Paradigm shift in discovering next-generation anti-infective agents: targeting quorum sensing, c-di-GMP signaling and biofilm formation in bacteria with small molecules

The discovery and development of several antibiotics between the late 1930s and 1960s allowed the effective management of previously lethal bacterial infections [1,2]. Penicillin G (2) dubbed the ‘magic drug,’ was responsible for reducing the number of deaths among soldiers in World War II infected with Streptococcus pneumonia[1,2]. When early resistance to penicillin was observed, second-generation antibiotics, such as methicillin (3), cephalothin (4) and imipenem (5), were developed [2]. In 1961, a methicillin-resistance strain of S. aureus (MRSA) was observed [3]. Soon thereafter other bacterial strains that were resistant to antibiotics such as streptomycin (8), chloramphenicol (9) and tetracycline (6) were identified [1,4–6]. Over the years, it has become clear that bacteria can develop resistance to almost any antibiotic.

Table 1 illustrates the severity of the antibiotic resistance problem. With the exception of a few antibiotics, such as vancomycin (14) and erythromycin (7), resistance to most antibiotics was observed only a few years after their introduction into clinical use. In the case of vancomycin and erythromycin, it took almost three decades between the year that the drug was introduced in the US market and the year resistance was first observed (Table 1)[1,7–20].

The three main strategies that are used by bacteria to develop resistance to antibiotics are:

• ▪ Overexpression of enzymes that can modify the antibiotic drug, rendering the antibiotic inactive;

• ▪ Mutation of the bacterial target site that allows the target site to maintain its functional role yet abrogates binding of drug to the target or transverse of the antibiotic across the bacterial cell wall;

• ▪ Export of antibiotic drugs to the extracellular media via multidrug-resistant (MDR) efflux pumps or loss of porin channels resulting in lower permeability of antibiotics.

Drugs such as penicillin G (2) and methicillin (3; β-lactams, see Figure 1) are inactivated via the overexpression of β-lactamases, which hydrolyze the antibiotics [21]. For an excellent review on the mechanism of action of β-lactamases, see elsewhere [22]. Resistance to β-lactams via β-lactamase-mediated hydrolysis can be ameliorated when the β-lactam antibiotic is used in combination with β-lactamase inhibitors such as clavulanic acid [23], tazobactam [23] and sulbactam [24]. Both clavulanic acid and sulbactam irreversibly inactivate β-lactamases [22–24]. Unfortunately some bacterial strains that are resistant to β-lactamase inhibitors have emerged [25]. Aminoglycosides, including streptomycin (8) and tobramycin, which inhibit bacterial rRNA are rendered ineffective by bacteria via the over-expression of antibiotic-modifying enzymes, such as nucleotidyltransferase/phosphotransferase, which transfer a monophosphate moiety from ATP to a hydroxyl on the ring of an aminoglycoside [26]. Other aminoglycoside modifying enzymes include N-acetyltransferases and O-adenyltransferases, which catalyze acetyl CoA-dependent acetylation of amino functionality and ATP-dependent adenylation of hydroxyl group, respectively [26]. Other antibiotics such as the streptogramin class and linezolid (10) are rendered ineffective via the modification of 23S ribosomal RNA [27]. Bacteria can use multiple pathways to inactivate an antibiotic. For example, resistance to vancomycin (14) can occur via thickening of the bacterial cell wall [28], changing the peptidoglycan precursor [1] or expression of an ‘abnormal’ ligase that makes the unusual Ala-lactate peptide bond [29].

Efflux pump proteins such as AcrB in E. coli[30], the Mex-Opr systems in P. aeruginosa[31,32] and NorA in S. aureus[33] export antibiotics such as ciprofloxacin (13)[33] and tetracycline (6)[34,35] out of bacterial cells, thereby raising the minimum inhibitory concentration values of antibiotics. Inhibitors of MDR efflux pumps such as the alkaloid reserpine, biricodar and timcodar (which are also mammalian multidrug efflux inhibitors) have been shown to potentiate antibiotic activity of conventional antibiotics [36]. Other small molecules such as L-phenylalanine-L-arginine-β-naphthylamine and carbonyl cyanide m-chlorophenylhydrazone (CCCP) have also been shown to inhibit the efflux of drugs such as levofloxacin and faropenam from bacterial cells [37,38]. The widespread use of MDR efflux inhibitors as combination therapeutics is probably not as popular as the use of β-lactamase inhibitors, due to the paucity of effective bacterial MDR efflux pump inhibitors.

After years of wavering success with the development of second- and third-generation antibiotics that would ultimately suffer from bacterial resistance, two new drugs were introduced at the beginning of the new millennium; daptomycin (15)[39,40] and linezolid (10)[41,42]. Linezolid (10) is a synthetic oxazolidinone antibiotic, which was initially effective against vancomycin-resistant Enterococci (VRE) and MRSA [41,42]. Linezolid inhibits protein synthesis by binding to the 23S site of the 50S ribosome [41,42]. Although its action is similar to the macrolides and phenylpropanoids, no cross-resistance was observed [41,42]. Regretfully, linezolid resistance was reported in VRE and MRSA only a few years after it was introduced [19,43]. Daptomycin (15), a natural lipopeptide antibiotic, was also introduced in the early 21st century [39,40]. Daptomycin kills bacteria via a novel mode of action involving binding to the bacterial cell wall and depolarizing the cell membrane [39,40]. Unfortunately, the early excitement regarding daptomycin’s ability to kill MRSA and VRE was diminished upon the identification of strains of MRSA and VRE that were resistant to daptomycin (15)[44].

Only a handful of antibiotics have been approved by the US FDA for clinical use in the past few years (see Table 2 and Figure 2) [45–47]. Worryingly, these newer antibiotics (with the exception of retapamulin) are merely derivatives of older generation antibiotics. Since the newer antibiotics are acting on the same targets, which have been shown to be evolvable, it would be naive to think that the newly introduced antibiotics will not ultimately suffer from bacterial resistance.

Retapamulin (21), introduced in 2007, belongs to a new class of antibiotics called pleuromutilins. Retapamulin inhibits protein biosynthesis by binding to the peptidyl transferase center of the 23S rRNA, which is located on the 50S subunit and, thus, prevents peptide bond formation [48]. This mode of action differs from the macrolide protein inhibitors such as erythromycin (7) and azithromycin, which also bind to the 23S rRNA segment of the 50S subunit, but do not interrupt the peptidyl transferase [48]. Retapamulin (21) may also destabilize tRNA in the P-site of the 50S subunit [48]. However, bacteria have shown in the past that ribosomal targets can be modified in order to abrogate antibiotic inhibition and there is no reason to suggest that the 50S ribosome is not evolvable.

The previous discussion highlights an unsavory truth: we are in a never-ending battle against pathogenic bacteria. The ability of bacteria to find several pathways to render a bacteriocidal or bacteriostatic antibiotic ineffective, coupled with the lack of newer druggable targets in bacteria and the paucity of new structural classes of antibiotics, implies that there might be a need for a paradigm shift in the strategies used to treat bacterial infections.

Over the last few years, researchers have begun to focus attention on finding ways to render pathogenic bacteria less virulent with small molecules. It is assumed that if one can find ways to prevent bacteria from being virulent (i.e., inhibit toxin production) without necessarily killing the bacteria, then there will be less evolutionary pressure for the bacteria to evolve resistance mechanisms. However, this presupposes that toxin production by bacteria does not confer a competitive advantage. Since biofilms are difficult to treat with conventional antibiotics, it is also expected that small molecules that can either inhibit biofilm formation or disperse an existing biofilm will find utility in combination antibiotic therapy. Studies from several laboratories have unraveled the complexities of the signal transduction pathways that lead to both toxin production and biofilm formation in bacteria. It is now known that the majority of bacteria toxin production is quorum sensing (QS)-controlled, vide infra[49]. Futhermore, a dinucleotide known as cyclic diguanylic acid (c-di-GMP) appears to be a master regulator of diverse bacteria strains [50,51]. In this article, the various signaling transduction processes in bacteria that result in virulence production or biofilm formation are briefly discussed and the small molecules that have been used to attenuate bacterial virulence as well as biofilm formation will be examined. Space limitation prevents us from discussing all of the molecules that have been shown by several groups to make bacteria avirulent. Therefore, the molecules that are chosen in this review are merely representative examples and are in no way exhaustive.

Quorum sensing

Quorum sensing is a form of bacterial cell-to-cell communication whereby bacteria secrete and detect signaling molecules known as autoinducers. The detection of a threshold concentration of autoinducers induces the activation/suppression of genes responsible for processes such as virulence gene expression [52–60], biofilm formation [61–64] and antibiotic resistance [65]. Different QS circuits have been identified. Some bacteria use multiple QS circuits to monitor their environment. Figure 3 illustrates six general QS systems described to date [66–72].

Bacteria use several different classes of molecules for cell-to-cell communication (Figure 4). Several Gram-positive bacteria use peptides (25–28) for intraspecies communication [73] whereas many Gram-negative bacteria utilize acylhomoserine lactones (AHL; also called AI-1; 29–39) [56,66,74] for intraspecies communication. Other molecules, such as g-butyrolactones (43), quinolones (40) and hydroxy ketones (42), are also employed as intraspecies communication molecules in Streptomyces[75], Pseudomonas aeruginosa[76] and Vibrio cholerae[70,71], respectively. AI-2 (collective name for the family of interconverting cyclic lactols 44–46; Figure 4) has been described as being used by bacteria for interspecies communication [67]. AI-2 has been found in several species of bacteria [77,78] and is derived from 4,5-dihydroxyl-2,3-pentadione (DPD; 47) [79,80]. Although over 70 bacterial strains have been shown to harbor the enzyme that makes AI-2, known as LuxS, only a handful of bacterial species have been shown to possess receptors that sense and respond to the presence of AI-2; the LuxP receptor in Vibrio harveyi and LsrB in E. coli/Salmonella bind to different forms of AI-2; LuxP binds to furanosyl borate diester [79] whereas LsrB binds to (2R,4S)-2-methyl-2,3,3,4-tetrahydroxytetrahydrofuran (R-THMF), which lacks boron [80]. Another protein that binds ‘processed’ AI-2 is the LsrR transcriptional regulator in E. coli/Salmonella that binds to phosphorylated AI-2 [67,81]. Despite the paucity of identified bona fide AI-2 receptors in bacteria, the role of this molecule as a cell-to-cell signaling molecule and mediator of virulence production in some bacteria is now established. For example, in Vibrio vulnificus[82,83], Serratia marcescens[84] and Clostridium perfringens[85], AI-2 has been implicated in the production of virulence factors. However, the receptors that mediate virulence-factor production in the aforementioned bacteria have yet to be fully characterized. In other bacteria, LuxS (the synthase for AI-2) might have an additional role in the metabolic pathway involved in methionine metabolism [77,86]. The LuxS-independent formation of AI-2 from ribulose-5-phosphate has been proposed [87–89]. The significance and implications of alternative AI-2 synthesis in QS remains to be fully elucidated.

Bacteria can also sense molecules produced by eukaryotic host cells and regulate genes in response to molecules produced by the host. For example, enterohemorrhagic E. coli (EHEC) serotype O157:H7, which is responsible for outbreaks of bloody diarrhea and uremic syndrome throughout the world, responds to the eukaryotic hormone epinephrine. β- and α-adrenergic antagonists have been shown to inhibit the bacterial response to this hormone [68]. EHEC and several other bacteria also produce an autoinducer known as AI-3 [69], whose structure still remains to be elucidated.

▪ QS in a select few human pathogens

Quorum sensing systems in Pseudomonas aeruginosa, Staphylococcus aureus, Vibrio cholerae, Salmonella typhimurium and Escherichia coli have been of interest to several researchers due to numerous common diseases that are attributed to these opportunistic bacteria. The QS circuitry in P. aeruginosa, a Gram-negative bacteria that thrives in many environments, has been well studied. P. aeruginosa can cause infections of the respiratory, urinary and GI tracts, as well as keratitis and bacteremia. Cystic fibrosis, burn or immunocompromised patients are particularly susceptible to P. aeruginosa infections. Two major QS systems in P. aeruginosa are the las and rhl systems [90]. The las and rhl QS systems are composed of transcriptional regulators that respond to autoinducers produced by LasI and RhlI synthases. LasI produces 3-oxo-C₁₂-HSL (35; Figure 4), which is sensed by LasR, whereas RhlI produces C₄-HSL (29), which is sensed by RhlR. The las and rhl QS systems are organized in a hierarchical manner; LasR: 3-oxo-C₁₂-HSL complex regulates the expression of the rhl system. Also, it is believed that the las system regulates the rhl system post-translationally. This is achieved via the competitive binding of 3-oxo-C₁₂ HSL to RhlR at low C₄-HSL concentration [90]. A transcriptional regulator, QscR represses the transcription of lasI and rhlI as well as lasB, pqsH, hcnAB and phenazine genes, phzA and phzA2[91,92]. P. aeruginosa also possesses a third QS molecule known as the Pseudomonas quinolone signal (2-heptyl-3-hydroxy-4-quinolone (40)) [76]. Microarray analysis has shown that a few hundreds of genes are under QS control in P. aeruginosa[93–95]. These genes regulate diverse toxic factors such as elastase (lasB) [96], LasA protease (lasA) [97], alkaline proteases (apr) [98] and secretion pathway (xcpP and xcpR) [99]. A fourth class of molecules found in P. aeruginosa is the dipeptide-derived piperazines (41)[100]. P. aeruginosa may use these piperazines to communicate with other bacteria such as P. mirabilis, C. freundii and E. agglomerans. More work is needed to confirm whether dipeptide-derived piperazines are indeed QS signaling molecules [101].

S. aureus is an opportunistic pathogen and is the major cause of postoperational infections as well as pneumonia, acute endocarditis and abscesses of the skin [52–54]. The accessory gene regulator (agr) QS system in S. aureus and other staphylococcal species plays a central role in the regulation of virulence factors during staphylococcal pathogenesis. The agr locus consists of the P2 operon (agrACDB) and the P3 operon, whose transcript, RNAIII encodes α-toxin (hld), in addition to having regulatory effects on agr expression [102]. In the agr QS system, the autoinducing peptide (AIP), consisting of seven to nine residues, is processed from AgrD protein via AgrB-mediated thiolactone ring formation between a conserved central cysteine and a C-terminal carboxyl group [53]. It is plausible that other proteins apart from the membrane bound AgrB might be involved in AgrD processing. The secreted AIP is recognized by the membrane-bound AgrC and triggers the transfer of a phosphate group between AgrC and AgrA. AgrA together with other proteins such as the transcriptional regulator SarA activates the transcription of the P2 and P3 operons [103]. The agr system regulates an array of virulence factors. These include entereotoxin B and C, proteases (splA, B, D, F and V8) and cell-surface associated capsular polysaccharides [104,105]. Interestingly, the AIP produced by one of the four distinct agr groups generally inhibits signaling by another Staphylococci that harbors a different agr group. The inhibition occurs via competitive binding to the AgrC receptor. For example, an AIP produced by S. epidermidis can inhibit all but type IV of S. aureus agr groups [52]. Although recent evidence suggest that agr expression in vivo is more complex than originally thought and might depend on several environmental factors [106], it is clear that inhibition of the agr QS system will attenuate S. aureus disease progression and persistence. This is illustrated in Figure 5. In Figure 5A, a mouse was infected with S. aureus that lacked the AgrC-1 receptor, whereas in Figure 5B, the mouse was infected with S. aureus that was agrC-1⁺. The mouse that was infected with S. aureus (agrC-1⁺) developed an abscess that was significantly larger than the mouse that was infected with S. aureus (agrC-1^–). Furthermore, when the mouse was infected with S. aureus (agrC-1⁺) and then treated with AIP-II (25b), the abscess that developed was similar to the agrC-1^– infection case. This showcases that the inhibition of one group of agr by AIP produced by another group also occurred under in vivo conditions. Other Gram-positive bacteria also use peptides for QS: E. faecium uses GBAP (28)[73], S. pneumoniae uses competence stimulating peptide (CSP; 26) [107] and B. subtilis uses competence and sporulation factor (CSF) and ComX (27)[66,107,108].

Another human pathogen, V. cholerae causes the deadly disease cholera [70]. V. cholerae is a Gram-negative bacterium and uses two signaling molecules for QS: (S)-3-hydroxytridecan-4-one (or C-AI-1 42) and AI-2 (44–46) [71]. (S)-3-hydroxytridecan-4-one (42) is the major QS molecule used in V. cholerae and it is structurally unique [71]. Compound 42 is a 13-carbon hydroxyl ketone (Figure 4) and is synthesized by CqsA and detected by CqsS [71,109]. Ironically, the QS circuitry is reversed in V. cholerae; expression of virulence and biofilm formation is activated at low cell density and repressed at high cell density [70,71]. The repression of biofilm formation and virulence expression at high cell density allows the bacteria to leave the host and be re-introduced into the environment [70,71].

Salmonella typhimurium is the leading cause of food-borne illnesses worldwide and causes gastroenteritis. Although S. typhimurium poisoning (‘food poisoning’) often clears on its own, it can be fatal in immunocompromised patients. In addition, biofilm formation on abiotic surfaces (e.g., restaurant kitchen counters) is a major cause of the spread of diseases. QS in S. typhimurium utilizes AHLs and AI-2 and their receptor proteins SdiA and LsrB, respectively [58–60]. SdiA detects acylhomoserine lactones, although S. typhimurium does not produce AHLs on its own [58–60]. The detection of AHLs is thought to be a defense mechanism that S. typhimurium uses to survive in environments with other diverse pathogens [58–60]. In S. typhimurium and E. coli, when a threshold concentration of AI-2 is reached, LsrB transports the molecule inside the cell where it is phosphorylated by LsrK kinase [67]. Phospho-AI-2 (48; Figure 6) binds and de-represses the LsrR repressor protein, which controls the lsr operon [67,81]. Ultimately, LsrG protein degrades phospho-AI-2 (48) into phosphoglycolic acid (PG; 49) and an unknown C3 compound (Figure 6)[67]. It remains unclear why Salmonella and E. coli import and subsequently degrade AI-2, although it is plausible that AI-2 transport and degradation is a defense mechanism to quench the AI-2 signal of other bacteria [110].

▪ Interception of QS signaling:

If QS is vital for the production of bacterial virulence, then it is reasonable to suggest that any of the following strategies will attenuate bacterial virulence production:

• ▪ Inhibition of enzymes that catalyze autoinducer synthesis;

• ▪ Inhibition of receptors (both cell surface and cytosolic) that bind to autoinducers or processed autoinducers;

• ▪ Inhibition of downstream proteins that link QS receptor occupancy to activation of QS-regulated genes via phosphorelay mechanisms;

• ▪ Inhibition of virulence factors that are expressed in response to QS.

In this article, we will discuss the nature of the small molecules that have been shown to inhibit QS at the various stages. Owing to space limitations, we are unable to list all of the chemical entities that have been shown to antagonize QS in pathogenic bacteria. We have only selected examples that have reasonable potency in attenuating QS processes. The majority of small molecules with EC₅₀/IC₅₀ values greater than or equal to 50 µM are excluded from this article.

▪ Inhibition of autoinducer biosynthesis

Acyl-homoserine lactones used for signaling in Gram-negative bacteria are synthesized by the reaction of the acyl group located on an acyl-carrier protein with S-adenosyl-methionine (50)[111]. Briefly, the acyl group on the acyl-carrier protein is transferred to the amino functionality in SAM (50). An intramolecular nucleophilic substitution reaction involving the carboxylate moiety of SAM and the thioether as a leaving group leads to the formation of AHLs and methylthioadenosine (51)[111]. The nucleosidase, Pfs, transforms methylthioadenosine (51) into methylthioribose (52) by cleaving the adenosine group. Methylthioribose is further degraded into methionine (53), which is recycled into the SAM biosynthesis pathway (Figure 7A). AI-2 biosynthesis also uses SAM as its starting point. In AI-2 biosynthesis, a methyltransferase removes the methyl group from SAM; converting it to S-adenosylhomocysteine (SAH; 54) [112]. Similar to the AHL synthesis, Pfs then cleaves the adenosine group of SAH (54) to form S-ribosylhomocysteine (SRH;55) [112]. Finally, LuxS converts SRH into DPD (47) and homocysteine (56; Figure 7B) [112].

Since AI-2 synthase, LuxS, has been found in over 70 bacterial species [77,78], inhibitors of LuxS could have broad-spectrum anti-QS effects. On the other hand, AI-1 synthases are species specific, therefore it might be difficult to find broad-spectrum AI-1 synthase antagonists. Attempts to utilize SAM analogs to inhibit AI-2 synthesis have been described [113], but it is unclear if this is a viable strategy since SAM is vital for several processes in both bacterial and human cells. On the bright side, inhibition of methylthioadenosyl nucleosidase, Pfs, involved in AI-1 and AI-2 biosynthesis has shown great promise. Transition state analog inhibitors of 5´-methylthioadenosine/S-adenosylhomocysteine nucleosidase (58; Figure 8) were found to inhibit 5´-methylthioadenosine/S-adenosylhomocysteine nucleosidase with K_i values as low as 47 fM [114]. More excitingly, these analogs (58a) have been shown to be effective against biofilm formation in a variety of bacteria such as V. cholerae and E. coli O157:H7 [115]. LuxS inhibitors are also promising drugs and several have been successfully synthesized [116–120]. One of the most effective LuxS inhibitors reported to date is compound 57(Figure 8). This compound was found to inhibit LuxS of B. subtilis with a K_i value of 0.37 µM [118]. Compound 57 had less affinity for the LuxS of E. coli and V. harveyi (K_i∼13 µM for both enzymes) [118]. Although all LuxS enzymes catalyze the formation of AI-2, subtle differences amongst LuxS from different bacterial strains might preclude the use of a single chemical entity to inhibit all LuxS enzymes.

▪ Inhibition of QS by targeting receptor proteins

Several research groups have pursued the development of novel QS inhibitors that target receptor proteins that bind autoinducers [121–132]. Since QS receptors might not be found in humans, targeting these proteins could be less cytotoxic than other anti-QS strategies. It is, however, plausible that eukaryotic hosts have receptors that sense bacterial signaling molecules and the receptors are yet to be found [133]. The autoinducer-binding domain of the LuxR-family of proteins resembles that of the GAF or PAS domain that is present in a number of mammalian signaling proteins and transcriptional factors [133]. AI-1 from P. aeruginosa has been shown to affect the production of cytokines by immune cells and also increase IL-8 mRNA levels in epithelial cells [134]. Several anti-QS small molecules have been developed that are structurally similar to natural autoinducers (Figure 4). Among Gram-positive bacteria, a cyclic peptide analog of AIP-I (25a), AIP-I D5A (60; Figure 9), has been shown to effectively inhibit AgrC3, while AIP-I D5N and AIP-IV Y5F were shown to inhibit AgrC4 and AgrC2, respectively, in S. aureus(Table 3)[121,122]. Several AIP analogs that lack a peptide tail have been synthesized and some of these truncated analogs act as inhibitors of AgrC. For example, trAIP-I D2A (62; truncated form of AIP-I) potently antagonizes AgrC1 (Table 3)[121,122].

The main functional groups that constitute AHLs are a lactone core, an amide moiety and an alkyl side chain. AHLs containing 3-oxo-amide side chains are also found (see compounds 35–39, Figure 4). Anti-QS small molecules based on the AHL structure with one or all of the three core motifs modified have been pursued. For example, the bromophenyl AHL compound (63; the alkyl side chain of this AHL analog contains a para-bromoethylbenzyl group, Figure 9) inhibits LasR, a LuxR-type protein found in P. aeruginosa, at micromolar concentrations [123]. Other AHL analogs, including compounds 64–70, inhibit QS-mediated processes in P. aeruginosa, A. tumefaciens or V. fischeri(Table 3)[123–126].

LuxR-type proteins bind to AHLs resulting in transcriptional response. AHL autoinducers bind to the amino-terminal module of LuxR-type proteins and, upon autoinducer binding, LuxR-type proteins dimerize and bind to specific DNA sequences near target promoters [135]. Schuster and Greenberg have proposed three different classes of LuxR-type proteins based on biochemical characterization [135]. These classes are:

• ▪ Class I proteins include LasR and CepR of Pseudomonas aeruginosa[136] and Burkholderia cenocepacia, respectively [137]. These proteins bind to AHLs in an unfolded state but, after AHL binding, class I proteins fold to bind AHL tightly;

• ▪ Class II proteins include the prototypical LuxR from V. fischeri[138] and QscR from P. aeruginosa[139]. These proteins also require AHL binding for folding but binding to the folded protein is not as tight as it is in the case of class I (i.e., AHL binding to class II is reversible);

• ▪ Class III proteins include RhlR from P. aeruginosa[140] and ExpR from Erwinia chyrisanthemi[141], which do not require AHL binding for folding. Binding of AHL to class III proteins is reversible; similar to binding to class II proteins.

Binding of AHLs to LuxR-type receptors results in the expression of virulence factors in several bacteria. For example, during infection by P. aeruginosa and Burkholderia cenocepacia (bacteria that usually coinfect patients with cystic fibrosis), QS-regulated virulence factors such as elastase (lasB), protease A (lasA), exotoxin (toxA) and alkaline protease (aprA) produced by P. aeruginosa[142], and extracellular proteases and siderophores produced by B. cenocepacia lead to fatal pulmonary decline [143]. LuxR-type receptors have also been implicated in bacterial biofilm formation. For example, compounds 63 and 64 attenuate the biofilm formation ability of P. aeruginosa(Figure 10)[132].

Despite the demonstration that AHL analogs such as 63 can successfully antagonize native AHL-induced virulence and biofilm formation in bacteria, it remains to be shown whether these analogs can survive lactonases that are capable of cleaving the lactone structure under in vivo conditions [144,145]. A recent finding by Greenberg et al. that lactonases produced in the human airway might serve as bacterial QS quenchers [146–148] cautions against the pursuit of lactone or thiolactone-based AI-1 analogs as anti-QS drugs. AHL-based analogs that have the lactone moiety replaced by cyclic structures such as cyclohexanone (69, Figure 9) and phenol (70) have also been pursued. Suga et al. have shown that analogs such as 68, 69 and 70, which are presumably lactonase resistant, can inhibit QS-regulated lasI and rhlI gene expression in P. aeruginosa[125]. Compounds 68, 69 and 70 caused more than 50% reduction in the expression of GFP reporter gene induced by 1 µM AI-1 at micromolar concentrations, providing encouragement that effective lactonase-resistant AHL-based anti-QS drugs could be developed in the future. PD12 (73) contains modifications of the lactone ring as well as the acyl side chain. PD12 (73) is a potent inhibitor of AHL-mediated QS with an IC₅₀ in the nanomolar range (Table 3)[149]. Another AHL analog that contains modification to the acyl side chain is the sulfide 67, which inhibits LasR monitor system with low micromolar IC₅₀[150].

AI-2 is important in bacterial QS but, despite its widespread utility, very few analogs based on the universal QS molecule AI-2 have been reported as anti-QS agents. Recently, the laboratories of Janda, Meijler and Sintim reported that C1-alkyl analogs of AI-2 did not have agonist activities but rather enhanced AI-2 agonism in V. harveyi’s bioluminescence (synergistic agonism) [129–131]. The C1-butyl analog of AI-2 (71) was shown to inhibit AI-2-induced β-galactosidase activity in S. typhimurium Met-844 with an IC₅₀ value of approximately 5 µM when tested against 50 µM AI-2 [129]. The exact mode of inhibition by the C1-butyl analog of AI-2 is unknown. It is, however, tempting to speculate that this particular analog inhibits LsrR. LsrR is a transcriptional repressor of the lsr operon in enteric bacteria. LsrR binds to phospho-AI-2 and this binding prevents LsrR from being a repressor [67,83].

QseC is a receptor for AI-3 and is found in EHEC [68]. Recently, high-throughput screening has identified LED209 (72; Figure 9) as a potent inhibitor of QseC [151]. Inhibition of QseC by 72 blocks the expression of virulence factors required for attaching and effacing lesions by EHEC (Figure 11).

Naturally occurring brominated furanones produced by algae and analogs (79–82, Figure 12) are inhibitors of QS and biofilm formation in a number of bacteria, including P. aeruginosa and S. typhimurium[152–154]. Compounds 79–81 are synthetic analogs of the natural furanone 82. Compound 79 inhibits P. aeruginosa biofilm formation (Figure 13)[153] and compound 80 significantly reduces the formation of abscess on mice lung tissue (Figure 14)[154]. Compound 81 is less potent than the natural furanones (e.g, 82) at reducing biofilm formation in S. typhimurium(Table 3) but is also less toxic [152]. In the S. typhimurium biofilm study, it was shown that flagellar biosynthesis was affected by brominated furanones but none of the known target genes of the AI-2 system and AHL receptor SdiA were affected. It therefore appears that brominated furanones affect biofilm formation in some bacteria by interfering with QS systems, for example in P. aeruginosa[153], whereas in other bacteria, alternative targets might be involved. The anti-QS activity of the halogenated furanones has previously been ascribed to inhibition of the transcriptional regulatory protein, LuxR-type. For example, in V. harveyi, the DNA-binding activity of LuxR is inhibited by the natural furanone (5Z)-4-bromo-5-(bromomethylene)-3-butyl-2(5H)-furanone [155]. A recent study has shown that halogenated furanones might also act to inhibit QS by inactivating AI-2 synthase, LuxS, via a covalent modification [156].

Several other inhibitors of QS have been discovered using high-throughput screening of both pure compounds and mixtures. Microarray analysis revealed that the addition of 4-nitropyridine-N-oxide (75; Figure 12) downregulates 37% of the QS regulon in P. aeruginosa including virulence factors such as lasB, rhlAb, aprA and phzAIBI[157]. Similarly, patulin (74) and penicillic acid (76) from the broth of Penicillium sp were shown to downregulate 45% and 60% of QS-controlled genes, respectively, in P. aeruginosa[158]. Similarly, GC-7 (77), an extract from garlic, inhibits LuxR but not LasR monitor systems [150]. Another extensive study looked at over 200 marine extracts in order to find potent QS inhibitors [159]. From this study, manoalide derivatives were identified as QS inhibitors. Extracts of the sponge Leucosolenia variabilis were also active QS inhibitors and inhibited lasB expression in P. aeruginosa[159]. The common food additive cinnamaldehyde (78) has been shown to inhibit AI-2 mediated QS by destabilizing the transcriptional regulator LuxR in V. harveyi[160].

▪ Macromolecules as anti-QS drugs

So far we have focused on small molecules that inhibit bacterial QS. The use of protein-based or oligonucleotide-based therapeutics has also been explored as an anti-QS strategy. It has been demonstrated that monoclonal antibodies raised against 3-oxo-C12 AHL or AIP-IV (25d) haptens were able to inhibit AHL- and AIP-mediated QS [161–164]. Monoclonal antibodies for AIP could protect mice from lethal P. aeruginosa infections [165] and inhibit biofilm formation in S. epidermidis[166]. In the natural environment, AHL lactonases (AiiA) and acylases have been shown to degrade AHL and quench AHL-mediated QS [167,168]. Stable versions of these enzymes and other QS-processing proteins could, in principle, be used as protein-based anti-QS therapeutics [169].

▪ Virulence factor inhibition

A high proportion of bacterial virulence factors are under QS control and, hence, inhibition of autoinducer synthases or receptors can reduce virulence factor production, as has been demonstrated in the preceding section of this article. An alternative strategy to curb bacterial virulence is to directly inhibit the factors that are detrimental to the host cell. These factors can be sub-divided as:

• ▪ Bacterial adhesion molecules that aid the attachment of bacteria to host cell;

• ▪ Type I–VI secretion systems that provide a channel for the delivery of toxic proteins or peptides from bacterial cells into host cells or the extracellular space;

• ▪ Toxic factors, which directly disable the normal cellular functions of the host cells.

Adhesion to host cells is usually a prerequisite for the successful establishment of infection by pathogenic bacteria. Adhesins, retractile type IV pilli or other surface macromolecules are typically used to establish host–pathogen crosstalk that leads to disease progression. Examples of adhesion molecules are the lectin-like E. coli, FimH [72], fibronectin-binding proteins A and B in S. aureus[105] and intimin in EHEC [72]. Since bacterial adhesion is mediated by interactions of carbohydrates that are present on the host cell surface, the development of carbohydrate-based anti-adhesion drugs holds promise [170,171]. Inspiration for the development of carbohydrate-based molecules as anti-adhesion therapeutics comes from the observation that the protective properties of human breast milk are derived from the presence of oligosaccharides [172] . For an excellent review on the anti-adhesive properties of milk oligosaccharides, see elewhere [172,173].

Compound 85 is a polyvalent inhibitor of anthrax toxin [173]. It blocks the host receptors that are used by B. anthracis to gain entry into the host cell. Compound 85 represents an important antibiotic class that might not suffer from bacterial resistance. This is because extensive alterations to the pathogen’s toxin has to be made in order to switch to another host receptor [173]. Synthetic carbohydrate-based compounds such as 84(Figure 15) have been shown to be effective inhibitors of fimbriae and FimH-mediated adhesion to guinea pig intestinal cells with 700-times higher affinity than methyl-mannoside [174]. Compound 83 is also carbohydrate based and it has been shown to inhibit the colonization of H. pylori in the GI tract of monkeys [175]. Carbohydrate-based compounds found in cranberry juice inhibit fimbriae-facilitated adhesion of E. coli to the urinary wall [176]. Other noncarbohydrate-based compounds such as 86(Figure 15) have been shown to be effective anti-adhesive compounds [177]. Compound 86 inhibits sortase A with a K_i^app of 0.3 µM. Sortase A is used by bacteria, such as S. aureus, for attachment to host cell wall [177]. The pilicide BibC10 (87) inhibits the formation of pili and surface adhesion fibers, in ureopathogenic E. coli[178,179]. BibC10 (87) blocks the usher–chaperone pathway by binding the chaperone, PapD [178,179].

Bacterial secretion systems are elaborate apparatus or systems that allow bacteria to secrete diverse molecules such as proteins, oligosaccharides and nucleic acids into the extracellular media or directly into host cells. These secretory apparatus thereby play important roles in bacterial pathogenesis. There are six well-known secretion systems in Gram-negative bacteria [180]. These systems secrete proteins either in a single step or with initial export into the periplasm via the Sec pathway and then use other machinery to cross the outer membrane [180]. The Sec system recognizes proteins for transport via a hydrophobic N-terminus signal sequence and then uses ATP to move the protein across the plasma membrane [180]. Type I secretion system (T1SS) is composed of the ABC transporter protein complex, membrane fusion protein and outer membrane protein (OMP). T1SS is important for the secretion of diverse biomolecules ranging in sizes (10–900 kDa) and substrate type (from proteins to polysaccharides). Several important bacterial toxins, such as the pore-forming RTX toxin hemolysin, which kill human immune and other cells are secreted to the exterior via T1SS [180]. The type II secretion system utilizes the Sec system for entry into the periplasm [181]. Once in the periplasm, proteins pass through the outer membrane via a secretion apparatus that is made of a multimeric complex of secretin and other membrane proteins. The type III secretion system (T3SS or TTSS) acts like a needle that penetrates and injects proteins into the host cell [182]. Type IV secretion system (T4SS or TFSS) is capable of transporting DNA and proteins into host cells [183]. Several pathogenic bacteria use the T4SS to deliver toxins such as pertussis toxin (from Bordetella pertussis) and CagA (from Helicobacter pylori). Type V secretion system (T5SS, also known as autotransporter system) also uses the Sec system to cross the membrane. Proteins, which use the type V secretion system, form a β-barrel with the C-terminus and inserts into the outer member [184]. The type V secretion proteins are then cleaved and the β-domain is released [184]. The type VI secretion system proteins (T6SS) presumably does not use the Sec system as they lack N-terminal signal sequences. It is now believed that T6SS are widespread in Gram-negative bacteria [185,186].

Quorum sensing has been implicated in the formation of bacterial secretion systems [187–189]. Therefore, general inhibitors of the QS circuitry might also inhibit bacterial secretion systems. Imine 89, at micromolar concentrations, has been shown to be an effective inhibitor of T3SS in enteropathogenic E. coli[188]. Compound 89 inhibits T3SS without affecting motility and flagellin expression in enteropathogenic E. coli[188]. Other T3SS inhibitors include compounds 90 and 91, which inhibit T3SS secretion at micromolar concentrations in Yersinia pseudotuberculosis[190]. The exact molecular targets of 89–91 are poorly characterized and, although they are not direct inhibitors of T3SS, they inhibit the expression of virulence-associated factors.

Compounds that inhibit transcription regulators of toxin genes or that inhibit the toxins directly have also attracted interest. For example, virstatin (93), identified via a high-throughput phenotypic screen, inhibits the transcription regulator ToxT in V. cholerae[191]. By inhibiting ToxT, virstatin (93) prevents the expression of cholera toxin and toxin coregulated pilus. These toxins are responsible for secretory diarrhea and V. cholerae attachment to human intestinal wall [191]. Compound 88 and peptoid 92 have been shown to bind to shiga [192] and cholera toxins [193], respectively. Shiga toxin, produced by Shigella dysenteriae type 1 and Shiga-like toxins from E. coli can cause serious illness to humans. These toxins belong to the bacterial AB5 toxins, which are made up of an enzymatic A subunit that is responsible for the toxin’s action and a homopentamer B5 subunit that is involved in oligosaccharide recognition of the host cells [194].

RNAIII-activating protein is responsible for a-toxin and serine protease expression in S. aureus[105]. The RNAIII-activating protein inhibitor peptide (RIP) inhibits RNAIII-activating protein and prevents biofilm formation and a-toxin and serine protease expression in S. aureus[195]. However, the potential for enzymatic degradation of RIP by peptidases in vivo could limit the therapeutic potential of RIP [195]. Hamamelitannin (95) also inhibits RNAIII-activating protein [195]. It has, however, not been demonstrated whether the ester functionalities in 95 will survive in vivo esterases. Compound 94, a metabolite extracted from Flustra foliacea, suppresses AHL-dependent production of proteolytic enzymes in P. aeruginosa at micromolar concentrations [196].

ClpP proteases have been identified as important virulence factors in S. aureus[197]. Clp proteolytic complexes degrade accumulated and misfolded proteins during pathogenesis by S. aureus[198]. Compound 96 has been identified as a micromolar inhibitor of ClpP and effectively prevents haemolysis and proteolysis in S. aureus[197].

▪ Biofilm inhibitors

Over 65% of hospital infections are caused by biofilm-forming bacteria [62–64]. Biofilms are micro-organisms embedded in a matrix and attached to surfaces [62–64]. Once incorporated into the biofilm community, bacteria are resistant to antibiotic action [62–64]. In addition, the biofilm environment increases the probability of antibiotic-resistant plasmid transfer between bacteria [62–64]. Despite the central role that bacterial biofilm plays in infection, there is currently no antibiofilm drug in clinical use. This is probably due to the fact that the mechanisms that underline biofilm formation are not completely understood. There is a paucity of small molecules that can inhibit biofilm or disperse mature biofilm.

Several of the QS inhibitors, such as brominated furanones (79–82), have been shown to inhibit biofilm formation [152–154]. AHL-analogs (63, 64, 69 & 70)[132] and hamamelitannin (95)[195] have also been shown to be effective in inhibiting biofilm formation. Hamameltannin inhibits the QS regulator RNAIII and shown to reduce biofilm formation in MRSA and methicillin-resistant S. epidermidis strains [195]. Other small molecules that reduce biofilm include baicalein (97), which is a natural product shown to inhibit P. aeruginosa biofilm formation at 20 µM [199]. Baicalein promoted proteolysis of TraR protein in E. coli, strongly suggesting that its biofilm-inhibition properties might be due to QS inhibition [199]. With the exception of few biofilm inhibitors including AHL analogs (63, 64, 69 & 70) and the brominated furanones (79–82), whose targets are known, the majority of biofilm inhibitor that have been discovered so far have unknown targets. These include 2-aminoimidazole/triazole (98)[200,201], several structurally diverse small molecules capable of inhibiting Pseudomonas aeruginosa biofilm formation (99–104; Figure 16)[202], SB2 (105) and ureidothiophene (106)[203]. 2-aminoimidazole/triazole (98) is a simple derivative of the marine natural product bromoageliferin, which can disperse established proteobacterial biofilm as well as inhibit biofilm formation [200,201]. 98 was able to disperse S. aureus biofilm with an IC₅₀ of 2.6 µM. The mechanism of biofilm formation or dispersion remains to be elucidated. (Z)-2-decenoic acid (107) is also a potent biofilm-dispersion inducer, being able to completely disperse P. aeruginosa biofilm at a concentration of 2.5 nM [204].

In an attempt to increase the pool of small molecules that can inhibit biofilm formation, the Clardy group developed an efficient high-throughput screen to identify small-molecules that are effective against P. aeruginosa biofilm development [202]. A total of 66,095 compounds were screened via a luminescence-based biofilm assay for both attachment and detachment of bacteria to solid surfaces. Several small molecules belonging to different structural classes were identified as effective biofilm formation inhibitors see Table 4.

SB2 (105) and ureidothiophene (106) are antibiofilm molecules that inhibit transcription in vitro against S. epidermidis[203]. It has been suggested that their antibiofilm activities may be connected with their action as RNA polymerase inhibitors but this claim has yet to be validated [203].

In an unexpected twist, recent studies have shown that known drugs such as azithromycin [205], ceftazidime and tobramycin [206] inhibit biofilm formation at subminimal inhibitory concentration. It remains to be seen if the antibiofilm activities of these established drugs are derived from their known targets, 23S rRNA (tobramycin), 23S rRNA (azithromycin) and bacterial cell wall (ceftazidine).

▪ New targets of antibiotic therapy

Recent efforts on antibiotic discovery have focused on new bacterial target proteins that have important roles in viability. Several RNA polymerase inhibitors including lipiarmycin (116; Figure 17) are currently in clinical trials [203]. Other new antibiotic targets include LpxC, an enzyme that is involved in the synthesis of lipopolysaccharide, Lipid A [207,208]. A few inhibitors of bacterial lipid biosynthesis are in clinical trials including Rs-DPLa (110), eritoran (117) and CHIR-090 (111; Figure 17)[207,208]. Type II fatty-acid synthesis proteins FabF and FabH are also important antibiotic targets. Platensimycin (108)[209] and platencin (109)[210], novel antibiotics produced by Streptomyces platensis, are nanomolar FabF inhibitor and FabF/H dual inhibitor, respectively. Simplified analogs thereof are expected to become useful next generation antibiotics [211–214]. However, a recent paper has questioned the validity of using type II fatty-acid synthesis inhibitors as effective antibiotics [215]. The FtsZ protein, which is homologous to tubulin and is important for prokaryotic cell division [216], has recently been shown to be a drugable target. It has been demonstrated that compound 113(Figure 17) can inhibit FtsZ protein leading to bacterial cell death [216,217]. RecA is an essential enzyme required in bacterial cells for the repair and maintenance of DNA [218]. The small molecule, suramin (112) has been shown to inhibit RecA’s ATPase activity [219]. Thiopeptide antibiotics, such as thiostrepton (119; Figure 17) and mirococcin, have also drawn the attention of medicinal chemists and biochemists. These thiopeptide antibiotics bind to the L11 domain of 23S ribosomal RNA [220] but, because of their poor solubility, none of the antibiotics in this family have found clinical use. The synthesis of soluble analogs of thiostreptin (119) may allow clinical application in the future [221,222]. Similarly lantibiotics, a class of antibiotics that are ribosomally synthesized and contain the unusually lanthionine amino acid residue, have bacteriocidal effects. This is due to the ability of lantibiotics to inhibit peptidoglycan formation [223]. Examples of lantibiotics include nisin (118; Figure 17) and epidermin, along with many others. Recent interest in the lantibiotics is due to the fact that they have been used in food preservation for decades without any observed resistance emerging. Nisin (118) is not currently in clinical use owing to the fact that it is degraded in the human digestive system [224]. The development of analogs, which are stable in the human intestine and gut, could allow these compounds to have clinical use in the future [225,226]. Inhibition of peptide deformylase is another novel strategy for antibiotic therapy, owing to its importance in protein synthesis [227]. The natural product actinonin (115) has been identified as an inhibitor of peptide deformylase [227] and other synthetic compounds that inhibit deformylase have been designed and are in clinical trials. Furthermore, inhibition of teichoic acid biosynthesis has been targeted through the inhibition of LtaS, the lipoteichoic acid synthase. Teichoic acid is a major component of Gram-positive bacteria cell walls. Teichoic acid attached directly to the cell wall (wall teichoic acid) or linked to membrane lipids (lipoteichoic acid) have been shown to be vital for bacterial survival.

▪ Novel c-di-GMP signaling presents new targets for novel antibiotic developments

In the coming decades, as more information from bacterial genome sequencing becomes available, it is anticipated that new proteins that are involved in bacterial regulatory pathways will be unveiled. The exact details of transduction processes such as c-di-GMP signaling will also be uncovered. c-di-GMP (Figure 18) has emerged as an important intracellular signaling molecule uniquely present in bacteria [228].

Since first being discovered as a cellulose synthase activator in Gluconacetobacter xylinus (formally known as Acetobactor xylinum) by Benziman over two decades ago [229], c-di-GMP has been shown to regulate diverse cellular processes, including the expression of virulence factors and genes that are important for the formation and maintenance of biofilm [51]. In addition, some advances have now been made regarding the elucidation of the roles of QS and c-di-GMP signaling in biofilm formation [230].

As c-di-GMP signaling and QS regulate the same complex processes, including biofilm formation and virulence, it has been argued that these two signaling pathways may be linked [149]. Accumulating evidence shows that c-di-GMP and QS signaling intersect [231]. For example in V. cholerae, QS autoinducers control the expression levels of small regulatory RNAs known as Qrr sRNAs [232]. These sRNAs are involved in an Hfq-mediated degradation of target mRNAs such as HapR (a master transcriptional regulator). Recently, the Bassler’s laboratory has shown that Qrr sRNA also disrupts an inhibitory stem loop structure of vca0939 mRNA [233]. This mRNA translates into an enzyme that forms c-di-GMP. Additionally, in V. cholerae a transcription factor known as AphA, which is regulated by QS, influences the expression of DGCs and PDEAs[234]. Recently, Wood showed that TpbA, a tyrosine phosphatase that is controlled by LasR, regulates biofilm formation by inhibiting intracellular c-di-GMP concentrations [230]. The exact details of c-di-GMP signaling is currently being studied by several laboratories and it is expected that analogs of c-di-GMP [235] or other small molecules that will inhibit the proteins that are involved in the biosynthesis of c-di-GMP will become useful as either antivirulence or antibiofilm drugs.

Future perspective

Appreciation of the amazing ability of bacteria to evolve mechanisms that render bacteriocidal or bacteriostatic antibiotics ineffective has led to a re-evaluation of how to develop new antibiotics and manage bacterial infections. In the last decade our understanding of signaling processes in bacteria (QS and c-di-GMP signaling, which leads to bacterial virulence factor production and biofilm formation) has increased. This development, coupled with the recent discovery that all bacteriocidal antibiotics lead to production of hydroxyl radical and oxidative damage (Figure 19)[236], has rekindled a belief that the antibiotic resistance problem might not lead to a plague in the future. Judging by the flurry of activities in the antibiotic research field, there is every hope that the new paradigm shift in antibiotic discovery will result in effective next-generation antibiotics. In the coming years, it is anticipated that several receptors for the universal QS molecule, AI-2, will be identified and biochemically characterized. Additionally, the structure of AI-3 is expected to be solved and more structural data for Lux-R type receptors that bind AHLs are also expected to be obtained. Furthermore, the adaptor proteins and RNA molecules that bind to c-di-GMP and transmit the binding event into virulence factors expression or biofilm formation are more likely to be identified and biochemically characterized. This projected wealth of information will give the medicinal chemist a powerful platform for developing the next generation antibiotics that aim to curb bacterial virulence instead of killing bacteria. Most of the anti-QS agents reported are experimental, and comprehensive pharmacokinetic data on these molecules are lacking. More anti-QS drugs that do not have in vivo instability issues and are drug like are desired. The demonstration that AI-1 and analogs thereof have immunological profiles [237] calls for a full scientific investigation of the immune suppressive activities of autoinducer analogs using relevant disease models. It is very plausible that a decade from now, bacterial infections will be treated with combination therapeutics consisting of multiple chemical entities that will function as antibiofilm, antivirulence, bacteriocidal or bacteriostatic and antiresistance. The real challenge will be finding a cocktail of such drugs that do not suffer from unwanted drug–drug interactions.