Correspondence: Birgit M. Prüβ, Department of Veterinary and Microbiological Sciences (7690), North Dakota State University, PO Box 6050, Fargo, ND 58108-6050, USA. Tel.: +1 701 231 7848; fax: +1 701 231 9692; e-mail: email@example.com
Current antibiotics continue to lose effectiveness for infectious diseases, especially in cases where the bacteria from a biofilm. This review article summarizes control mechanisms for bacterial biofilm, with an emphasis on the modification of signal transduction pathways, such as quorum sensing and two-component signaling, by externally added metabolic intermediates. As a link between central metabolism and signal transduction, we discuss the activation of two-component response regulators by activated acetate intermediates in response to signals from the environment. These signals constitute ‘nutrients’ for the bacteria in most cases. Depending on the identity of the nutrient, biofilm amounts may be reduced. The nutrient may then be used for the development of both novel prevention and treatment options for biofilm-associated infectious diseases.
Bacterial biofilm is a complex aggregation of single species or multi-species bacterial communities that forms on a solid substrate, at a liquid–air interface or sometimes even intracellularly. This phenomenon has major impacts in clinical, environmental, agricultural and bio-industrial settings. With this review article, we will cover recent advances in biofilm research, with an emphasis on control mechanisms that will reduce biofilm in clinical environments. We will focus on the enterobacterium Escherichia coli and provide examples of other bacterial species as needed.
The old problem
Biofilm and bacterial infectious disease
The CDC and NIH estimate that 60–80% of all human bacterial infections involve biofilm. Many of the infectious diseases that Escherichia coli is associated with and that are typically attributed to specific pathotypes of E. coli are exacerbated by the formation of biofilm. Table 1 summarizes examples of these pathotypes and the biofilm-associated diseases they can cause. UPEC stands for uropathogenic E. coli, a pathotype that causes infections of the urinary tract, often acquired in a hospital in connection with long-term catheterization. The adherence of UPEC to the host cells is mediated by short adhesive fibers, such as curli (Wu et al., 2012) and type I fimbriae (Mulvey et al., 1998). Intriguingly, UPEC can form intracellular biofilm-like structures in the host cell cytosol which increase the persistence of the chronic illness (Anderson et al., 2003; Goller & Seed, 2010). EaggEC, enteroaggregative E. coli, adhere tightly to one another, in part by means of AafA fimbriae, which have been targeted by nitazoxanide to inhibit biofilm formation and hemagglutinin production (Shamir et al., 2010). STEC is a classification that combines all E. coli that are capable of Shiga toxin production. Typically, the STEC genome includes the LEE pathogenicity island (locus of enterocyte effacement). An interesting STEC variant are LEE-negative E. coli that can cause hemolytic uremic syndrome by means of the Sab autotransporter, contributing to biofilm formation and adherence (Herold et al., 2009). As a final example, a number of interesting variants of E. coli have recently been identified in periprosthetic joint infection. These were all deficient in typical characteristics of E. coli, such as the production of ß-galactosidase, flagella, indole and the resistance towards aminoglycosides (Sendi et al., 2010).
Table 1. Examples of biofilm-associated infectious diseases caused by Escherichia coli
E. coli pathotype
Major virulence factors
Uropathogenic E. coli (UPEC)
Urinary tract infections
Curli, type I fimbriae, K-capsule
Wu et al. (2012), Goller & Seed (2010), Mulvey et al. (1998)
Since biofilm contributes to numerous different bacterial infectious diseases, the need for novel biofilm control mechanisms persists. In the remainder of this article, we present a selection of control mechanisms that were developed for bacterial infectious diseases (see section below). We then advance to recent developments in interfering with the bacterial signal transduction pathways. Instead of killing the bacteria, this would manipulate them into behaviors and phenotypes that are less harmful to us. The two mechanisms that are discussed are quorum sensing (QS) and two-component signaling (2CSTS), as well as one example of a 2CSTS that contributes to QS. As one link between central metabolism and signal transduction, acetate metabolism is discussed, together with the activation of 2CSTS response regulators.
Controlling bacterial infectious disease
Conventional antibiotics fail to work on biofilm-associated infectious diseases due to a number of bacterial phenotypes that are specific for the biofilm environment. Understanding the regulation of these phenotypes may offer novel approaches to reducing biofilm, often as a combination treatment with a conventional antibiotic. As one example of such a phenotype, the matrix is composed of many substances, among them extracellular DNA. Treating the biofilm with DNAse greatly enhances the effect of the antibiotic against many bacterial species, including E. coli (Tetz et al., 2009). Another mechanism that increases antibiotic tolerance in biofilms are so-called persister cells that constitute a dormant phenotype and exhibit a much reduced energy metabolism (Lewis, 2012). Among the few techniques that are currently available for the eradication of E. coli persister cells are cationic membrane-penetrating peptides, which will then enhance the susceptibility of the biofilm to ofloxacin (Chen et al., 2011) and aminoglycosides that have improved the treatment of chronic infection in a mouse model (Allison et al., 2011). One other example of an antibiotic resistance mechanism would be plasmid exchange (Sorensen et al., 2005), including the transfer of antibiotic resistance plasmids (Krol et al., 2011) between different bacterial species (Habimana et al., 2010). The increase in genetic variation brought on by the inclusion of multiple different species of microorganisms increases biofilm robustness, as well as increasing its ability to cope with host immunity and antibiotic treatments (Rickard et al., 2003). The following sections introduce QS and 2CSTS as two examples of bacterial signal transduction pathways whose modification by external signals is currently being proposed to control E. coli biofilm.
Signal transduction as a biofilm control mechanism: quorum sensing (QS)
Quorum sensing (QS) is defined as intercellular communication, permitting bacteria (and some lower eukaryotes) to coordinate shifts in gene expression as a community as a whole (Fuqua et al., 1994). QS is a control mechanism of many pathological processes, including biofilm formation (Dickschat, 2010) and toxin production (Dobretsov et al., 2009). Because of these factors, QS is considered a virulence factor (Lim et al., 2012). As a consequence, QS offers ample opportunity as a drug target mechanism (Njoroge & Sperandio, 2009; Raina et al., 2009; Bassler, 2010).
The first QS system was described in Vibrio fischeri (Fuqua et al., 1994), where QS is permitted by an acyl homoserine lactone (AHL) type autoinducer (AI) produced by LuxI. The AI can permeate the membrane and gets released into the environment. As the population increases, so does the extracellular AHL concentration. Eventually, the AI diffuses back into the cells and activates the transcription of the luxICDABEG operon through LuxR. The bioluminescence produced as the final outcome of this regulatory mechanism enables the bacterium to fulfill its role in the symbiotic relationship with its squid host.
Quorum sensing (QS) contributes to the production of virulence factors and toxins by several pathotypes of E. coli, such as avian pathogenic E. coli (APEC) (Han et al., 2013) and STEC (Pacheco & Sperandio, 2012), as well as to biofilm production (Kim et al., 2012). Escherichia coli does not synthesize AHL. It does, however, have a LuxR homolog in SdiA (Smith et al., 2011) that acts as a strong repressor of biofilm-associated cell surface organelles, such as flagella and curli (Sharma et al., 2010). Escherichia coli also synthesizes an autoinducer 2 (AI-2) that is produced by LuxS (Surette et al., 1999), regulates 5.6% of the E. coli genome (DeLisa et al., 2001) and has lately been recognized as a target of the next generation antimicrobials (Zhu & Li, 2012; Roy et al., 2013).
Signal transduction as a biofilm control mechanism: two-component signaling (2CSTS)
Whereas QS is a mechanism by which bacteria communicate with each other, 2CSTS permit bacteria to respond to signals from the environment (for recent review articles on 2CSTS and their use in the genetic engineering of novel cell functions, please see Kenney, 2010; Galperin, 2010; Ninfa, 2010; Stock et al., 2000). In E. coli K-12, a total of 37 2CSTS control many metabolic phenotypes in response to a diversity of signals from the environment (Zhou et al., 2003). Pathogenic strains of E. coli, which often have up to 1000 genes in excess of E. coli, harbor additional 2CSTS that are used to regulate their virulence genes (Tobe, 2008). Altogether, 2CSTS exist in most bacteria but not in higher eukaryotes (e.g. humans), making 2CSTS suitable for the development of novel prevention and treatment techniques for biofilm-associated infectious diseases in humans, animals and plants.
A 2CSTS that was investigated early on and contributed to defining 2CSTS as a new paradigm of gene regulation is the EnvZ/OmpR system which regulates the relative expression of the outer membrane porins OmpF and OmpC in response to changes in osmolarity (Igo et al., 1990). We use this system to explain the concept of 2CSTS. EnvZ constitutes the histidine kinase (Forst et al., 1987), which is membrane bound and acts as the osmolarity sensor. In response to increases in osmolarity, autophosphorylation occurs at a conserved histidine within the transmitter domain of EnvZ. The OmpR response regulator (Hall & Silhavy, 1981) receives the phosphate from EnvZ at a conserved aspartate within its N-terminal receiver domain, causing a conformational change at the C-terminus. Through differential affinities to the OmpR-P binding sites on the ompF and ompC promoters, low levels of phosphorylated OmpR favor the expression of ompF, whereas high levels of phospho-OmpR favor ompC expression. As one example of a slightly different type of 2CSTS, the more complex colanic activator RcsCDB consists of three proteins (Gottesman et al., 1985) and has two each of the receiver and transmitter domains.
Two-component signaling (2CSTS) that were described as having a role in the regulation of E. coli biofilm formation include RcsCDB (Gottesman et al., 1985), EnvZ/OmpR (Prigent-Combaret et al., 2001), CpxA/CpxR (Dorel et al., 1999; Hung et al., 2001), BarA/UvrY (Suzuki et al., 2002; Sakamoto et al., 2012), BasS/BasR (Ogasawara et al., 2012) and QseC/QseB (Rasko et al., 2008; Kostakioti et al., 2009). The CpxR response regulator responds to envelope stress and decreases biofilm formation on hydrophilic surfaces (Ma & Wood, 2009). In addition to curli (Jubelin et al., 2005), the Cpx signaling pathway also regulates P pili (Hung et al., 2001). BarA/UvrY contributes to virulence of APEC by increasing the synthesis of type 1 fimbriae and P pili (Herren et al., 2006) and the virulence of UPEC by increasing the production of hemolysin and lipopolysaccharide, while inducing the cytokine and chemokine host immune response (Palaniyandi et al., 2012). Recently, Sakamoto and coworkers demonstrated UvrY and CpxR were responsible for the previously described enhancement of biofilm formation and cell viability by polyamines (Sakamoto et al., 2012). Among the 20 sites BasR binds to, the csgD operon has the most relevance to biofilm formation, suggesting BasR may be involved in controlling irreversible attachment (Ogasawara et al., 2012). QseC/QseB is particularly interesting since it regulates type I fimbriae in UPEC during urinary tract infection (Kostakioti et al., 2009), and flagella and motility in EHEC (Sperandio et al., 2002). At the same time, it contributes to the QS system. This system is discussed in more detail below.
QseC/QseB, one example of a 2CSTS that contributes to QS
Although the previous two chapters discussed QS and 2CSTS as two independent mechanisms, there are numerous 2CSTS that are involved in QS. In fact, Gram-positive bacteria do not produce AHL type autoinducers at all but instead depend on 2CSTS. QS by means of 2CSTS may be particularly important, as the existence of AHL in the mammalian intestine has been postulated but has yet to be confirmed (Swearingen et al., 2013). In E. coli, QseC/QseB constitutes one example of a 2CSTS that contributes to QS. The formation of intracellular UPEC biofilm is dependent on the phosphatase activity of QseC, as dephosphorylated QseB is needed to express type I fimbriae, curli, flagella (Kostakioti et al., 2009), as well as many metabolic genes (Hadjifrangiskou et al., 2011). A mutant in qseC, while attenuated for virulence already, shows an additional reduction in the establishment of chronic infection after a prophylactic dose of an antagonist for FimH, the mannose-specific tip of the type I fimbrium (Kostakioti et al., 2012). In EHEC, QseC has been proposed as a drug target as well (Fig. 1). LED209, a small molecule that was identified in a high-throughput screen, inhibits the expression of LEE and shiga toxin (stx2) genes by abolishing the binding of the host signal norepinephrine to QseC and preventing autophosphorylation (Rasko et al., 2008). An intriguing piece of regulation is the induction of qseB by the toxin/antitoxin system MqsA/MqsR, which was highly up-regulated in persister cells and whose deletion caused reduced persistence (Ren et al., 2004; Barrios et al., 2006).
As a final 2CSTS discussed in this context, QseE/QseF may not be involved in EHEC biofilm formation but does contribute to virulence by its role in iron uptake and by regulating the expression of additional 2CSTS, such as RcsCDB and the phosphoregulation system PhoP/PhoQ (Reading et al., 2010). Taken together, the Qse systems offer ample opportunity as drug targets in order to prevent UPEC and EHEC infections.
Developing acetate metabolism into a biofilm control mechanism
While all the above examples of 2CSTS required the respective sensor kinase to phosphorylate its cognate response regulator, some of the response regulators can also be activated by intermediates of acetate metabolism. The first example of such an activation would be the acetylation of RcsB (Thao et al., 2010), CpxA and UvrY (Ma & Wood, 2011) by acetyl-CoA, the endproduct of glycolysis. The second example is the phosphorylation of OmpR (Shin & Park, 1995) and numerous others (McCleary et al., 1993; Wanner, 1993; Klein et al., 2007) by acetyl phosphate, the activated acetate intermediate between acetyl-CoA and acetate (Brown et al., 1977; Holms, 1986; el-Mansi & Holms, 1989). This metabolic activation of the response regulators constitutes an important connection between central metabolism and signal transduction (Wolfe et al., 2003). It also controls important cellular processes, such as the cell division rate (Prüß & Matsumura, 1996) and biofilm amounts (Prüß et al., 2010) in response to signals from the environment.
The cell division rate was controlled by the addition of serine to the bacterial growth medium in our first study of this kind (Prüß & Matsumura, 1996). It was hypothesized that serine led to an increase in the acetyl phosphate levels, an increase in the levels of phosphorylated OmpR, a decrease in flhD (encodes flagellar master regulator) expression, and the simultaneous increase in the cell division rate and decrease in flagellar synthesis (Fig. 2). The hypothesis was built upon the observations that (i) mutants in flhD (Prüß & Matsumura, 1996) and ackA (encodes acetate kinase) (Prüß, 1998) divided to higher cell densities while getting smaller, (ii) mutants in ompR and pta (encodes phosphotransacetylase) divided to lower cell densities while getting longer (Prüß, 1998), (iii) the addition of carbon sources that get converted to acetate also increased the final cell number (Prüß, 1998) and (iv) phosphorylation of OmpR led to a 10 times increased binding affinity for the flhD promoter (Shin & Park, 1995). The flhD operon encodes FlhD and FlhC (Bartlett et al., 1988), the two components of the heterohexameric (Wang et al., 2006) FlhD/FlhC complex that acts as a transcriptional regulator for all flagellar genes (Silverman & Simon, 1973) and genes of anaerobic metabolism and sugar acid degradation (Prüß et al., 2001, 2003). It also impacts colonization of the mouse intestine (Horne et al., 2009). In E. coli O157 : H7, FlhC inhibits cell division simultaneously with biofilm and the synthesis of LEE genes when bacteria grow on meat (Sule et al., 2011).
After the discovery of the impact of the acetate-OmpR-FlhD/FlhC pathway on the cell division rate, a partial network of regulation was summarized that centered on FlhD/FlhC, included numerous 2CSTS, and regulated all the biofilm-associated cell surface organelles (Prüß et al., 2006). A computational study by the same authors expanded on the list of 2CSTS that may be involved in the regulation of biofilm beyond the previously described ones (Denton et al., 2008). This included DcuS/DcuR, a 2CSTS that is involved in the regulation of C4-dicarboxylic acids (Golby et al., 1999; Scheu et al., 2011), and ArcB/ArcA, which is involved in anaerobic metabolism (Iuchi, 1993; Alvarez & Georgellis, 2010). Intriguingly, the increased amounts and three-dimensional structures of biofilm that were formed by an ackA mutant strain (Prüß et al., 2010) were abolished by the introduction of an additional mutation in dcuR (Prüß et al., 2010) or arcB (L. Nessa and B.M. Prüß unpublished data) into this strain.
While the role of acetyl phosphate (and the subsequent phosphorylation of the 2CSTS response regulators) in biofilm formation has been known for a long time (Wolfe et al., 2003; Fredericks et al., 2006; Klein et al., 2007; Schwan et al., 2007), an important piece of evidence for an involvement of additional acetate intermediates was obtained from a high-throughput experiment that determined genetic and environmental factors affecting biofilm amounts (Prüß et al., 2010). This hypothesis was further supported by microscopy observations where ackA and pta ackA mutant strains produced larger amounts of biofilm than their isogenic parent strain, as well as more three-dimensional structures within the biofilm. This was true for both the ackA mutant strain, which accumulates acetyl phosphate, and the pta ackA mutant strain, which is unable to produce acetyl phosphate (Prüß & Wolfe, 1994). We postulated that acetyl-CoA rather than acetyl phosphate affected biofilm amounts in this study (Prüß et al., 2010). Support for this postulate was obtained with mutations in ldhA (encodes lactate dehydrogenase) and pflA (encodes pyruvate formate lyase) that also impact intracellular acetyl CoA levels. These mutations had a similar effect as was seen for the ackA and ackA pta mutations in a phenotype microarray experiment (Mugabi et al., 2012).
As a conclusion of this review article, we propose the use of this link between acetate metabolism and 2CSTS as a mechanism to control biofilm amounts by the addition of nutrients to the bacterial growth medium. As one example of this type of control, E. coli grown on C6-sugars formed particularly large amounts of biofilm, whereas C5-sugars only gave rise to minimal amounts of biofilm (Prüß et al., 2010). C6-sugars are converted to acetyl-CoA during glycolysis and have been shown to increase acetyl phosphate levels (McCleary et al., 1993). This observation is similar to our earlier observation where carbon sources that get converted to acetyl-CoA increased the cell division rate through phosphorylation of OmpR (Prüß, 1998).
Intriguingly, some carbon sources that do not get converted to acetate can still permit the formation of large amounts of biofilm. For example, several C4-dicarboxylic acids allowed larger amounts of biofilm for the E. coli parent strain, but not for the dcuR mutant (Prüß et al., 2010). This may be an example where the pathway from the nutrient to biofilm amounts may not involve acetyl phosphate or acetyl-CoA directly. The information may be signaled directly by the 2CSTS that is responsible for these nutrients through the respective sensor kinase, DcuS in this case. Altogether, we believe nutrient supplementation and 2CSTS allow for ample opportunity to control biofilm in many bacterial species and for multiple biofilm-associated medical problems. The precise nutrient that works best likely depends on the combination of available 2CSTS in each strain and the specific biofilm-associated disease.
Conclusion and outlook
Escherichia coli biofilm contributes to many infectious diseases. Genes that contribute to biofilm are estimated at more than 250 (Domka et al., 2007) and regulation of these genes is complex. Two gene regulatory mechanisms that are particularly promising for the development of novel biofilm control techniques are quorum sensing and two-component signaling. The latter is linked to the bacterial central metabolism through several activated acetate intermediates whose concentrations vary with the identity and concentration of the available nutrient sources. Other signaling molecules may feed into specific 2CSTSs through their sensor kinase. An even larger group of metabolic intermediates may be entirely unrelated to two-component signaling (Chant & Summers, 2007; Garavaglia et al., 2012). Supplying the bacteria with any such signaling molecules may, 1 day, constitute a novel mechanism to control the amounts of biofilm.
The authors would like to thank Drs Phil Matsumura (University of Illinois, Chicago, IL) and Alan J. Wolfe (Loyola University, Chicago) for countless discussions over the years that led to the development of this research and Dr Shelley Horne (North Dakota State University, Fargo, ND) for critically reading the manuscript. The anonymous reviewers are thanked for one of the most thorough reviews we ever got on a manuscript. P.S. and B.M.P. were funded by grant 1R15AI089403 from the NIH/NIAID. T.L. and B.M.P. were funded by the North Dakota Agricultural Experiment Station. The authors declare that they have no conflict of interest.