Biofilms are microbial aggregations adhering to biological or non-biological surfaces. They can be found on any implant but also on the mucosa, e.g. in the cystic fibrosis lung [1], the middle ear [2] or gastric mucosa [3]. Biofilm formation is a key factor for microbial survival in hostile environments, representing a protected mode of growth that allows cells to survive and disperse [4]. Biofilm infections are a huge problem in the clinic and cause many deaths and high health costs, e.g. 12% to 25% of patient mortality are attributable to catheter-related bloodstream infections [5].

Biofilm infections are difficult to eradicate because the genetic program (gene expression) of bacteria in this structure is fundamentally changed [6] resulting in a much better protection against macrophages and antibiotics, compared to free living cells [7]. The eradication of Escherichia coli biofilms required 220 times higher antibiotic concentrations than for the same strain in serum [8]. The reasons for the higher antibiotic resistance of bacteria in biofilms are many fold and still not very well understood. In biofilms, up to 20% of all bacterial genes are expressed differently [9]. Bacteria grow slower and the reduced metabolic activity renders them less prone against most antibiotics. They also produce compounds actively detoxifying antibiotics, e.g. Pseudomonas aeruginosa produces only in biofilms a periplasmic cyclic glucan which complexes the antibiotics tobramycin andgentamicin, inactivating them [10]. This shows that the extracellular matrix of the biofilm is not only a passive diffusion barrier for antibiotics but is also actively shaped by species within the microbial biofilm communities. Interestingly, it has recently been reported that swarming also causes increased antibiotic resistances and that quorum-sensing is not involved in this effect [11].

Before forming biofilms, bacteria have to synchronize their gene expression in a process called quorum-sensing (QS) [12]. They do this by secreting small extracellular signal molecules acting as autoinducers to start genetic programs [13]. When a certain threshold of autoinducer concentration is reached, cells attach to the surface forming a biofilm and starting the production of virulence factors. Quorum-sensing is necessary both to start biofilm formation and, to a lesser extent, to maintain the biofilm. This process, however, is not absolute. It depends on the environmental conditions experienced by the cells and varies between the different species [14,15]. Iron is essential for most bacteria and its availability is often limited during the infection process. It has been shown that iron modulates the stress response in biofilms and overrides the expression of superoxide dismutases in P. aeruginosa, which is otherwise induced by QS [16]. Transcriptome results indicate that the QS system in P. aeruginosa, and probably in other bacteria as well, is far more complex than previously thought and many QS-regulated genes encoding virulence products were expressed in all conditions investigated [17]. In most pathogens studied so far, blocking QS does not abolish biofilm formation, but does make the biofilm more susceptible to antimicrobials and immune reactions.

Currently, more than 50 different autoinducers are known and the best studied ones, called autoinducer-1 (AI-1), are N-acyl-L-homoserine lactones (AHL) 1–6, which are only known from Gram-negative bacteria. The acyl side chain of AI-1 has between 4 and 16 carbons and is usually saturated, although mono-unsaturated side chains are also known. AI-1 compounds 3–6 have also a hydroxyl or oxo-function at C-3 [18] (Figure 1). Another group of autoinducers, autoinducer-2 (AI-2), is known both from Gram-positive and -negative bacteria. The first AI-2 was found in Vibrio harveyi and identified as the cyclic boronic ester 7 [19]. Later, the epimer 8 lacking the boronic ester was detected in Salmonella typhimurium [20]. AI-2 is produced and detected by many bacteria and regarded as autoinducer for interspecies communication between bacteria [21]. A contradictory finding is that the only genes known to be regulated by the AI-2 system in bacteria other than Vibrio species encode the ABC transporter, LsrR, which is responsible for the uptake of the AI-2 signal in S. typhimurium and Escherichia coli. In the cell, AI-2 interacts with LsrR, which then represses the lsr operon [22,23]. This leaves the exact action and function of AI-2 in these organisms open, and a non-quorum sensing role for luxS in most bacteria has been suggested [24]. Such a non-quorum-sensing role of AI-2 has been demonstrated for Staphylococcus aureus. When grown together with the wild type in mixed culture, mutants of various S. aureus strains that are unable to produce AI-2 showed reduced ability to compete for growth under sulphur limitation. Inactivation of AI-2 production did not affect biofilm formation, nor virulence-associated traits, such as production of hemolysins and extracellular proteases. Interestingly, AI-2 production does not appear to contribute to the overall fitness of S. aureus during intracellular growth in epithelial cells [25].

Instead of acyl homoserine lactones (AHL), Gram-positive bacteria use small peptides as autoinducers (AIP). All peptide signalling molecules are formed by posttranslational modification of precursor peptides which are then actively secreted. Sensory information is relayed into the cell by phosphorylation cascades altering gene expression [26]. The cyclic octa- and nonapeptide AIPs from several Staphylococcus species are thiolactone peptides. Only the central cysteine and its distance to the C terminus are conserved but the primary sequence of the pheromones vary even in subgroups within one species [27,28]. For example, the AIPs 9 and 10 have been identified from S. epidermidis isolates [29]. The only exception is the AIP 11 of S. intermedius where the cysteine is replaced by a serine forming not the usual thiolactone, but a lactone. The AIPs produced by some strains of the same species inhibit other strains even of the same species, probably to exclude them from infection or colonization sites, or both [30].

Most bacteria live in multi-species microbial communities, therefore, the quorum-sensing signals are not only received and processed by cells of the same species, but also by foreign species [31]. This has been shown for E. coli, where both gene expression and phenotype alter in response to AHLs of foreign bacteria [32]. Burkholderiacepacia reacts to AHLs produced by Pseudomonas aeruginosa, but P. aeruginosa does not respond to B. cepacia AHLs [33]. The consequence is that B. cepacia always occurs together with P. aeruginosa in the cystic fibrosis lungs while P. aeruginosa is also found without a B. cepacia co-infection [34]. A much more complex autoinducer signal interaction concerning autoinducer AI-2 is found between the oropharyngeal flora and P. aeruginosa in cystic fibrosis. In this case, AI-2 is produced by the oropharyngeal flora and P. aeruginosa responds to it, although it does not produce AI-2 itself [35]. Additionally, antibiotic activity has been demonstrated for 3-oxo-dodecanoyl-AHL 6 and is directed against some Gram-positive bacteria, but was not observed for Gram-negative species [36].

Quorum-sensing is currently regarded as a key mechanism in biofilm development and the wealth of autoinducers offer an attractive new target for non-antibiotic control of biofilm infections [68]. The sophisticated action of autoinducers in different bacteria species, the existence of several QS-control mechanisms in one cell, and the complex interaction between autoinducers of different species, demonstrated that the goal of controlling biofilm infections is not easily achievable. Obviously, searching for biofilm control beyond the interference with the known QS-autoinducers seems to be an attractive goal.

There are several reports stating that the virulence of some pathogens is reduced when they are growing in biofilms. At the ASM Biofilms 2009 conference in Cancun, Mexico, David Davies presented data showing that the application of the biofilm dispersing factor 2-cis-decenoic acid (53) to Pseudomonas fluorescens infected leaves led to the dispersion of the biofilm on the one hand, but also to an enhanced virulence of the pathogen [88]. Further evidence comes from studies of the type III secretion systems (T3SS) in many bacteria, translocating pathogenicity factors into eukaryotic host cells. For P. aeruginosa, a balance was reported between active repression of the T3SS, leading to the establishment of P. aeruginosa in biofilms, and expression of T3SS in these biofilms, leading to the formation of virulence factors [89]. In contrast to these results, P. aeruginosa small colony variants, isolated from chronically infected cystic fibrosis patients, have both an increased capacity to form biofilms and increased T3SS gene expression and cytotoxicity [90]. The report that planktonic cells do not express T3SS, but cells in biofilms do, underlines the fact that the relations between biofilm mode of growth and T3SS expression is still not completely understood [91]. It seems that a homeostasis exists which might contribute to the persistence of chronic infections. As a consequence, biofilm repression might lead to an activation of T3SS and increased export of virulence factors.

In S. aureus, sarZ deletion mutant clones presented increased expression of SarA protein and a strong down-regulation in agr RNAIII transcription, both important components of the virulence gene regulation in staphylococci. Accordingly, the expression of virulence factors (alpha-hemolysin and serine protease sspA) in these mutants is markedly reduced, while the biofilm formation ability is actually enhanced. The results suggest the promotion of virulence by SarZ includes repression of sessile bacteria establishment in biofilms, hence maintaining active S. aureus infections [92].

Another interesting mechanism by which bacteria can prevent eradication of infection is called indirect pathogenicity. In this polymicrobial interaction, an antibiotic-sensitive pathogen can be commensally protected by resistant bacteria of low intrinsic virulence. Studies in a model using Escherichia coli strains differentially susceptible to antibiotics showed that the biofilm environment limited the beneficial interaction between the strains in ampicillin-containing medium, the decline of susceptible population being two orders of magnitude lower in planktonic cultures than in biofilms [93].

These reports demonstrate that merely dissolving biofilms may not be sufficient to control biofilm infections. Nevertheless, quorum-sensing systems are important targets to address the sensitivity of bacteria to antibiotic compounds and the host immune system itself. This was demonstrated, for example, using P. aeruginosa deletion mutants for QS-receptor (ΔlasR rhlR). Mutant biofilms were significantly more sensitive towards reactive oxygen and polymorphonuclear leukocytes activity, being more rapidly cleared in an in vivo model of lung infection. Also, QS-deficient biofilms were more susceptible to killing by tobramycin [94]. Although it might be enough to dissolve the biofilm, attenuating the antibiotic resistance of the pathogens to the levels of single cells and leaving the rest to the immune reaction of the host, it is more advisable to support this process by a suitable antimicrobial therapy.

Dissolving pathogenic biofilms exposes the bacteria to the immune system of the host. It may be possible that the immune reaction is sufficient to clear the pathogen, but it is more likely that some bacteria escape the immune response and establish again biofilms. To address this problem, several attempts have been made to kill the bacteria released from the biofilm by treatment with standard antibiotics. In a pulmonary model of chronic lung infection ,the quorum-quenching furanone 29 was administered to mice infected with Pseudomonas aeruginosa two days previously. Attenuation of virulence factor expression and much better clearance of the bacteria by the immune system were observed [95]. From the fungus Penicillium, patulin (17) and penicillic acid (18) were identified as quorum-quenching compounds. Furthermore, both compounds activate polymorphonuclear neutrophils, breaking the blockage of the oxidative burst. When patulin (17) and penicillic acid (18) were tested in the mouse model, results similar to the furanone 29 were observed [96]. All three compounds increased the sensibility of P. aeruginosa to tobramycin, opening another window to control this bacterium in the lung of cystic fibrosis patients.

The triterpene ursolic acid (22) was found to inhibit biofilm formation in Escherichia coli, P. aeruginosa and Vibrio harveyi, (10 µg mL^-1) when added upon inoculation or to a 24-h biofilm, but it does not interfere with quorum sensing, as shown with V. harveyi AI-1 and AI-2 reporter systems [97]. The closely related asiatic acid (23) and corosolic acid (24) are even more effective and enhance the susceptibility of P. aeruginosa to antibiotics [98] (Figure 2). This activity is limited to biofilms younger than 24 h; however, biofilm infections are usually caused by much older biofilms, limiting any application of these compounds in medicine.

In Agrobacterium tumefaciens, TraR is the receptor for AHLs. A computer-directed search using the 3D-structure of TraR for natural compounds fitting into the receptor pocket identified, among other compounds, baicalein (20) as a possible candidate. Baicalein could indeed dissolve P. aeruginosa biofilms and showed synergistic effects with ampicillin, causing eradication of P. aeruginosa biofilms at 2 μg mL^-1, a concentration not effective to kill the bacteria without baicalein [99]. No in vivo experiments have been reported and it is not known whether baicalein in combination with ampicillin can also eradicate old P. aeruginosa biofilms, e.g. in the lung of cystic fibrosis animals.

As described above, low concentrations of NO cause P. aeruginosa biofilms to disperse, converting the bacteria in the planktonic phase to their normal susceptibility to antibiotics. This led to the suggestion that a combined exposure to both NO and antimicrobial agents may be a way to control pre-established P. aeruginosa biofilm infections [100]. A similar approach has been suggested by Melander et al., who combined the anti-biofilm activity of a synthetic compound with photodynamic inactivation involving a suitable dye to kill Acinetobacter baumannii biofilms [101]. The same group reported on a dramatically increased sensitivity against conventional antibiotics when biofilms of multi-resistant Staphylococcus aureus (MRSA) or A. baumannii were treated with biofilm-controlling 2-aminoimidazoles in combination with antibiotics [102].

The enzymatic activity of the β-N-acetyl-glucosaminidase dispersin B is able to dissolve mature biofilms of S. epidermidis. As a promising tool for preventing biofilm colonization in medical devices, dispersin B was shown to directly decrease biofilm formation in polyurethane surfaces, and most importantly, to enhance the activity of the antibiotic cefamandole nafate against adherent cells in these surfaces [103].

The prevalence of bacterial pathogens living in biofilms, where they are much more resistant to antibiotics and clearance by the immune system, has promoted the search for new strategies to control biofilm infections. The discovery that bacteria have to coordinate their genetic programs to form a biofilm, and that they use small molecule autoinducers to do so, led to the search for quorum-quenching drugs. Impressive progress has been made in this area, revealing the huge diversity of autoinducers and their mechanisms of action, and convincing in vitro results have been published. However, none of the compounds made it yet to the clinical phase II stage. One problem is that many compounds can prevent the formation of biofilms but are not effective in dissolving existing ones and are, therefore, of little use for the treatment of already existing biofilm infections. Furthermore, for some compounds, even merely the concentration can switch between agonizing and antagonizing quorum-sensing [104]. In addition, redundancies among QS systems of several pathogens can render inhibitors of a single system ineffective, as has been shown e.g. for the AHLs of Pseudomonas aeruginosa or the AIPs of Staphylococcus aureus. Most of these studies focus on few, well studied pathogens [105] but most bacteria occur in complex microbial communities. An advantage for pathogens in microbial communities is that the communication advances to a complex interspecies cross-talk, favouring cheating, where single cells do not bother to take the metabolic burden to produce the autoinducer, but take it from the neighbour. To deal with such new challenge for drug development, other components for biofilm maintenance than quorum-sensing have been explored, opening novel strategies for control. Furthermore, several studies have revealed a balance for pathogenicity and the biofilm mode of growth [106].

The two problems: high specificity of autoinducers combined with a complex network of—often redundant—regulation circuits, and an increased virulence of pathogens released from biofilms, are real challenges in the development of quorum-quenching compounds for medical applications. As a consequence, many researchers have combined quorum-quenching with antibiotic treatments and demonstrated in animal studies that it worked well. This approach is based on the observation that most quorum-quenching compounds enhance the sensitivity of pathogens to antibiotics, even if the quorum-quenchers could not achieve complete dissolution of the biofilms. In animal models, this strategy of combination has already been successfully validated and could overcome many shortfalls of the application of quorum-quenching or other biofilm-controlling compounds alone. Such a combination is attractive and holds high expectations, but one should also bear in mind that we are dealing here with drugs influencing each other and complicating pharmacokinetics. Future biofilm infection treatment will probably use sophisticated drugs tailored to the causing pathogen and comprising both biofilm-controlling compounds and antibiotics.