Bidirectional Cell-Cell Communication via Indole and Cyclo(Pro-Tyr) Modulates Interspecies Biofilm Formation

ABSTRACT

The extracellular signaling molecule indole plays a pivotal role in biofilm formation by the enteric gammaproteobacterium Escherichia coli; this process is particularly correlated with the extracellular indole concentration. Using the indole-biodegrading betaproteobacterium Burkholderia unamae, we examined the mechanism by which these two bacteria modulate biofilm formation in an indole-dependent manner. We quantified the spatial organization of cocultured microbial communities at the micrometer scale through computational image analysis, ultimately identifying how bidirectional cell-to-cell communication modulated the physical relationships between them. Further analysis allowed us to determine the mechanism by which the B. unamae-derived signaling diketopiperazine cyclo(Pro-Tyr) considerably upregulated indole biosynthesis and enhanced E. coli biofilm formation. We also determined that the presence of unmetabolized indole enhanced the production of cyclo(Pro-Tyr). Thus, bidirectional cell-to-cell communication that occurred via interspecies signaling molecules modulated the formation of a mixed-species biofilm between indole-producing and indole-consuming species.

IMPORTANCE Indole is a relatively stable N-heterocyclic aromatic compound that is widely found in nature. To date, the correlations between indole-related bidirectional cell-to-cell communications and interspecies communal organization remain poorly understood. In this study, we used an experimental model, which consisted of indole-producing and indole-degrading bacteria, to evaluate how bidirectional cell-to-cell communication modulated interspecies biofilm formation via intrinsic and environmental cues. We identified a unique spatial patterning of indole-producing and indole-degrading bacteria within mixed-species biofilms. This spatial patterning was an active process mediated by bidirectional physicochemical interactions. Our findings represent an important step in gaining a more thorough understanding of the process of polymicrobial biofilm formation and advance the possibility of using indole-degrading bacteria to address biofilm-related health and industry issues.

INTRODUCTION

Bacteria live in complex and dense multispecies communities on biotic or abiotic surfaces. In fact, they often depend on each other for metabolic substrates (1–3). Considering the polymicrobial nature of biofilm, the interspecies cell-to-cell communication system via the release and recognition of small signaling molecules has been established among many microorganisms (4–6). Although several studies have examined this cell-to-cell communication network (7–9), the mechanism by which bidirectional interaction modulates mixed-species biofilm formation among Proteobacteria with different metabolisms of the signaling molecule remains unclear.

When Escherichia coli and other indole-producing gammaproteobacteria are exposed to indole, it induces acid tolerance, drug resistance, and biofilm formation (10, 11). These indole-mediated cell responses are common to many species as well as many strains within a species (10). Indole mediates biofilm formation in commensal and pathogenic E. coli strains (10, 12–14), while betaproteobacteria produce various diffusible signaling molecules, including diketopiperazines (DKPs), which are novel quorum-sensing (QS) signals for a broad range of Gram-negative bacteria. DKPs also act as interspecies cross-talking signals that allow communication with other phyla and classes of bacteria (15–17).

As the exogenous indole concentration modulates biofilm formation by E. coli (18), we hypothesized that manipulation of exogenous indole levels under various conditions could inhibit biofilm formation in a manner that would prove useful to health care providers and industrial producers. In this paper, we used an experimental model to test whether bidirectional communication between indole-producing and -degrading bacteria affected their physicochemical associations, including biofilm formation. We identified different phenotypes of cocultured biofilms mediated by bidirectional communication and cocultured indole-degrading Burkholderia unamae with biofilm-forming E. coli to study its responses to an extracellular indole. We also characterized an interspecies signaling molecule generated by B. unamae, whose production increased in the presence of indole and gallic acid. We note that gallic acid, which is widely distributed throughout the vegetable kingdom and commonly consumed by humans, plays an important role as an inhibitor of indole degradation via B. unamae. We also found that supplementation with a B. unamae-derived DKP enhanced E. coli biofilm formation by enhancing indole synthesis, while unmetabolized indole enhanced DKP production by B. unamae. This study provides a conceptual framework to explain how indole-degrading bacteria can modulate biofilm formation by degrading indole and releasing a signaling molecule.

RESULTS

Indole-dependent biofilm formation.

As an autoregulatory signaling molecule whose production depends on the cell population density, indole functions as a QS signal in stationary-phase cells (10). As shown in Fig. 1, in an E. coli monoculture biofilm, accumulation was closely associated with the increased production of extracellular indole between 4 and 12 h, while it reached its threshold at 12 h. This is consistent with the results obtained by Ryan and Dow, who determined that biofilm formation was largely regulated by the cell growth phase in response to cell-to-cell signaling molecules, including DKPs, autoinducer-2 (AI-2), and indole (19). Accordingly, we sought to determine whether extracellular indole affected interspecies biofilm formation (Fig. 1). We employed a coculture system consisting of indole-producing E. coli and indole-degrading B. unamae to analyze the interspecies communication between the producers and the consumers of the signaling molecules during mixed-species biofilm formation. In the B. unamae coculture, extracellular indole concentrations markedly decreased between 12 and 24 h (Fig. 1A and B). Notably, a large number of dispersed planktonic cells were present when indole degradation occurred in this coculture system. Between 12 and 24 h, B. unamae actively degraded indole to 3-hydroxyindoxyl (3-HI) (20), and the resulting relatively low level of indole suppressed biofilm formation. In sum, reduced biofilm formation resulted from a profound change in the cell density, which was regulated by the cell cycle-dependent association in the presence of indole derivatives (3, 21). After assessing the effects of exogenous indole (produced by E. coli in monoculture and degraded by B. unamae in coculture systems), planktonic cells, and the process of biofilm formation over time, we determined that 24 h marked a critical point in the process of interspecies biofilm formation.

FIG 1.

Changes in extracellular indole concentrations and biofilm formation in E. coli monoculture and coculture with B. unamae. The effects of the presence of extracellular indole on planktonic cell density and biofilm formability were evaluated in a time course manner (0, 4, 8, 12, and 24 h) for an E. coli monoculture and a coculture of E. coli with B. unamae. (A) Biofilm formed on the bottom of a petri dish stained with crystal violet. The biofilm was stained with crystal violet, washed with water, air dried, and then subjected to microscopic observation in phase-contrast mode. (B) Changes in extracellular indole, planktonic cells (optical density at 660 nm [OD₆₆₀]), and total biofilm (absorbance at 570 nm [A₅₇₀]) at 0, 4, 8, 12, and 24 h. E. coli biofilm accumulation was drastically increased between 4 and 12 h, while the mixed-species biofilm was disrupted after 12 h with decreased exogenous indole, and planktonic cells were dispersed more than those of the E. coli monoculture at 24 h. Values are averages from duplicates (n = 2). Biofilm formation of the E. coli monoculture (●) and the coculture with B. unamae (■) closely correlated with the extracellular indole concentration (r² = 0.97 and 0.99, respectively).

We also determined that biofilm formation in the E. coli monoculture and the B. unamae coculture was closely correlated with the presence of extracellular indole (r² = 0.97 and 0.99, respectively) (Fig. 1B, right E. coli and E. coli + B. unamae panels).

Effect of gallic acid on indole degradation and B. unamae swimming motility.

The effects of phenolic additives, including gallic acid, on indole degradation were evaluated. Previous chromatographic and spectrometric analyses indicated that B. unamae CK43B can hydroxylate indole to 3-hydroxyindoxyl and simultaneously degrade it to catechol via several intermediary compounds such as anthranilic acid (22) (Fig. 2A). In oxygen-saturated shaken cultures, B. unamae-induced indole degradation was accelerated by the presence of gallic acid as a typical pyrogallol-type phenolic compound but not in the case of catechol-type phenolic additives (see Fig. S1 in the supplemental material). In contrast, gallic acid inhibited indole degradation in a static culture (Fig. 2A). Chromatograms showed that in static cultures, indole was degraded to 3-hydroxyindoxyl, while B. unamae-induced indole degradation was inhibited by the presence of gallic acid (Fig. 2A and B). As B. unamae neither produced indole nor actively formed biofilms on its own (Fig. 2C), gallic acid must have played a crucial role in regulating the physicochemical properties of B. unamae (including indole metabolism). Gallic acid also enhanced B. unamae swimming motility, which showed a larger spread than E. coli, suggesting that gallic acid modulated flagellum-mediated chemotaxis. Indole in conjunction with gallic acid, however, inhibited the swimming motility of B. unamae (Fig. 2D). B. unamae chemotaxes toward indole and gallic acid might be important for mixed-species biofilm formation with indole-producing bacteria in gallic acid-rich environments.

FIG 2.

Effect of gallic acid on indole metabolism and B. unamae swimming motility. (A) Indole degradation pathway by B. unamae under exposure to phenolic compounds. The postulated pathway was modified from a previous report (22). (B) HPLC profiles of the indole metabolite 3-hydroxyindoxyl (3-HI) and unmetabolized indole in the culture of B. unamae in the presence or absence of gallic acid. A static culture of B. unamae was grown in 0.2 mM indole-containing LB medium at 28°C for 24 h. Indole and indole metabolites were extracted from each culture fluid with EtOAc. These profiles were obtained using an isocratic solvent system consisting of CH₃CN-H₂O (50:50) with 0.1% HCOOH. The peaks at retention times (tR) of 4.6, 6.3, and 14 min are of 3-HI, the internal standard (2,7-naphthalenediol), and indole, respectively. (C) Validation of indole production by B. unamae. An aliquot of 5 μl, which is equivalent to that of the culture fluid (2 ml), was injected into the HPLC system, and the peak intensity (absorbance) of indole (indicated by an arrowhead) was monitored at 280 nm. A 5-μl portion from 2 ml of a 10 μM indole solution contains 25 pmol (2.9 ng) indole; therefore, approximately 0.3 ng is the minimum detection level in this HPLC analytical system for indole. The background level of indole in LB broth was also analyzed (blank). (D) Swimming motilities of E. coli and B. unamae. For swimming motility assays, indole (0.2 mM), gallic acid (1.0 mM), indole (0.2 mM) with gallic acid (1.0 mM), or no chemical (control) was supplemented in LB broth solidified with 0.3% agar. Colony 1, point-inoculated E. coli K-12; colony 2, E. coli Top10, an indole-producing but less-biofilm-forming strain; colony 3, E. coli DH5α, an AI-2-negative, less-biofilm-forming strain; colony 4, B. unamae CK43B.

Indole-degrading B. unamae, in the presence of gallic acid, modulates biofilm formation by an E. coli strain.

E. coli and B. unamae cocultures with and without gallic acid were developed to assess the process of interspecies biofilm formation between indole-producing E. coli and indole-degrading B. unamae and compared to an E. coli monoculture. In cocultures with B. unamae, biofilm formation was suppressed by as much as 67% compared to an E. coli monoculture (Fig. 3A). However, in the presence of 1 mM gallic acid, which inhibited indole degradation by B. unamae, mixed-species biofilm formation was enhanced, similar to what was observed for the E. coli biofilm (Fig. 3A to C). The presence of supplemented indole affected the biofilm formation of the coculture of E. coli with B. unamae, although indole concentrations of >0.5 mM repressed E. coli biofilm formation (Fig. 3D). We further tested 0.2 mM indole to evaluate the physiologically relevant concentration at which indole alters interspecies biofilm formation. We determined that 0.2 mM indole (the threshold level of exogenous indole produced by E. coli in the biofilm model) did not affect the total biofilm formation of E. coli. However, supplementing 0.2 mM indole with gallic acid prompted the formation of an abundant biofilm, more so than in the coculture and the coculture with supplemented indole (P < 0.01) (Fig. 3E).

FIG 3.

Biofilm formation in the E. coli and B. unamae coculture in the presence of gallic acid. (A) Biofilm formation quantified by analyzing the A₅₇₀ of the coculture. (B) Changes in total extracellular indole levels in the coculture grown with or without gallic acid. The total amount of extracellular indole was quantitatively analyzed using HPLC. Statistical significance was calculated by one-way ANOVA with post hoc Duncan’s multiple-range test. Data represent the means ± SD (n = 3). Scripts a to c denote significant differences (**, P < 0.01; ***, P < 0.001). (C) HPLC profiles of exogenous indole in E. coli monoculture and coculture with B. unamae with and without gallic acid. These profiles were obtained using an isocratic solvent system of CH₃CN-H₂O (50:50) containing 0.1% HCOOH. The retention times (tR) of gallic acid (‡), the internal standard (int. std) (§) (2,7-naphthalenediol), and indole (arrows) were 3.9, 6.3, and 14.0 min, respectively. (D) Effect of indole additives (with different concentrations [0 to 2 mM]) on biofilm formation of an E. coli/B. unamae monoculture and their coculture. (E) Biofilm formation of the E. coli monoculture and coculture with B. unamae in the presence or absence of indole and gallic acid. Statistical significance was calculated by one-way ANOVA with post hoc Duncan’s multiple-range test. Data represent the means ± SD (n = 3). Scripts a to c denote significant differences (**, P < 0.01). IN, indole (0.2 mM); GA, gallic acid (1.0 mM).

Physical associations between indole-producing and -degrading bacteria.

We also sought to determine how the communal structure of the mixed culture was modulated by interspecies communication. Using fluorescence in situ hybridization (FISH) for species-specific labeling, we observed that E. coli and B. unamae formed segregated clusters within biofilms (Fig. 4A). These physical associations between E. coli and B. unamae are relevant to cell-to-cell interactions and play a crucial role in the spatiotemporal development of biofilms in polymicrobial communities (5, 23, 24). When B. unamae-induced indole degradation was inhibited by the addition of 1 mM gallic acid, the unmetabolized indole accelerated the formation of a mixed-species biofilm between E. coli and B. unamae (Fig. 3 and Fig. 4A). Using a polystyrene petri dish, coculture systems of both bacteria were dispersed when the E. coli biofilm was disturbed by indole-degrading B. unamae. Gallic acid, however, mediated the coaggregation of E. coli and B. unamae within the mixed-species biofilm, which possessed a biomass similar to that of the E. coli-monocultured biofilm (Fig. 3 and Fig. 4A). Indeed, a unique organization pattern of a microbial community that is comprised of indole-producing and -degrading bacteria occurs in a response to gallic acid (Fig. 4). Ultimately, the specific spatial arrangement by which different species appear within biofilms is influenced by the relative fitness benefits of cooperative and competitive species (25).

FIG 4.

Spatial organization of the coculture biofilm between E. coli and B. unamae in the presence of gallic acid. (A) Fluorescence images of cell aggregates obtained from E. coli cocultured with B. unamae with and without gallic acid. Bacterial cells cocultured with gallic acid or without gallic acid were observed under fluorescence with labeling by the B. unamae CK43B-specific probe BKH70 and the E. coli-specific FISH probe ENT186. (B) Spatial arrangement pattern of mixed-species biofilm determined by Daime using the linear dipole algorithm. Shaded areas depict means ± SD (n = 5). Polynomial regression analysis was conducted to estimate the spatial arrangement of variable biofilms. A polynomial regression model, including the coefficient of determination (r²) and fitting lines (depicted in red), is shown in each panel. (C) Proposed spatial arrangement between E. coli and B. unamae afforded by interspecies interactions. Dispersed cells resulted from the disturbed E. coli biofilm by indole-degrading B. unamae, while an intermixed interspecies community was present in the aggregated cluster. (D) Relative abundances of E. coli and B. unamae within the biofilm. The relative proportional occupancy was calculated from the ratio of each bacterial biovolume (voxel) (labeled with E. coli- and B. unamae-specific probes) to the reference biovolume (labeled with a universal probe [EUB338] for all bacteria) using Daime software (version 2.2.2).

In light of the distinctive spatial organization of dispersed and aggregated populations that we observed, we performed a computational quantitative analysis to determine the cell-cell interactions occurring via pair correlation between E. coli and B. unamae within the biofilm. We performed a computational analysis using the linear dipole algorithm developed by Daims et al. (26), to assess the overall spatial organization of E. coli and B. unamae in the biofilm community. We found that the two bacteria were located in proximity within cell clusters, i.e., “intermixed” communal organizations (pair correlation value of >1) (Fig. 4B). This spatial intermixing probably enables reciprocal benefits and cooperation between the two species, suggesting that cell-cell communication contributes to mixed-species biofilm formation in cases where the presence of unmetabolized indole enhances its formation (Fig. 4C). In contrast, a random distribution of the two populations (pair correlation value of around 1) was observed in the “disturbed” communities, suggesting that the two species remained physically separated (Fig. 4B). This pattern is associated with extracellular indole, where E. coli QS was disturbed by B. unamae-induced indole degradation (Fig. 4C). In sum, the spatial organization patterns that defined the relationship between the two species were regulated by gallic acid. However, there were no differences in the relative abundances of E. coli cocultured with B. unamae regardless of the presence or absence of gallic acid (Fig. 4D).

Effect of cyclo(Pro-Tyr) on interspecies biofilm formation.

The spatial organization within each biofilm indicated that the intermixed communal structures were mediated by gallic acid. Gallic acid inhibited indole degradation by B. unamae, and the unmetabolized indole, in turn, enhanced interspecies biofilm formation. We then tested whether the interspecies interaction was mediated bidirectionally. To determine whether any reciprocal cell-to-cell communication was occurring within the biofilm, the B. unamae conditioned (spent) medium was collected from a static culture and tested for its effect on the biofilm formation of E. coli. Interestingly, the extracellular indole concentration increased when E. coli was cultured in the spent medium (Fig. 5A). Biofilm formation by E. coli was also enhanced by exposure to spent medium (Fig. 5B).

FIG 5.

Effect of B. unamae-derived diketopiperazines on E. coli biofilm formation. (A) Extracellular indole concentration of E. coli K-12 in LB medium supplemented with the spent medium of B. unamae CK43B. Statistical significance was calculated using Student’s t test. Data represent the means ± SD (n = 3). (B) The turbidity of the culture medium corresponding to planktonic cells and the total biofilm mass were evaluated by analyzing the optical density at 660 nm (OD₆₆₀) (left) and the absorbance at 570 nm (A₅₇₀) of crystal violet-stained E. coli biopolymers (right). Statistical significance was calculated by Student’s t test for a pairwise comparison. *, P < 0.01 relative to plain LB medium; NS, not significant. (C) Enhancement of E. coli biofilm formation in a dose-response curve for cyclo(Pro-Tyr). Cyclo(Pro-Tyr) at a concentration range of 0 μM (vehicle [DMSO] control) to 1,000 μM was added to 2 ml LB broth in petri dishes. Bacterial cells were statically cultured in petri dishes at 28°C for 24 h. Total biofilm was evaluated by analyzing the A₅₇₀ of biopolymers stained with crystal violet and dissolved with 30% acetic acid. Data represent the means ± SD (n = 6). ***, P < 0.001 relative to plain LB medium (vehicle [DMSO] only). (D) Changes in sdiA, tnaA, and luxS of E. coli at the transcriptional level mediated by cell-to-cell signaling molecules derived from B. unamae. Bacterial cells were statically cultured in LB medium supplemented with no chemical additives (normal), noncultured LB medium eluents (control), spent medium of B. unamae (spent), and 50 μM cyclo(Pro-Tyr) (DKP) in petri dishes 28°C for 24 h. Statistical significance was calculated by one-way ANOVA with post hoc Duncan’s multiple-range test. Data represent the means ± SD (n = 3). Scripts a to c denote significant differences (**, P < 0.01).

Using a molecule elucidation procedure via column chromatography and spectrometric analyses (see Materials and Methods), a possible diffusible DKP-type signaling molecule, cyclo(Pro-Tyr), was identified in the spent medium. We therefore tested the effect of cyclo(Pro-Tyr) and other DKPs on biofilm formation by E. coli in a dose-dependent manner. Among the DKPs tested, cyclo(Pro-Tyr) was the most active in the biofilm bioassay; other DKPs were either less active or inactive (data not shown). A total of 50 μM synthetic cyclo(Pro-Tyr) enhanced the biofilm formation of E. coli (P < 0.001) in the same manner as a natural product (Fig. 5C).

We next sought to determine the threshold concentration of cyclo(Pro-Tyr) that would modulate the expression of various genes using quantitative real-time PCR (qRT-PCR). The genes selected for qRT-PCR are related to a transcription factor of the LuxR family (sdiA) and the biosyntheses of indole (tnaA) and AI-2 (luxS) in E. coli. LuxS is a downstream enzyme of carbon-sulfur lyase functioning in the biosynthetic pathway of AI-2 and is the species-nonspecific interspecies communication pathway (27), while tryptophanase (TnaA) is involved in the interspecies communication system as well as species-specific intraspecies reactions (10). In this manner, we evaluated luxS and tnaA gene expression together with sdiA gene expression to confirm whether B. unamae-derived signaling molecules modulate interspecies communication signaling of E. coli. Notably, spent media derived from B. unamae and cyclo(Pro-Tyr) upregulated tnaA but did not induce any dynamic changes in sdiA and luxS transcription (Fig. 5D), leading us to conclude that increased indole production caused by the upregulation of tnaA may result in enhanced interspecies biofilm formation.

Effect of indole and gallic acid on the production of cyclo(Pro-Tyr).

DKPs have been identified as putative QS signals that activate or antagonize N-acyl-homoserine lactone biosensors, including LuxR and its homologue TraR or LasR, in Burkholderia spp. and some other proteobacteria (15, 16). Accordingly, we evaluated the roles of indole and gallic acid in the regulation of the B. unamae QS system. A B. unamae autoinducer synthase gene (unaI) was upregulated in the presence of 0.2 mM indole and 0.2 mM indole plus 1 mM gallic acid. The transcription of the B. unamae PAS/PAC sensor transduction histidine kinase (unaR), however, was repressed by supplementation with indole plus gallic acid (Fig. 6A). This kinase belongs to the LuxR-type superfamily that responds to redox signals (28). A thin-layer chromatogram showed that indole plus gallic acid enhanced cyclo(Pro-Tyr) production, allowing it to play a greater role in modulating the physical association between E. coli and B. unamae. Due to an antagonistic effect of DKPs against LuxR-type biosensors (15), increased production of cyclo(Pro-Tyr) in the presence of indole plus gallic acid probably modulated unaR gene expression of B. unamae.

FIG 6.

Transcriptional changes in B. unamae CK43B and enhanced production of diketopiperazines (DKPs) in the presence of indole and gallic acid. (A) Bacterial cells were statically cultured in LB medium supplemented with no chemical additives, indole (0.2 mM), gallic acid (1.0 mM), and indole (0.2 mM) plus gallic acid (1.0 mM) in petri dishes at 28°C for 24 h. arhd, aromatic-ring-hydroxylating dioxygenase; unaI, autoinducer synthase; unaR, luxR homologue for the biosensor. Statistical significance was calculated by one-way ANOVA with post hoc Duncan’s multiple-range test. Data represent the means ± SD (n = 4). Scripts a to c denote significant differences (for each gene) (*, P < 0.05; **, P < 0.01). (B) Semiquantitative analysis of DKPs using thin-layer chromatography (TLC). The production of DKPs in B. unamae CK43B was promoted by 0.2 mM indole plus 1.0 mM gallic acid. A total of 200 ml of LB medium supplemented with 0.2 mM indole, 1.0 mM gallic acid, and 0.2 mM indole plus 1.0 mM gallic acid was statically cultured at 28°C for 24 h. Culture fluid was extracted twice with an equal volume (200 ml) of acidified EtOAc, and the presence of DKPs was confirmed using TLC. TLC was developed in hexane-EtOAc (1:2) and sprayed with a vanillin-sulfuric acid reagent.

DISCUSSION

Cell-to-cell communication is mediated by specific extracellular signaling molecules that tend to accumulate in the vicinity of their origin. Among known signaling molecules, indole is known to serve as an interspecies signal, a quality that has engendered a great deal of interest among researchers (10). Indole signaling occurs across kingdom boundaries. For example, bacterium-derived indole represses the growth of Penicillium strains and attenuates the virulence of Candida albicans despite the fact that these fungi are incapable of producing indole (29). Additionally, many non-indole-producing bacteria and fungi possess oxygenases that are capable of degrading indole and producing indole derivatives, including hydroxyindoles. These indole derivatives also assume various biological roles in interspecies/interkingdom communities (10, 30). Finally, indole is able to attenuate inflammatory reactions by preventing the attachment of pathogenic E. coli to intestinal epithelial cells (31).

Although the mechanism of indole biosynthesis in E. coli is well known, the real biological functions of indole in drug resistance, virulence, and biofilm formation in multispecies niches have only recently begun to be revealed (10, 30). Indole signaling is dynamic in many microorganisms, prompting us to investigate the effect of exogenous indole on an indole-producing E. coli monoculture and a coculture with indole-degrading B. unamae. We determined that mixed-species biofilm formation between E. coli and B. unamae is responsive to extracellular indole and that indole facilitates aggregation and biofilm formation between indole-producing gammaproteobacteria and indole-degrading betaproteobacteria through the exchange of extracellular signaling cues, particularly in the presence of gallic acid. Our findings suggest that the colonization of interspecies biofilms that consist of indole-producing and indole-metabolizing bacteria is mediated by bidirectional cell-to-cell communication.

We recognize the complexity of this interspecies association. Although we focused on the bidirectional interaction between indole-producing E. coli and indole-degrading B. unamae, the interactions between these bacteria are mutually beneficial when indole degradation is inhibited. B. unamae is unable to form biofilm on its own (20), although unmetabolized indole mediated B. unamae biofilm formation and B. unamae-derived DKP enhanced E. coli biofilm formation. This reciprocal exchange of signaling molecules (indole and DKP) mediated interspecies biofilm formation. Thus, metabolic events linked to indole can alter bacterial behavior via the sensing of other exogenous molecules (e.g., intrinsic QS molecules or other microbial products) and extrinsic molecules (providing environmental cues) in a multispecies community (10, 30, 32).

Indole derivatives are widely distributed in the human body, especially in the intestine, although their role in interacting with the gut microbiota remains unclear (29, 33, 34). It is certainly possible that indole-degrading microbes modulate the microbial composition of the gut. How do indole metabolites interact with exogenous phenolic compounds to affect physiological processes relevant to human health? It will be interesting to determine, through future research, the impact of indole-metabolizing bacteria on the colonization of gut microbiota and to identify the specific metabolites that modulate cell-to-cell interactions. The ingestion of indole-degrading microbes along with phenolic-rich food ingredients may modulate inflammatory and metabolic disorders associated with gut microbiota homeostasis (35, 36).

The cometabolic effect of pyrogallol-type phenolic compounds on N-heterocyclic aromatic compound (NHAC)-degrading microorganisms may enhance phytobioremediation efforts (22). In this context, some betaproteobacteria (including other saprophytic bacteria) may have their biodegradation modulated by polyphenols. Thus, phenolic compounds may play important roles in the rhizosphere or for wastewater treatment or the phytobioremediation of soil or water polluted with NHACs (37).

In summary, this study provides new insights into the cell-to-cell communication between specific producers and consumers (degraders) of signaling molecules modulated by intrinsic and extrinsic compounds. Using an experimental biofilm model and chromatographic analyses, we demonstrate that the biofilm formation of indole-producing E. coli is modulated by indole-degrading B. unamae, mediating a dispersed cell arrangement. In contrast, when indole degradation by B. unamae was inhibited by gallic acid, E. coli and B. unamae coaggregated within the interspecies biofilm. Further chemical analysis showed that cyclo(Pro-Tyr) from B. unamae spent medium enhanced E. coli biofilm formation. While further studies using a comprehensive transcriptome and time-lapse imaging modality will allow an improved understanding of the precise mechanisms of bidirectional communication at different biofilm development stages, the existing data are sufficient to show that the systems of cell-to-cell communication are feasible targets for the manipulation of interspecies interactions and modulation of biofilm development.

MATERIALS AND METHODS

Bacteria used.

B. unamae strain CK43B has been characterized as an active indole-degrading bacterium (20, 22). B. unamae CK43B, which is identifiable by the presence of a 16S rRNA gene sequence (GenBank accession number AB714631), was used in the coculturing experiment, along with indole-producing and biofilm-forming E. coli K-12 IFO 3301 (12).

Bacterial culture and biofilm formation assay.

All bacteria were cultured in 2 ml of Luria-Bertani (LB) broth in polystyrene petri dishes (diameter [ϕ] of 35 by 10 mm; BD Biosciences, San Jose, CA, USA) at 28°C as a static culture. Bacterial growth was assessed by monitoring the optical density at 660 nm (OD₆₆₀) values of triplicate samples using a microplate reader (Sunrise; Tecan, Männedorf, Switzerland). In the coculture assay, E. coli and B. unamae were simultaneously inoculated and then incubated under conditions similar to those for the monoculture. A biofilm formation assay was performed to evaluate the biofilm-inducing signals using the crystal violet staining method as described previously (20).

Cell response to extracellular indole in coculture growth.

To determine the effect of extracellular indole on biofilm formation, indole production and total biofilm mass of E. coli K-12 and an E. coli coculture with B. unamae at 28°C were evaluated at 0, 4, 8, 12, and 24 h. Moreover, the cell responses of E. coli K-12, B. unamae CK43B, or the mixed culture to exogenous indole (0.2 mM; effective concentration to induce biofilm), gallic acid (1.0 mM; sufficient concentration to inhibit indole degradation without growth-inhibitory effects on both bacteria), and indole plus gallic acid (0.2 plus 1.0 mM) were examined after 24 h of incubation at 28°C.

Quantification of indole in culture fluid.

The culture was centrifuged at 15,650 × g for 10 min. The resulting supernatant was adjusted at pH 4.0 with 1 M HCl and extracted twice with an equal volume of ethyl acetate (EtOAc). The resulting organic layer was dried over anhydrous Na₂SO₄ and then concentrated and redissolved in 2 ml of acetonitrile for quantitative analysis of extracellular indole using high-performance liquid chromatography (HPLC). The HPLC system consisted of a Hitachi (Tokyo, Japan) L-6320 intelligent pump, a Hitachi L-4250H UV-visible (UV-Vis) detector, and a Hitachi D-2500 Chromato-integrator. An Inertsil Prep-Ods 10-μm column (6.0-mm internal diameter [i.d.] by 250 mm; GL Science Inc., Torrance, CA, USA) was used as a reverse-phase column, with acetonitrile-H₂O (50:50) containing 0.1% (vol/vol) formic acid as a mobile phase (20). As an internal standard, 25 μM 2,7-naphthalenediol was used for quantifying indole at a UV absorbance of 280 nm with a linear correlation (r² = 0.99; y = 6.046x + 0.004 [where y is indole {millimolar} and x is the peak intensity ratio {indole/internal standard}]).

Indole degradation by B. unamae with supplementation with a phenolic additive.

B. unamae’s indole-degrading ability in the presence of other phenolic compounds was investigated by adding up to 3 mM pyrogallol (gallic acid and pyrogallol)- and catechol (4-methylcatechol and protocatechuic acid)-type phenolic compounds to a total volume of 200 ml of modified Winogradsky’s liquid medium containing indole (200 mg/liter). B. unamae strain CK43B was grown using this medium for 48 h in static and shaking cultures at 28°C. Metabolites in the cultured medium were extracted with EtOAc or a reverse-phase Sep-Pak column (Mega BondElut C₁₈; Varian Inc., Lake Forest, CA) and detected by TLC (CHCl₃-MeOH-H₂O, 65:25:4). Quantitative analysis of indole degradation using HPLC was performed as described previously (22).

Agar plate assay for assessment of swimming motility and competitiveness.

For swimming motility assays, indole (0.2 mM), gallic acid (1.0 mM), indole (0.2 mM) with gallic acid (1.0 mM), or no chemicals (control) were added to LB broth solidified with 0.3% agar. To prepare the inocula of B. unamae CK43B, E. coli K-12 IFO 3301, E. coli Top10, and E. coli DH5α, each strain was shake cultured in 10 ml of LB broth overnight at 100 rpm at 28°C. An aliquot 100-fold diluted by LB broth (OD₆₆₀ of ca. 0.2; 5 × 10⁵ CFU/ml) was spotted onto the plate and then incubated at 25°C for 48 h.

Fluorescence imaging of E. coli cocultured with B. unamae in LB broth medium.

E. coli and B. unamae were cocultured to observe whether they accepted or competed against each other. Aliquots of 10 μl each of diluted bacterial colonies of B. unamae (OD₆₆₀ of 0.1 to 0.2) and E. coli (OD₆₆₀ of 0.2) were inoculated together in 2 ml of LB broth medium with or without 1 mM gallic acid as the indole degradation inhibitor under anaerobic culture conditions. The coinoculated medium was statically cultured in a polystyrene petri dish (35 by 10 mm) at 28°C for 24 h. For fluorescence in situ hybridization (FISH) of the biofilm formed by the B. unamae and E. coli coculture, microbial cell fixation, in situ hybridization, and imaging were performed according to procedures described previously (20).

Computational quantitative image analysis.

To examine the spatial organization of the interspecies community across each biofilm’s architecture, we used Daime software (version 2.2.2) (https://dome.csb.univie.ac.at/daime/download-daime). Briefly, we selected spatial arrangement analysis from the available software dialogues and characterized segmented images of mixed-species communities by pair correlation using the linear dipole algorithm (26). The reference population was E. coli, while B. unamae was the analyzed population. Pair correlation values of >1 indicate coaggregation, values of <1 indicate avoidance, and values of ∼1 indicate a random distribution of two populations over a certain distance. The proportional occupancy (relative abundance of each bacterium) was described as the ratio of the bacterial biovolume (voxel) (labeled with E. coli- and B. unamae-specific probes) to the reference biovolume (labeled with a universal probe for all bacteria).

Preparation of spent medium.

To confirm the effect of B. unamae CK43B-produced signaling molecules on E. coli biofilm formation, B. unamae cells were cultured in LB medium (35- by 10-mm-diameter petri dish, total of 40 ml [2 ml × 20]) at 28°C for 24 h. The culture fluid obtained after centrifugation at 5,650 × g was subjected to solid-phase extraction using a 50-ml reverse-phase Sep-Pak column (Mega BondElut C₁₈; Varian Inc., Lake Forest, CA, USA) with no pH adjustment. Chemical molecules in the spent medium were recovered from the column cartridge by elution with 200 ml of methanol (MeOH). The concentrated MeOH eluents were redissolved in 2 ml of dimethyl sulfoxide (DMSO). E. coli was inoculated into 2 ml of LB medium containing an additional 20-μl aliquot of the concentrated spent medium equivalent to 1 ml of culture fluid, which was then subjected to static culture at 28°C for 24 h.

Identification of signaling molecules in the B. unamae spent medium.

The culture fluid of the B. unamae static-culture medium (200 ml × 5, cultured at 28°C for 48 h) obtained by centrifugation was acidified (0.1 ml acetic acid/ml) and extracted twice with an equal volume of EtOAc (38). Five microliters of the volumetric solution was charged by normal-phase thin-layer chromatography (TLC) and developed in CH₃Cl-MeOH (9:1). Chemical spots were observed under UV at 254 nm and purified using silica gel column chromatography (30-g spherical silica gel [60 N; Kanto Chemical, Tokyo, Japan] in a 2.0- by 20-cm open glass column) and preparative TLC. The extract (187.4 mg) was eluted by stepwise increases in CHCl₃-MeOH (10 to 100%). Compounds 1 to 3 (total of 77.3 mg) were obtained from fraction 3 (20%), and the concentrated fraction was further purified by preparative TLC. The chemical structures of compounds 1 to 3 were separately elucidated by ¹H and ¹³C nuclear magnetic resonance (NMR) (EX-270; JEOL, Tokyo, Japan) at 270 and 75 MHz and using field desorption (FD)- and high-resolution (HR)-FD-mass spectrometry (MS) or electron ionization (EI)-MS (JMS-T100GCv or JMS-SX102A; JEOL). Cyclo(Ala-Leu) (compound 1)—EI-MS (relative intensity [percent]): m/z 182 ([M]⁺, 20), 167 (5), 140 (2), 126 (100), 97 (52); FD-MS (relative intensity [percent]): m/z 183 ([M + H]⁺, 100), 182 ([M]⁺, 88), 126 (13); HR-FD-MS: m/z 182.1059 (C₉H₁₄N₂O₂, calculated for 182.1055). Cyclo(Pro-Leu) (compound 2)—FD-MS (relative intensity [percent]): m/z 211 ([M + H]⁺, 47), 210 ([M]⁺, 100), 182 (6), 168 (12), 154 (20); HR-FD-MS: m/z 210.1360 (C₁₁H₁₈N₂O₂, calculated for 210.1368). Cyclo(Pro-Tyr) (compound 3)—EI-MS (relative intensity [percent]): m/z 260 ([M]⁺, 11), 154 (100), 152 (2), 107 (42), 70 (15), 41 (2); FD-MS (relative intensity [percent]): m/z 283 ([M + Na]⁺, 5), 261 ([M + H]⁺, 27), 260 ([M]⁺, 100); HR-FD-MS: m/z 260.1168 (C₁₄H₁₆N₂O₃, calculated for 260.1161). ¹H NMR (δ_H in CDCl₃): 8.32 (br s, 1H), 6.99 (d, J = 8.2 Hz, 1H), 6.73 (d, J = 8.2 Hz, 1H), 5.83 (br s, 1H), 4.15 (br d, J = 6.9 Hz, 1H), 4.04 (t-like, J = 7.3 Hz, 1H), 3.61 to 3.49 (m, 2H), 3.40 (dd, J = 13.2, 9.2 Hz, 1H), 2.68 (dd, J = 14.5, 4.6 Hz, 1H), 2.33 to 2.21 (m, 1H), 2.06 to 1.86 (m, 3H); ¹³C NMR (δ_C in CDCl₃): 169.58, 165.17, 155.65, 130.31, 126.99, 116.10, 59.12, 56.24, 45.45, 35.93, 28.23, 22.46. Based on the results of the ¹H and ¹³C NMR analyses, a structure of cyclo(l-Pro-l-Tyr) was assigned, a result that is consistent with that reported in a previous study (15).

Quantitative real-time PCR analysis.

The expression of indole production- and signal transduction-related E. coli genes (sdiA, tnaA, and luxS) in response to the presence of B. unamae spent medium and cyclo(Pro-Tyr) (Bachem, Bubendorf, Switzerland) was analyzed using quantitative real-time PCR (qRT-PCR). B. unamae genes (unaI, unaR, and arhd) related to QS and indole degradation were also evaluated to determine cell responses to indole (0.2 mM), gallic acid (1.0 mM), and indole plus gallic acid (0.2 plus 1.0 mM). To identify the target gene for qRT-PCR analysis, genomic DNA of B. unamae was extracted as described previously (22). B. unamae and E. coli total RNA was extracted from 2-ml cultures in LB medium (OD₆₆₀ of ∼0.6; 10⁸ bacterial cells) using a NucleoSpin RNA II kit (Macherey-Nagel GmbH & Co., Düren, Germany). The quantity of RNA in each sample was determined by measuring the absorbance at 260 nm (A₂₆₀) with a spectrophotometer (GeneQuant Pro; Amersham Pharmacia Biotech, Uppsala, Sweden). RNA integrity was assessed by measuring the ratio of the absorbance at 260 nm to that at 280 nm (values obtained of between 1.8 and 1.9). cDNA was synthesized from each RNA sample using a PrimeScript 1st-strand cDNA synthesis kit (TaKaRa) containing 250 ng total RNA and 5 μM random hexamers in a final volume of 20 μl. Total RNA was reverse transcribed with the PrimeScript 1st-strand cDNA synthesis kit (TaKaRa) using random hexamers. Next, qRT-PCR was performed for each sample using SYBR Premix ExTaq II (TaKaRa), 50 ng of the cDNA mixture, and 0.4 μM gene-specific primer pair for each, with a final volume of 25 μl. Thermal cycling was performed using a Dice real-time thermal cycler system (TaKaRa) under the following reaction conditions: 95°C for 1 min for initial denaturation and 30 cycles of 95°C for 10 s, followed by 55°C for 30 s and 72°C for 1 min. All transcriptional changes were relative to those of an internal control, i.e., RNA polymerase sigma factor (rpoD) for B. unamae and recombinase A (recA) for E. coli, which were previously used as internal control genes for qRT-PCR analysis (2^−ΔΔCT [C_T is the crossing point]) (39–41). The primer sequences used in this study are shown in Table 1.

TABLE 1.

qRT-PCR primers used in this study

Primer	Sequence	Description (reference)
E. coli primers
recA-F	5′-TGCGTTTATCGATGCTGAACA-3′	Recombinase A (39)
recA-R	5′-GAGCACAGCAGGTTGTCGATATC-3′
tnaA-F	5′-TGGATCGCAGCAAAATGGT-3′	Tryptophanase (42)
tnaA-R	5′-GCACGGTACAGCCGTTGAT-3′
sdiA-F	5′-TGCAACGGGAAAAGGACAA-3′	Transcription factor of the LuxR family (42)
sdiA-R	5′-GCGGTGTCACTCAGTATTTAATGC-3′
luxS-F	5′-CATACCCTGGAGCACCTGTT-3′	Autoinducer-2 synthase (43)
luxS-R	5′-TGATCCTGCACTTTCAGCAC-3′
B. unamae CK43B primers
rpoD-F	5′-CCAAGTGGCCCGTTTTCCTTTG-3′	RNA polymerase sigma factor (this study)
rpoD-R	5′-ATTCTTATTGACGGCAGCGGA-3′
arhd-F	5′-TTCAAAGATGTTCGCGTGCC-3′	Aromatic-ring-hydroxylating dioxygenase (this study)
arhd-R	5′-AGATCGTGCCGAAGGAGAAC-3′
unaI-F	5′-AGCCGCAAATTTCCCCGTTCG-3′	Autoinducer synthase (this study)
unaI-R	5′-GTACCGTTATTCGCGTGTTCGT-3′
unaR-F	5′-CGCTATCAAAGCGACTACGC-3′	luxR homologue for biosensor (this study)
unaR-R	5′-TAGATGCCGGACAACAGACC-3′

Statistical analysis.

Values are expressed as the means ± standard deviations (SD). Multiple comparisons were carried out using Student’s t test and one-way analysis of variance (ANOVA) with post hoc Duncan’s multiple-range test using SPSS 18.0 software (IBM, Armonk, NY, USA). Differences were considered statistically significant when the P value was less than 0.05. GraphPad Prism 9 software (GraphPad, San Diego, CA, USA) was used to perform the polynomial regression analysis (y = pair correlation; x = distance [micrometers]) in Fig. 4B.