Keywords

1 Dispersion as a Flight Response

Biofilms are defined as a structured community of bacterial cells enclosed in a self-produced polymeric matrix and adherent to inert or living surfaces [1]. The ability to form a biofilm is a common trait of a diverse array of microbes, including lower order eukaryotes, with biofilms being the predominant mode of bacterial growth in nature [2]. The sessile lifestyle affords bacteria multiple protective advantages, allowing bacteria to remain within a favorable environmental niche or host. Compared to free-swimming bacteria, biofilms are better adapted to withstand nutrient deprivation, pH changes, oxygen radicals, biocides, and antimicrobial agents [3]. However, being in a biofilm is not only advantageous. As a biofilm grows in size, some cells will become increasingly separated from the bulk liquid interface and essential sources of energy or nutrients. Accumulation of waste products and toxins in the interior of biofilms pose additional challenges. Being trapped deep within a biofilm can, therefore, threaten cell survival. Thus, biofilm cells have evolved mechanisms which enable escaping the sessile mode of growth as a means of self-preservation, by liberating themselves from matrix-encased biofilms, and reverting back to the planktonic mode of growth. The transition to the planktonic mode of growth is referred to as dispersion [4,5,6,7]. Moreover, dispersion is considered a mechanism by which enables dissemination to new locales for colonization [4, 5, 8]. First described by Davies in 1999 [9], dispersion is apparent by single cells actively escaping from the biofilm, leaving behind eroded biofilms and microcolonies having central voids (Fig. 31.1) [4,5,6,7, 10,11,12,13,14,15]. Dispersion rarely involves the entire biofilm, with no more than 80% of the biofilm biomass being removed upon induction of dispersion [10, 13, 16,17,18]. Instead, selected microcolonies or areas within a biofilm will undergo a dispersion event at any particular time, in a manner often dependent on microcolony diameter [8].

Fig. 31.1
figure 1

Appearance of the remaining biofilm architecture post induction of dispersion. Representative drawings and confocal images of P. aeruginosa biofilms prior to and post dispersion are shown

Full size image

2 Dispersion Induces a Switch in the Mode of Growth

Considering that dispersed cells escape from the biofilm as single cells suggests that dispersion is a way for bacteria to transition from the surface-associated to the planktonic mode of growth. Transition to and from the surface have been linked to the modulation of the intracellular signaling molecule bis-(3′-5′)-cyclic dimeric guanosine monophosphate (cyclic di-GMP). More specifically, biofilm formation or the sessile lifestyle have been associated with high levels of cyclic di-GMP, with elevated cyclic di-GMP levels, in turn, resulting in increased production of biofilm matrix components, adhesiveness/autoaggregation, and antimicrobial tolerance, but repressed motility [19,20,21,22,23,24,25,26]. In contrast, low cyclic di-GMP levels have been associated with a motile or planktonic existence. Levels of cyclic di-GMP are enzymatically modulated by diguanylate cyclases (DGCs), proteins containing a GGDEF domain, and phosphodiesterases (PDEs) harboring either an EAL or HD-GYP domain. In agreement with dispersion coinciding with single cells actively escaping from the biofilm and transitioning toward a motile mode of growth, dispersed cells are motile, characterized by increased expression of fliC (encoding flagellin type B), and cyclic di-GMP levels comparable to or lower than those found in planktonic cells [12, 13, 27, 28].

3 Translation of Dispersion Cue Perception into the Modulation of the Intracellular Cyclic di-GMP Pool

How is dispersion induced and how does dispersion result in a reduction in cyclic di-GMP levels? Dispersion occurs in response to a number of cues and signals including fatty acid signaling molecule belonging to the family of diffusible signaling factors (DSF), pH, ammonium chloride, heavy metals, and nitric oxide (NO), host factors such as bile salts, and availability of oxygen, iron, amino acids, and carbon sources (Table 31.1). The mechanism of dispersion cue perception has been determined in more detail for a select number of dispersion agents including fatty acid signals, carbon sources, and NO. In each case, dispersion cue sensing was found to require at minimum a membrane-bound sensory protein, and a protein involved in the modulation of cyclic di-GMP levels such as a phosphodiesterase or in the case of NO sensing, a bifunctional enzyme harboring GGDEF-EAL domains (Fig. 31.2). The components form a signal transduction cascade that upon dispersion cue perception likely initiate a phosphorelay to the cyclic di-GMP modulating enzyme, resulting in the activation of the phosphodiesterase activity, and thus, the reduction of cellular levels of cyclic di-GMP (Fig. 31.2). Our current understanding of selected dispersion signaling pathways resulting in the alteration in the cellular level of cyclic di-GMP is discussed in detail below.

Table 31.1 Cues and signals linked to biofilm dispersion
Full size table
Fig. 31.2
figure 2

Dispersion cue perception and relay resulting in the modulation of cyclic di-GMP levels. (a) Model of the DSF/Rpf signaling cascades in X. campesitris. RpfF synthesizes DSF. RpfC is a sensor kinase and the response regulator RpfG exhibits phosphodiesterase activity. The signaling cascade is shown at low (left) and high (right) DSF concentrations. (b) Model of the signaling cascade involved in sensing nutrient dispersion cues in P. aeruginosa. Conditions at low (left) and high (right) concentrations of nutrient dispersion cues are shown. (c) Nitric oxide-induced signaling events mediated by the heme-based sensor domain H-NOX. Left, nitric oxide-induced signaling events in Legionella pneumophila or Shewanella woodyi. NO sensing via the H-NOX protein results in reduced cyclic di-GMP levels by stimulating the PDE activity (EAL) domain. Right, in Vibrio cholerae or S. oneidensis, interaction of the NO-bound H-NOX domain with a coupled histidine kinase (H-NOK) controls the phosphorylation activity of the kinase. Specific phosphorylation events lead to a decrease in cyclic di-GMP levels, either by stimulating PDE or by controlling the transcriptional response through a dedicated transcription regulator (HTH). (d) Possible mechanism of NO-induced biofilm dispersal via the MHYT domain-containing NbdA protein in P. aeruginosa. While BdlA, DipA, and RbdA have been shown to be required for NO-induced dispersion, it is unclear whether they form a signaling cascade with NbdA. Binding of NO is supposed to occur via copper ions (Cu) located in the MHYT domain. . Conditions at high NO concentrations are shown. With the exception of NicD and NbdA, no domain resolution is shown. For BdlA, domains are indicated only to indicate non-processive cleavage for BdlA activation. DSF, cis-11-methyl-2-dodecenoic acid. PDE, phosphodiesterase. DGC, diguanylate cyclase. HTH, helix-turn-helix, transcription regulator. DISMED2, predicted periplasmic sensory domain. MHYT, predicted periplasmic sensory domain for diatomic gases. P, phosphorylation. Arrows indicate increased enzyme activity. Width of gray arrows indicates level of activity

Full size image

3.1 Fatty Acids as Dispersion Signals

Akin to cell–cell signaling molecules, fatty acid signals are involved in intra-species, inter-species, and cross-kingdom communication where they regulate community-associated behavior including biofilm dispersion. The fatty acid acting as a dispersion autoinducer of P. aeruginosa biofilms has been identified as cis-2-decenoic acid (cis-DA) [10]. Additional dispersion autoinducer molecules include cis-11-methyl-2-dodecenoic acid (DSF) that has been shown to disaggregate flocs by Xanthomonas campesitris in liquid [46], and the Burkholderia cenocepatia cis-2-dodecenoic acid (BDSF) [47]. The dispersion response to cis-unsaturated fatty acids is fairly conserved, as cis-DA has been shown to induce dispersion of biofilms by Escherichia coli, Klebsiella pneumoniae, Proteus mirabilis, Streptococcus pyogenes, Bacillus subtilis, Staphylococcus aureus, and yeast Candida albicans biofilms [10], while the B. cenocepatia BDSF has been shown to trigger dispersion of Francisella novicida biofilms [47]. Amari et al. [51] demonstrated that production of cis-DA requires an enoyl-CoA synthetase encoded by dspI (PA14_54640, a PA0745 ortholog), with dspI inactivation resulting in significantly reduced dispersion and defective swarming motility. Microarray analysis furthermore suggested cis-DA to affect the expression of 666 genes encoding proteins involved in motility, chemotaxis, cell attachment, TCA cycle, exopolysaccharides and LPS synthesis and secretion, virulence, iron uptake, and respiration [52]. Rahmani-Badi et al. [52] furthermore predicted PA4982–PA4983 encoding a two-component system, to be involved in cis-DA signal perception. No experimental evidence, however, for cis-DA being sensed via PA4982–PA4983 exist. While mechanism by which cis-DA is perceived by P. aeruginosa and relayed to result in dispersion has yet to be elucidated, much is known about DSF signal sensing by X. campesitris (Fig. 31.2a) which requires the rpf gene cluster. While RpfF directs the synthesis of DSF, a two-component sensory transduction system comprising the hybrid sensor kinase RpfC and the response regulator RpfG has been implicated in the perception of the DSF signal and signal transduction [46]. DSF perception by RpfC is believed to lead to its autophosphorylation and subsequent phosphorelay to RpfG. RpfG is unique in that it contains no DNA-binding domain, but an HD-GYP domain that exhibits phosphodiesterase activity capable of degrading cyclic di-GMP to GMP [46, 53,54,55]. Phosphorylation is thought to activate RpfG for cyclic di-GMP degradation. In this way, RpfC/RpfG link perception of the cell–cell signal DSF to alteration in the cellular level of cyclic di-GMP.

3.2 Nutrient-Induced Dispersion

Both starvation and the sudden excess of nutrients have been shown to induce dispersion by Pseudomonas sp (Table 31.1). Nutrient cues include glucose, glutamate, succinate, citrate and induce dispersion when biofilms are exposed to a sudden increase in the carbon source concentration. Basu Roy and Sauer [11] demonstrated that while L-glutamate supported growth of P. aeruginosa, D-glutamate did not. However, both D- and L-glutamate were capable of inducing dispersion, indicating that nutrient cues are not metabolized in order to induce dispersion. Instead, in P. aeruginosa, nutrient cues including glutamate, citrate, and glucose are sensed by the diguanylate cyclase NicD belonging to a family of seven transmembrane (7TM) receptors (Fig. 31.2b). NicD directly interacts with BdlA and the phosphodiesterase DipA, with NicD contributing to the membrane association of the protein complex [11, 37]. Nutrient cue perception by NicD is believed to lead to dephosphorylation, with posttranslational modification coinciding with increased cyclase activity. Thus activated NicD contributes to the non-processive proteolysis and activation of the chemotaxis transducer protein BdlA, via phosphorylation and temporarily elevated cyclic di-GMP levels [11, 14]. BdlA activation requires an unusual, non-processive proteolytic cleavage found to be stimulated by increased cyclic di-GMP levels, and dependent on the protease ClpP, the chaperone ClpD, and BdlA phosphorylation [14, 37]. BdlA, in turn, activates the phosphodiesterase DipA, and recruits a second phosphodiesterase, RbdA, to ultimately reduce cellular cyclic di-GMP levels (Fig. 31.2b). An additional player is the diguanylate cyclase GcbA. GcbA contributes to BdlA cleavage during biofilm growth and has been shown to play an essential role in allowing biofilm cells to disperse in response to a variety of substances including carbohydrates, heavy metals, and NO [14, 15]. Likewise, BdlA and DipA appear to be required for P. aeruginosa biofilm dispersion in response to NO and heavy metals [12, 13], indicating GcbA, BdlA, and DipA to play a central role in the translation of a large variety of dispersion cues into the modulation of the intracellular cyclic di-GMP pool.

3.3 NO-Induced Dispersion

The diatomic gas nitric oxide (NO), a well-known signaling molecule in both prokaryotes and eukaryotes, is able to induce the dispersal of P. aeruginosa and other Gram-negative bacterial biofilms (Table 31.1). NO was first suggested by Webb et al. [56] to stimulate the release of planktonic cells from an established P. aeruginosa biofilm. The finding of NO serving as dispersion inducer was confirmed by Barraud et al. [18, 57] using several NO donors. Moreover, the studies linked NO to low cyclic di-GMP levels and changes in phosphodiesterase activity in a dose-dependent manner. In most species, NO is sensed by H-NOX (heme-nitric oxide/oxygen-binding) domain proteins, by NO binding to the heme moiety of H-NOX. H-NOX can directly interact with DGC to regulate cyclic di-GMP synthesis and degradation. In Legionella pneumophila or Shewanella woodyi, the H-NOX protein interacts upon NO binding with the bifunctional GGDEF-EAL (HaCE) protein, and lowers cyclic di-GMP levels by inhibiting the DGC activity but stimulation the PDE activity of the HaCE [58] (Fig. 31.2c). In Vibrio cholerae or S. oneidensis, interaction of the NO-bound H-NOX domain with a coupled histidine kinase (H-NOK) controls the phosphorylation activity of the kinase (Fig. 31.2c). Specific phosphorylation events lead to a decrease in cyclic di-GMP levels, either by stimulating the hydrolysis of cyclic di-GMP by a cognate PDE (via the fused REC domain) or by controlling the transcriptional response through a dedicated transcription regulator (HTH) [58]. In P. aeruginosa, NO sensing likewise involves the activation of cyclic di-GMP-specific phosphodiesterases in P. aeruginosa, ultimately leading to cyclic di-GMP decrease and biofilm dispersal (Fig. 31.2d). However, P. aeruginosa does not encode H-NOX proteins. Instead, NO sensing in P. aeruginosa was found to be linked to NbdA, an MHYT domain harboring phosphodiesterase [38, 58]. MHYT is a transmembrane domain of seven predicted membrane spanning helices and proposed to possess putative sensory function for diatomic gases like oxygen, carbon monoxide, or NO through protein-bound copper ions [59]. Considering that inactivation of bdlA, dipA, and gcbA impairs dispersion by P. aeruginosa biofilms in response to NO, the signaling cascade likely requires, in addition to NbdA, also BdlA, DipA, (Fig. 31.2d), and GcbA.

4 The Dispersion Phenotype

The cellular cyclic di-GMP levels noted upon dispersion cue sensing and induction of dispersion have been reported to be comparable to or lower than those found in planktonic cells [12, 13, 27, 28]. The low cyclic di-GMP levels explain much of the similarities found between dispersed and planktonic cells. Relative to biofilms, both are motile, and are susceptible to antimicrobial agents [16, 60,61,62]. However, dispersed cells are not identical to planktonic cells [4, 60, 61]. Instead, dispersed cells were found to be highly virulent when tested using various acute and chronic virulence models, to produce more matrix degrading enzymes, to be more primed to re-attach following egress from the biofilm, and to exhibit protein production and gene expression profiles that are distinct from planktonic cells and biofilms from which they escaped [4, 39, 60, 61, 63]. The distinct phenotype of dispersed cells, however, was found to be reversible and short-lived. Using qRT-PCR and antimicrobial susceptibility assays, Chambers et al. [60] demonstrated that in P. aeruginosa, differences between planktonic and dispersed cells remained for 2 h post-dispersion, with additional time being required for dispersed cells to display expression of genes indicative of exponential growth.

5 Cyclic di-GMP Levels and Downstream Pathways

The finding of dispersed cells being characterized by reduced cyclic di-GMP levels has led to the hypothesis that dispersed cells can be generated by reducing the intracellular cyclic di-GMP content through modulation of PDEs [28]. However, the cyclic di-GMP activated pathways have not been fully elucidated. Considering that dispersion coincides with biofilm erosion and single cells escaping the biofilm structure (Fig. 31.1), dispersion likely relies on factors that weaken the biofilm matrix. The biofilm matrix is composed of polysaccharides, eDNA, and adhesins [64], with Pseudomonas sp. using cyclic di-GMP regulated adhesins to reinforce the biofilm matrix. These adhesins have been identified in P. putida and P. fluorescens as the large outer-membrane protein LapA [27, 65,66,67], and CdrA in P. aeruginosa [67, 68]. Elevated cyclic di-GMP levels contribute to the localization of LapA to the cell surface, while low cyclic di-GMP levels result in LapA being released from the outer membrane via cleavage by the periplasmic cysteine protease LapG [66]. Gjermansen et al. [27] demonstrated that in P. putida, carbon starvation decrease the level of LapA, with LapA release resulting in biofilm dispersal, a response that was absent in ΔlapG mutant biofilms. Additionally, a plethora of matrix degrading factors such as proteases, deoxyribonucleases, and glycoside hydrolases have been linked to biofilm dispersal [39, 61, 69,70,71,72]. However, most studies have relied on inducing dispersion by the exogenous addition of these factors. For instance, PslG, a glycosyl hydrolase involved in the synthesis/degradation of a key biofilm matrix exopolysaccharide Psl in P. aeruginosa, disassembles existing biofilms within minutes at nanomolar concentrations when supplied exogenously [71]. However, as PslG is not predicted to be released from the cell, it is unlikely that PslG indeed contributes to matrix degradation during dispersion or activated in a low cyclic di-GMP environment. As for now, specific matrix degrading factors remain elusive.

Dispersion furthermore coincides with bacteria escaping from the biofilm being susceptible to antimicrobial agents. Recent findings suggested a link between cyclic di-GMP and drug susceptibility [60, 73, 74]. For instance, Gupta et al. [73] demonstrated that P. aeruginosa planktonic cells were rendered more resistant to antimicrobial agents upon increasing intracellular cyclic di-GMP, from 10–30 pmol/mg, to cyclic di-GMP levels more commonly found in biofilm cells (≥80 pmol/mg). Additionally, drug tolerance by P. aeruginosa biofilms and dispersed cells has been linked to the cyclic di-GMP-responsive transcriptional regulator BrlR [60, 75]. BrlR contributes to biofilm drug tolerance by activating the expression of multidrug efflux pumps and ABC transporters [76,77,78]. Low cyclic di-GMP levels, however, negatively impact BrlR levels and BrlR-DNA binding [26].

6 Concluding Remarks

Being a near-ubiquitous second messenger that coordinates diverse aspects of bacterial growth and behavior, it is not surprising that cyclic di-GMP has become known as the “second messenger extraordinaire” [79]. However, despite the large number of bacterial behavior and functional outputs that have been characterized since its discovery in the late 1980s, there is still much to learn, especially when it comes to the role of cyclic di-GMP in biofilm dispersion. While future experiments will be required to elucidate cyclic di-GMP dependent pathways leading to dispersion, indirect evidence suggests a role of AmrZ and FleQ. Originally described to inversely regulate alginate production and swimming motility in P. aeruginosa, AmrZ is now recognized as a global regulator of multiple virulence factors, including cyclic di-GMP, extracellular polysaccharide production including Pel and Psl polysaccharides, and flagella [80]. Support for AmrZ playing a role in dispersion stems from AmrZ affecting gcbA expression and inversely regulating exopolysaccharide production and motility [80]. Additionally, Chua et al. [61] demonstrated amrZ to be differentially expressed in dispersed relative to planktonic cells using RNA-seq. Similarly, FleQ may contribute to the cyclic di-GMP dependent pathways to induce dispersion. This cyclic di-GMP responsive transcriptional inversely contributes to the expression of pel genes required for Pel polysaccharide biosynthesis and of flagellar genes in response to cyclic di-GMP [81,82,83]. At high cyclic di-GMP levels, FleQ induces the expression of the pel operon while at low cyclic di-GMP levels, FleQ regulates the expression of flagellar genes but represses transcription of the pel operon required for Pel polysaccharide biosynthesis. It is of interest to note that FleQ is under the transcriptional control of AmrZ [80].