not affect mMDH refolding (data not shown). In our microarray data, 12 heat-shock or protease genes including GroEL and GroES were upregulated by indole. Our transcriptome and protein-refolding assay supported our hypothesis that indole might interfere with protein folding. It is worth noting that another hypothesis, fully consistent with the array data and with the experiments on MDH refolding, would be that indole directly inhibits the action of GroEL and GroES.

Figure 5.

Indole inhibits protein folding. Detailed procedures for the protein-folding assay were described in ‘'Materials and methods'’. Denatured mMDH was refolded in ATP and GroEL/GroES 1 : 1 mixture at 37 °C. One millimolar indole was added during this refolding process. The activity of mMDH was assayed at 25 °C, and the oxidation of NADH by mMDH was monitored at 340 nm. (a) Time-dependent protein folding. (b) The activity of refolded mMDH after 60 min of refolding, expressed as a percentage of native MDH control activity.


Studies of the effect of indole on bacterial cells have been focused on E. coli, which produces indole. High temperature (50 °C), low pH, and the presence of the antibiotics affect indole production in E. coli (Han et al., 2011). Thus, indole production may be affected by the surrounding environment, and this will in turn affect bacterial communities and their life. The impact of exogenous indole on the physiological characteristics of non-indole-producing bacteria has been poorly explored. Indole production during mixed-culture growth between P. aeruginosa and E. coli prevented pyocyanin production and quorum sensing–regulated virulence factors in P. aeruginosa (Chu et al., 2012). Indole could be degraded by various oxygenases from bacteria, fungi, and plants (Lee & Lee, 2010) and quickly reduced in P. aeruginosa (Lee et al., 2009). Degradation and incorporation of indole into tryptophan biosynthesis may be the first mechanism to prevent stress caused by indole. Our microarray data also showed the highest expression of trpB genes in the presence of indole (Table S2). If non-indole-producing bacteria cannot degrade or use excess indole, defense mechanisms such as molecular chaperones and proteases are turned on. Our data demonstrate that indole increases biofilm formation and reduces cell size in P. putida. These results are inconsistent with previously well-known studies using E. coli strains. We speculate that the roles of indole may differ between Pseudomonas species and indole-producing bacteria. Among indole derivatives, indole-3-acetic acid (IAA) has been extensively studied in terms of a symbiotic relationship with soil bacteria and plants. When E. coli K-12 was treated with IAA, the majority of genes encoding cell envelope components and adaptation-related proteins were differentially expressed (Bianco et al., 2006ab). IAA-treated cells had enhanced biofilm formation due to an increased production of lipopolysaccharide (LPS) and exopolysaccharide. In addition, IAA induced higher levels of the heat-shock protein DnaK, as also seen in our data. In E. coli, sigma factor, σ32, encoded by the rpoH gene, positively regulates the induction of heat-shock proteins such as DnaK, DnaJ and GrpE, and GroEL/ES (Kobayashi et al., 2011). These heat-shock proteins have been extensively studied in other bacterial cells. Many heat-shock proteins, which were highly expressed at high temperatures, include hslU, hslV, htpG, grpE, dnaK, ibpA, clpB, lon, and hflK in E. coli (Richmond et al., 1999); htpG, grpE, dnaK, groEL, and groES in Bacillus subtilis (Helmann et al., 2001); dnaK, groEL, groES, clpB, and lon in Mycoplasma pneumonia (Weiner et al., 2003); and ibpA, groEL, and groES in Yersinia pestis (Motin et al., 2004). Many of these heat-shock proteins were highly expressed in our microarray analysis with indole (Table. S2) and are reported to be expressed under various environmental stresses, such as nutrient starvation, exposure to pollutants, and changes in pH or osmolarity (Koide et al., 2006). In case of P. putida KT2440, toluene or o-xylene treatment induced those genes, such as groES, groEL, lon-1, lon-2, ibpA, htpG, danK, grpE, hslV, and hslU (Domínguez-Cuevas et al., 2006).

Although indole has been reported to cause oxidative stress (Vega et al., 2012), our microarray data did not show any change in oxidative stress–related genes. To investigate the degree of cellular oxidative stress in the presence of indole, superoxide production was measured using the nitroblue tetrazolium (NBT) assay. Exponentially growing cells were treated with different concentrations of indole and ethanol for 30 min. Concentrations of 1–3 mM indole did not cause oxidative stress (data not shown). Therefore, indole probably stresses bacterial cells by disrupting the electron transport system or energy generation and protein folding, rather than by producing oxidative stress, as seen in our study.

In this study, we have shown that indole, at concentrations above a certain level that appears to be toxic to non-indole-producing Pseudomonas strains, altered the expression of many genes in three functional categories: (1) proteases; (2) molecular chaperones; and (3) TCA cycle enzymes. These gene products play important roles in conditions of indole-induced stress. The expression of these genes might cause phenotypic and physiological changes such as in cell morphology, biofilm formation, the NADH/NAD+ ratio, and ATP concentrations. Our data provide evidence that indole, which has been recently spotlighted as a beneficial signal molecule for E. coli, also exerts deleterious effects on non-indole-producing Pseudomonas strains.


This work was supported by a grant (2012-0005277) from the MEST/NRF program.