Bacteria display two distinct life forms during growth and proliferation. In one form, the bacteria appear as single, free-floating planktonic cells, in the other, bacteria are organised into sessile aggregates. The latter form is commonly referred to as the biofilm mode of growth. These two life forms have serious implications for bacterial infections in humans. Acute infections that involve planktonic bacteria are generally treatable with antibiotics. However, in cases where bacteria succeed in forming a biofilm within the human host, the infection might prove to be untreatable and will often develop into a chronic state. The important hallmarks of chronic, biofilm-based infection are an extreme tolerance to antibiotics and conventional antimicrobial agents, and an almost infinite capacity to evade the host defence systems – in particular, the host innate immunity. In order to devise novel intervention principles for biofilm control, we have to understand the molecular mechanisms that underlie biofilm formation and this remarkable tolerance. One important control element that has a role in persistent biofilm infection is cell-to-cell communication, which is also known as quorum sensing (QS).

N-acyl-L-homoserine lactone (AHL)-based signalling systems are the most well studied examples of QS. By 2002, more than 70 Gram-negative bacterial species had been reported to harbor AHL-based systems (Ref. Reference Taga and Bassler1). QS controls a wide range of functions in Gram-negative bacteria, including Ti-plasmid conjugation in Agrobacterium tumefaciens, antibiotic production in Erwinia carotovora (Ref. Reference Taga and Bassler1), virulence gene expression in Vibrio cholerae (Ref. Reference Zhu2), Burkholderia cepacia (Ref. Reference Gilligan3) and Pseudomonas aeruginosa (Ref. Reference Whiteley, Lee and Greenberg4), and surface motility by means of swarming in Serratia liquefaciens (Ref. Reference Eberl5), P. aeruginosa (Ref. Reference Kohler6) and B. cepacia (Ref. Reference Huber7).

QS systems have also been identified in Gram-positive bacteria. Rather than AHL molecules, Gram-positive bacteria use small peptides, usually 5–17 amino acids in length, as signalling molecules (Ref. Reference Federle and Bassler8). Examples of Gram-positive QS-controlled behavior include: development of genetic competence and sporulation in Bacillus subtilis (Ref. Reference Magnuson, Solomon and Grossman9), virulence expression in Enterococcus faecalis (Ref. Reference Qin10) and Staphylococcus aureus (Ref. Reference Kong, Vuong and Otto11) and the somewhat controversial biofilm regulation in S. aureus (Ref. Reference Boles and Horswill12). The existence of a third type of QS system – the autoinducer-2 (AI-2) system – has been proposed (Ref. Reference Winans and Bassler13); however, this and the Gram-positive QS systems are outside the scope of the present review (see also Fig. 1).

Figure 1.

The basic quorum sensor. R represents the regulatory protein and I represents the synthetase protein that generates signal molecules. When the critical cell density is reached, a complex of the R protein and a signal molecule enables autoinduction of the QS sensor and expression of the target genes. A list of QS inhibitors is shown on the right; asterisks indicate QS inhibitors shown to compete with the binding of the signal molecule to the R protein. Crosses indicate putative inhibitory sites. The list below the illustration shows the variety of QS-regulated phenotypes.

Pseudomonas aeruginosa

P. aeruginosa is a well characterised bacterium that is capable of causing chronic infections. It belongs to the genus Pseudomonas, which is in the γ-subclass of Proteobacteria. Characteristics of the genus are simple growth requirements and tolerance to a wide range temperatures (4–42°C). Pseudomonas is one of the most diverse and ecologically significant groups of bacteria known. It can be isolated from most environments, including soil, marshes, coastal marine habitats, plants and mammal tissue (Ref. Reference Stover14). In hospital environments, it grows in moist reservoirs such as food, cut flowers, sinks, toilets, floor mops, dialysis equipment and even in disinfectant solutions (Ref. Reference Murray15). P. aeruginosa is an opportunistic human pathogen, which is not part of the normal human flora, but it can give rise to acute infections in hospitalised patients, in particular immunocompromised individuals, such as patients with AIDS and neutropaenic patients undergoing chemotherapy. P. aeruginosa is a significant source of infections in burn victims and patients having foreign bodies inserted. It causes pneumonia in patients receiving artificial ventilation in intensive care units, as well as keratitis in users of extended-wear soft contact lenses (Refs Reference van Delden and Iglewski16,Reference Lyczak, Cannon and Pier17). It is also the predominant cause of chronic lung infection and mortality in cystic fibrosis (CF) patients (Ref. Reference Koch and Hoiby18). Other susceptible patients include those suffering from venous leg ulcers and diabetic foot ulcers.

Pseudomonas aeruginosa biofilms

The various definitions of bacterial biofilm all seem to agree that biofilms consist of numerous bacteria organised in a structured community composed of microcolonies embedded in a self-produced matrix. We believe that attachment to a surface is not an absolute requirement and propose the following definition of a biofilm as: ‘a coherent cluster of bacterial cells embedded in a biopolymer matrix, which compared with planktonic cells, shows increased tolerance to antimicrobials and resists the antimicrobial properties of the host defence’.

The ability of P. aeruginosa to readily form biofilms is likely to contribute to its ecological versatility. Biofilm formation is a survival strategy in hostile environments, which include that of the host organism. Biofilm formation might therefore also be considered to be an important pathogenicity trait. Generally, biofilms consist of bacteria encapsulated in an extrapolymeric substance (EPS) that is either produced by the bacteria or sometimes additionally adapted from the host. Biofilm architecture promotes the development of nutrient and oxygen gradients (Ref. Reference Walters64). The EPS consists of polysaccharides, extracellular DNA and other macromolecular components such as proteins, lipids and biosurfactants. It seems to constitute the scaffolding component of the aggregating bacteria in the biofilm (Refs Reference Nivens65, Reference Allesen-Holm66), acts as scavenger of free oxygen radicals (Ref. Reference Simpson, Smith and Dean67) and binds many antibiotics, including aminoglycosides (Ref. Reference Allison and Matthews68).

The role of Pseudomonas aeruginosa in chronic infections

P. aeruginosa biofilms have a key role in many persistent infections, in particular in non-healing leg ulcers (chronic wounds) (Refs Reference Bjarnsholt117, Reference Kirketerp-Moller118) and the chronic lung infection of CF patients (Ref. Reference Barnsholt119). Chronic skin wounds can arise from pressure wounds or cuts in patients with reduced or impaired blood circulation in the lower extremities. The chronic wound remains at an inflammatory stage, which is characterised by a continuing influx of polymorphonuclear neutrophils (PMNs) that release cytotoxic enzymes, free oxygen radicals and inflammatory mediators, causing extensive collateral damage to the host tissue.

CF is a monogenic autosomal recessive, multiorganic disease, although the main cause of morbidity and mortality has its root in the patient's increased susceptibility to lung infections. The genetic cause of CF was identified in 1989 as a defect in the gene encoding the cystic fibrosis transmembrane conductance regulator (CFTR), which is located on chromosome 7 (Refs Reference Kerem120, Reference Riordan121). This causes a dysfunction of a chloride ion channel, which depletes the otherwise normal mucus of water and impairs the mucociliary clearing mechanism of the bronchia. Inhaled microorganisms are therefore able to multiply and an inflammatory response is recruited, leading to highly viscous mucus. The lower respiratory tract in patients with CF is recurrently colonised with a variety of bacteria such as Haemophilus influenzae, S. aureus and P. aeruginosa, before it is chronically infected with P. aeruginosa (80% of adult patients). However, aggressive antibiotic treatment, which delays the onset of chronic infection, remarkably prolong the life expectancy of patients with CF (Ref. Reference Hansen, Pressler and Hoiby122).

There are obvious similarities with respect to the bacterial infection found in patients with CF and those with chronic wounds. First, the conditions remain chronic because of the continuous presence of bacteria (in particular P. aeruginosa). Microscopic inspection of ex vivo samples from wounds and lungs of CF patients reveals that the bacteria are assembled in structures consisting of aggregated cells, which are surrounded but not penetrated by inflammatory cells, in particular phagocytic cells such as PMNs (Fig. 2c). We propose that bacterial persistence, including the extreme tolerance to antibiotics and evasion by the immune system is caused by the bacteria's ability to form these aggregations, which represent bacterial biofilms in vivo. The presence of the matrix component alginate can be demonstrated by using fluorescently labelled antibodies (Refs Reference Bjarnsholt123, Reference Kirketerp-Moller124). Second, both kinds of patients are immunocompromised and suffer from defects in the primary line of defence. The chronic wound consists primarily of granulation tissue composed of a network of collagen fibers, new capillaries and extracellular matrix, together with PMNs, macrophages and fibroblasts. Aggregated microcolonies of bacteria were found embedded in this matrix. This is in accordance with observations that P. aeruginosa is also detected in aggregated microcolonies from the chronic infected CF lung within the lumen of the bronchial airways (Ref. Reference Barnsholt119). Since neither disease causes defects in cellular defence, the PMNs would be expected to eradicate the bacteria. However, similarly to in vitro P. aeruginosa biofilms overlaid with freshly isolated human PMNs (Ref. Reference Jensen125), the bacteria resist phagocytosis by eradicating the PMNs upon contact (Fig. 2c). Recent results demonstrate that P. aeruginosa produces rhamnolipids (Ref. Reference Jensen19), detergents that lead to cellular necrosis and elimination of the PMNs (Refs Reference Jensen19, Reference Alhede126, Reference van Gennip127). Measurements of in vitro biofilms reveal that rhamnolipids are not excreted into the surrounding flow medium but rather stick to the aggregated bacterial biofilm. Accordingly, the rhamnolipids function as a PMN shield (Fig. 2c,d), and the capability of mounting this shield is proposed to significantly contribute to the ability of P. aeruginosa to persist in the lung, as well as in chronic wounds. The result of sustained PMN lysis is that antimicrobials, as well as tissue-devastating compounds, spill out. Examples of such compounds are myeloperoxidase, elastase and matrix metallopeptidase 9 (MMP9); chronic wound fluids are particularly rich in these compounds, in contrast to their lower levels in fluid from acute wounds in humans (Ref. Reference Wysocki, Staiano-Coico and Grinnell128). The same is also true for CF patients compared with patients suffering from acute respiratory failure (Ref. Reference Gaggar129). It is also suggested that the elimination of PMNs facilitates additional colonisation. Antibiotic treatment is likely to reduce the number of planktonic bacteria in the wound and in the lower respiratory areas of the CF lung; however, the biofilm-embedded P. aeruginosa survive, keeping the wound and the bronchial regions chronically infected.

Figure 2.

In vitro and ex vivo P. aeruginosa biofilms. (a) In vitro flow cell P. aeruginosa biofilm with glucose as the carbon source, showing characteristic mushroom structures. (b) In vitro flow cell P. aeruginosa biofilm with citrate as the carbon source, showing a flat uniform biofilm. (c) P. aeruginosa biofilm detected in an explanted lung from a cystic fibrosis patient with chronic infection. (d) P. aeruginosa in vitro flow cell biofilm overlaid with human polymorphonuclear neutrophils (PMNs). The PMNs are unable to phagocytose the biofilm bacteria because of rhamnolipid production. Images in a, b and d were obtained using confocal laser-scanning microscopy; c was obtained using epifluorescence microscopy. Bacteria are visualised by GFP staining in a, b and d, and by PNA staining in c. PMNs are visualised by DAPI staining in c, and syto62 in d. (Images in a and b are reprinted from Ref. 170, with permission from John Wiley & Sons)

Control over rhamnolipid synthesis is governed by QS, in particular the PQS and rhl systems (Ref. Reference Fazli130). Rhamnolipid production constitutes a ‘shield’ function to promote necrosis of PMN leukocytes (upon biofilm contact), which subsequently causes collateral tissue damage and promotes development of inflammation. This suggests that quora (which are likely to be represented by the observed bacterial biofilm aggregates) have amassed at certain locations in chronic wounds (Refs Reference Kirketerp-Moller118, Reference Alhede131) or in the CF lung (Ref. Reference Barnsholt119). Such quora are then capable of eliminating PMNs by the production of rhamnolipid (Ref. Reference Morici132). Interestingly, AlgR was shown to reduce expression of RhlR-controlled gene expression in the biofilm mode (Ref. Reference Wu133). Concurrently, the concentration of rhamnolipids was below the detection limit in in vitro biofilms (Ref. Reference Morici132). However, exposure to freshly isolated PMNs superseded AlgR repression and significantly increased expression of rhamnolipids up to a level of 50 µg per g biofilm, which in pure form would causes lysis of PMN within 30 minutes of exposure (Ref. Reference Jensen19). The inducing effect was found to require a functional QS system to induce transcription initiation of the genes regulated by RhlR and PQS, and it is possible that P. aeruginosa, in order to upregulate its PMN shield, receives and responds to signal molecules originating from the host defence cells. Specific effects of cytokines such as interferon gamma (IFN-γ; IFNG) are transmitted through binding of IFN-γ to OprF, resulting in the expression of the QS-dependent virulence determinants PA-I lectin, N-butanoyl-L-homoserine lactone (BHL) and pyocyanin (Ref. Reference Kristiansen134).

Cross-kingdom signalling has a significant role in bacteria–host interactions. Recent evidence also supports the view that QS signal molecules have distinct targets and alter gene expression in mammalian cells (Ref. Reference Skindersoe135). These alterations include modulation of proinflammatory cytokines and induction of apoptosis. In particular, it was observed that P. aeruginosa QS signals decrease the production of IL-12 by murine bone-marrow-derived dendritic cells (BM-DCs) without affecting the release of IL-10. BM-DCs exposed to QS signals during antigen stimulation exhibit a decreased ability to induce specific T-cell proliferation. These results suggest that the P. aeruginosa QS signal molecules impede the T-cell-stimulatory effects of DCs, and function as immunomodulators, removing the host-protecting proinflammatory TH1 from the milieu, thereby enabling the establishment of bacterial infection within the host (Ref. Reference Wu136).

Successful treatment of a nonhealing wound depends on the identification and treatment of the factors that are preventing the healing process. Adaptations from the CF experience might be useful. Early eradication therapy against P. aeruginosa and other bacteria has long been recognised to prevent the onset of the chronic infection (Ref. Reference Hansen, Pressler and Hoiby122). Control of the host response, for example by using growth factors and MMP inhibitors, as well as local modulation of the invading bacteria by the topical application of silver ions, might also promote healing. However, the above-described bacteria–host interactions and shielding open up new treatment scenarios. Lately, several studies have proved the concept of inhibiting QS as a functional and relevant antimicrobial treatment strategy (Refs Reference Hentzer35, Reference Bjarnsholt41, Reference Rasch137, Reference Hodson138). It was recently demonstrated that a single genetic inactivation of rhlA, a gene that is important for rhamnolipid synthesis led to significant attenuation in two animal models of infection (Ref. Reference van Gennip127) The rhlAB operon is responsible for the formation of monorhamnolipids. We propose that QS inhibitor (QSI) compounds conduct functional antimicrobial effects in vivo by preventing the establishment of the PMN shield, and consequently restore proper antimicrobial functions of the PMNs. Since PMN lysis is then likely to be reduced, local inflammation and tissue destruction will be kept to a minimum. We also suggest that release of bacterial eDNA will be obstructed and thereby will reduce the viscocity of the bronchial secretion in CF, which in turn, has been shown to increase and maintain lung function (Ref. Reference Hentzer and Givskov139). Both effects will be expected to tip the balance in favor of bacterial eradication and subsequent healing.

Clinical implications

The previous 15 years of research have shown that QS controls the expression of virulence (but not growth) in a number of pathogens. Despite the fact that QS systems do not control expression of essential genes, QS systems have been shown to constitute new promising antimicrobial drug targets (Refs Reference Bjarnsholt40, Reference Bjarnsholt41, Reference Rasch137, Reference Hodson138, Reference de Nys140). Disruption of the QS signalling cascade might be a particularly valuable strategy in the treatment of chronic biofilm infections. Likewise, it has to be emphasised that the rationale behind this approach is to specifically interfere with inflammation and disease-causing traits, rather than restricting the growth of the bacteria. As production of virulence factors is inhibited, the pathogen can no longer maintain its position in the increasingly hostile host environment and consequently will be cleared by the innate host defences. Given that the mere attenuation of the pathogen does not impose a harsh and direct selective pressure on the cells, it is anticipated that resistant mutants will not arise as readily as in the case of conventional antibiotic treatment. We have denoted this the ‘antipathogenic’ drug principle (Ref. Reference de Nys140)

The first example that nature has already applied QSIs as an antibacterial strategy was demonstrated by the production of halogenated furanones by the Australian macroalga Delisea pulchra (Refs Reference Kjelleberg141, Reference Givskov142, Reference Wu143). These natural brominated furanones do not inhibit the QS of P. aeruginosa. A series of synthetic furanones has since been synthesised. Among these, compounds C-30 (Ref. Reference Hentzer35) and C-56 (Ref. Reference Hentzer114) showed QSI potential in P. aeruginosa. It was shown that biofilms grown with and without furanone C-30 were not significantly different in biofilm formation and biomass (Ref. Reference Hentzer35). However, when treating 3-day-old biofilms with 100 µg/ml tobramycin, grown either with or without furanone C-30, a significant number of dead cells were detected in the furanone C-30-treated biofilm. The antimicrobial properties of both furanone C-30 and furanone C-56 were also demonstrated in a chronic pulmonary mouse model (Refs Reference Hentzer35, Reference Kim144), an implant model (Ref. Reference Christensen42) and a fish model for vibriosis (Ref. Reference Hodson138). The bacteria were cleared significantly faster in mice treated with 1 µg furanone per gramme body weight compared with placebo-treated mice. Other furanone derivatives have been synthesised (Ref. Reference Persson145); however, these furanones also block biofilm formation of P. aeruginosa. This indicates that these furanone derivatives are highly toxic to the bacteria, because QS is not required for biofilm formation. Since the discovery of the halo-furanones, a number of papers have reported the synthesis of QSI compounds by means of organic chemistry methods (Refs Reference Riedel146, Reference Persson, Givskov and Nielsen147, Reference Olsen148, Reference Muh149, Reference Borlee150, Reference Geske, Mattmann and Blackwell151, Reference Geske, O'Neill and Blackwell152, Reference Geske153, Reference Geske154, Reference Mattmann155, Reference Bottomley156). However, a detailed overview of this is outside the scope of the present review.

Many of these compounds have been developed on agonist-based designs and probed for their action by means of specific QS screens (see below). In silico analysis showed an ID₅₀ of 30 nM for the 12-akyl-chain-substituted tetrazol PD12 (Refs Reference Persson, Givskov and Nielsen147, Reference Borlee150). This compound is so far the most active, functional QSI drug ever designed; however, target specificity and functionality have never been validated by transcriptome analysis or animal experiments. It must be emphasised that physiologically relevant concentrations of P. aeruginosa QS signals are likely to be in the range of 1 to 20 µM. Recently, the crystal structure of a LasR peptide bound to OdDHL became available (Ref. Reference Amara157). Thus it is now possible to apply in silico analysis of receptor–ligand interactions (Ref. Reference Yang158). Direct covalent interaction of a reactive isothiocyanate AHL derivative with an amino acid present in the binding site was reported. Based on the three-dimensional structure of the LasR protein (without ligand), computer-based virtual screening has been used to search for putative quorum-sensing inhibitors from a database comprising approved drugs and natural compounds (Ref. Reference Rasmussen159). Three potential QSIs that had effects on P. aeruginosa virulence factor expression were identified.

QSI compounds are found in food, herbal and fungal sources. Several years ago we developed a new tool for high-throughput screening of fresh samples, as well as crude extracts and pure QSI compounds (Refs Reference Rasmussen39, Reference Bjarnsholt41, Reference Bjarnsholt160). The presence of QSI compounds in natural foods is extremely interesting, because, in most cases, vegetables and herbs are nontoxic to humans and are easily available. One of the most potent QSIs ever detected is within extract of garlic (Ref. Reference Bjarnsholt41). The fact that fungi also produce QSIs (Ref. Reference Rasmussen44) supports the view that organisms devoid of functional immune systems rely on chemical defence systems against competing or detrimental bacteria.

The antibiotic azithromycin was also reported to inhibit QS of P. aeruginosa (Ref. Reference Tateda161). This led us to investigate whether QS inhibition is a common feature of antibiotics. Twelve antibiotics were screened for their QS-inhibitory activities using the QSI selector system (Ref. Reference Skindersoe52). Three of the antibiotics tested, azithromycin, ceftazidime and ciprofloxacin, were active in the assay and were further examined for their effects on QS-regulated virulence factor production in P. aeruginosa. Consistent results from the virulence-factor assays, reverse-transcription-PCR, and the DNA microarrays support the finding that these three antibiotics decrease the expression of a range of QS-regulated virulence factors. The data suggest that the underlying mechanism is mediated by changes in membrane permeability, thereby influencing the flux of 3-oxo-C12-HSL (Ref. Reference Skindersoe52). Azithromyzin inhibits or reduces the production of several of the virulence factors of P. aeruginosa, such as elastase and rhamnolipids (Ref. Reference Hoffmann162), and interferes with alginate polymerisation (Ref. Reference Kai163). Recently another group reported that azithromycin inhibits AHL synthesis by blocking unknown pathways involved in regulation of lasI and rhlI (Refs Reference Kai163). It is proposed that azithromycin interferes with one of the precursors of the AHL molecules, indicating that the QSI activity is a secondary effect. Azithromycin does improve the respiratory function in CF patients (Ref. Reference Saiman164, Reference Southern and Barker165).

The principal role of QSIs is to attenuate the virulence of the bacteria, and in our opinion, if the production of the excreted rhamnolipids is attenuated, P. aeruginosa is dramatically weakened. As a result of this, the venom alkaloid solenopsin A is of great interest (Ref. Reference Park166). It is extracted from the fire ant Solenopsis invicta and targets the rhl QS system specifically, thereby inhibiting rhamnolipid production.