The Primary Physiological Roles of Autoinducer 2 in Escherichia coli Are Chemotaxis and Biofilm Formation

Autoinducer 2 (AI-2) is a ubiquitous metabolite but, instead of acting as a “universal signal,” relatively few phenotypes have been associated with it, and many scientists believe AI-2 is often a metabolic byproduct rather than a signal. Here, the aim is to present evidence that AI-2 influences both biofilm formation and motility (swarming and chemotaxis), using Escherichia coli as the model system, to establish AI-2 as a true signal with an important physiological role in this bacterium. In addition, AI-2 signaling is compared to the other primary signal of E. coli, indole, and it is shown that they have opposite effects on biofilm formation and virulence.


Introduction
Quorum sensing (QS) is the process by which bacteria communicate via secreted signals (autoinducers); once the concentration of the autoinducers reaches a threshold, the signal is detected, and gene expression is altered [1]. The roles of QS are diverse and include population density detection, virulence, biofilm formation, and the maintenance of the stress response [2]. Although inhibitors of QS (quorum-quenching compounds) are still promoted as a means to reduce virulence without promoting resistance [3], these compounds will indubitably and unfortunately fail. The main problem is that the inhibition of QS leads to pleiotropic effects that affect growth; hence, lab strains and clinical isolates rapidly evolve resistance to these compounds [4][5][6]. Clearly, it is imperative to have a better understanding of QS in order to be in a position to better control bacteria to prevent diseases, such as stomach cancer and ulcers caused by Helicobacter pylori and Lyme disease by Borrelia burgdorferi [7], and to utilize them for synthetic biology applications. Therefore, in this opinion piece, we probe the physiological role of AI-2 by focusing on the best-studied bacterium, Escherichia coli.

Autoinducer-2
Commensal E. coli has several QS pathways, including one system based on indole ( Figure 1) [8][9][10], which is produced by TnaA from tryptophan, and another system based on autoinducer 2 (AI-2) ( Figure 1) [11], which is produced by LuxS from Sribosylhomocysteine [12]. It appears AI-2 is used primarily for communication inside the gastrointestinal tract at 37 • C, while indole is used primarily at lower temperatures (30 • C and lower) when the bacterium is outside of its eucaryotic host [9]. Although E. coli can detect homoserine lactones through the autoinducer-1 sensor SdiA (a LuxR homolog), it lacks a homoserine lactone synthase to produce the homoserine lactone signal, so E. coli uses SdiA to eavesdrop on signals of other bacteria [13]. Moreover, there is an interaction between these systems in that SdiA has been shown to be important for indole signaling in E. coli [8]. tion by the AI-2 receptor LsrB [15]. In addition to LsrB in E. coli, LuxP (e.g., Vibrio harve and the dCACHE-domain proteins PctA/TlpQ (Pseudomonas aeruginosa) are receptors f AI-2 [15], so there are at least three forms of AI-2 receptors in different bacteria. Furthe more, upon import, AI-2 is phosphorylated by LsrK in E. coli, and phosphorylated A binds and inhibits the repressor LsrR, which leads to changes in gene expression primar at 37 °C [9].
The first report of AI-2 and biofilm formation was indirect and based on masking A 2 signaling in E. coli with the QS inhibitor (5Z)-4-bromo-5-(bromomethylene)-3-buty 2(5H)-furanone (henceforth furanone) from the alga Delisea pulchra; in this report, biofil formation was reduced by 60 µg/mL furanone [22]. Later reports of AI-2 influencing b film formation were based on luxS mutants rather than purified AI-2. For example, a lu mutation in Streptococcus gordonii influenced mixed-species biofilm formation with P phyromonas gingivalis [23], a luxS mutation had a small impact on the architecture Klebsiella pneumoniae (although there was no effect for a luxS mutant for intestinal colon zation and colonization on polystyrene) [24], and a luxS mutant increased biofilm fo mation in Helicobacter pylori [25]. Unfortunately, these early results related to AI-2 via lu Once produced by LuxS, the AI-2 precursor 4,5-dihydroxy-2,3-pentanedione is converted spontaneously into R-2-methyl-2,3,3,4-tetrahydroxytetrahydrofuran (R-THMF) in E. coli (Figure 1), and R-THMF is the active form of AI-2 [7]. Hydrophilic AI-2 is transported from the cell by the membrane protein TqsA [14]. Once a threshold concentration is reached in the late exponential phase, AI-2 is imported into E. coli through its recognition by the AI-2 receptor LsrB [15]. In addition to LsrB in E. coli, LuxP (e.g., Vibrio harveyi) and the dCACHE-domain proteins PctA/TlpQ (Pseudomonas aeruginosa) are receptors for AI-2 [15], so there are at least three forms of AI-2 receptors in different bacteria. Furthermore, upon import, AI-2 is phosphorylated by LsrK in E. coli, and phosphorylated AI-2 binds and inhibits the repressor LsrR, which leads to changes in gene expression primarily at 37 • C [9].
The first report of AI-2 and biofilm formation was indirect and based on masking AI-2 signaling in E. coli with the QS inhibitor (5Z)-4-bromo-5-(bromomethylene)-3-butyl-2(5H)-furanone (henceforth furanone) from the alga Delisea pulchra; in this report, biofilm formation was reduced by 60 µg/mL furanone [22]. Later reports of AI-2 influencing biofilm formation were based on luxS mutants rather than purified AI-2. For example, a luxS mutation in Streptococcus gordonii influenced mixed-species biofilm formation with Porphyromonas gingivalis [23], a luxS mutation had a small impact on the architecture of Klebsiella pneumoniae (although there was no effect for a luxS mutant for intestinal colonization and colonization on polystyrene) [24], and a luxS mutant increased biofilm formation in Helicobacter pylori [25]. Unfortunately, these early results related to AI-2 via luxS mutations do not provide compelling evidence due to pleiotropic changes resulting from the luxS mutations.
The first direct demonstration that AI-2 was responsible for influencing biofilm formation was the 4-to 24-fold increase in biofilm formation in microtiter plates for three E. coli strains upon the addition of 11 µM of purified AI-2 [11]. Moreover, AI-2 failed to stimulate biofilm formation for an lsrK AI-2 regulation mutant, and AI-2 stimulated biofilm formation five-fold in flow cells [11]. A decade later, the Sourjik group rediscovered that AI-2 increases E. coli biofilm formation and extended the original results to show AI-2 increases aggregation through the adhesin antigen 43 and curli [26]. They [26] also confirmed that the AI-2 Lsr uptake/processing pathway influences E. coli biofilm formation [27].
The first direct report of AI-2 as a chemoattractant for any species was the 2008 discovery that Escherichia coli O157:H7 (EHEC) is attracted to purified AI-2 [29]. For EHEC, AI-2 also increases both swimming motility and attachment to HeLa cells [29]. For non-pathogenic E. coli, microfluidic devices were used a year later to show AI-2 is an attractant [30]. Later, similar to their studies on biofilm formation, the Sourjik group confirmed that AI-2 attracts E. coli [26]. Furthermore, as with biofilms, indole signaling is opposite that of AI-2 since indole repels enterohemorrhagic EHEC [31], whereas AI-2 attracts EHEC [29] (Figure 1).
The mechanism by which AI-2 is detected in E. coli was determined to be the chemotactic receptor Tsr, which previously was known for its recognition of L-serine [32]; LsrB, the AI-2 receptor, was also shown to be necessary [32]. As with chemotaxis and biofilm formation, chemotaxis through Tsr was corroborated by the Sourjik group [26]. Furthermore, the Manson group also verified that AI-2 increases biofilm formation in E. coli and found that biofilm formation in this strain is enhanced by chemotaxis to AI-2 [33]. Therefore, AI-2 stimulates biofilm formation in E. coli by increasing aggregation and chemotaxis (Figure 1).

AI-2 and Virulence
The two main E. coli signals influence pathogens in an opposite manner-indole decreases EHEC chemotaxis, motility, biofilm formation, and adherence to epithelial cells at the physiologically relevant concentration of primarily 0.5 mM [31]; these results that indole decreases EHEC virulence were largely confirmed 12 years later by the Sperandio group [34,35] (Figure 1). Indole from E. coli also reduces the virulence of P. aeruginosa by masking its QS [36], prevents P. aeruginosa from resuscitating [37] from the dormant persister state [38], and tightens the epithelial cell junctions of the human host [39]. Indole and its derivatives also kill persister cells [40,41]. In contrast, AI-2 at 100 µM to 500 µM increases EHEC chemotaxis, motility, and adherence to epithelial cells and induces biofilm-related genes [29]. Moreover, AI-2 induces the expression of 23 genes of the locus of enterocyte effacement of EHEC [29]. Hence, in pathogenic E. coli, indole reduces pathogenicity, while AI-2 increases it.

Perspectives
The discovery that the E. coli AI-2 signal secreted by cells attracts other E. coli cells and leads to increased biofilm formation indicates that E. coli cells actively seek other E. coli cells to form communities [42]. Hence, it illustrates how bacteria can seek kin to increase their fitness, i.e., cells seek others to build communities (biofilms) to protect themselves from myriad stresses [43] and to increase their pathogenicity.
The chemoattractant property of AI-2 has also led to several synthetic biology applications. For example, biological nanofactories have been devised that detect and bind cancer cells and then produce AI-2 at the surface of the cancer cells, which attracts E. coli homing cells that internalize the synthesized AI-2 and then produce a biomarker or potentially an anti-cancer compound from an AI-2-induced promoter [44]. In this way, healthy cells could be discriminated from diseased ones. Therefore, the better understanding of the roles AI-2 and indole play in E. coli physiology has had a significant impact, both in our understanding of how communities are formed and in synthetic biology. Hence, AI-2 and indole are true and important signals in E. coli.