The bacterial cell surface
Bacterial attachment to surfaces or other cells can be seen as a physicochemical process determined by Van der Waals, electrostatic and steric forces acting between the cells and the attachment surface. A theory to quantitatively describe this interaction of charged surfaces through a liquid medium, designated the Derjaguin, Verwey, Landau and Overbeek (DLVO) theory, has been developed in the 1940s. Later, an extended DLVO theory was developed, which incorporated besides these long-range forces also hydrophilic/hydrophobic and osmotic interactions, resulting in more accurate predictions of bacterial adhesion [reviewed by Strevett and Chen (2003)]. These theories are not reviewed here, instead the wide variety of individual outer cell surface structures and molecules that are exposed on, or protrude from, the cell surface are described in detail. These structures shape the physicochemical surface properties of bacterial cells, and hence determine attachment and biofilm formation properties. However, the presence or absence of a certain structure on initial attachment or biofilm formation should be evaluated with care because multiple structures can be present, each with their own specific effects, and different structures could have diverse roles depending on the bacterium and the attachment surface.
Many bacteria are motile by virtue of peritrichous or polar flagella, and motility is generally regarded as a virulence factor facilitating the colonization of host organisms or target organs by pathogenic bacteria. Flagellar motility is critical for initial cell-to-surface contact and normal biofilm formation under stagnant culture conditions for Escherichia coli (Pratt and Kolter 1998), Listeria monocytogenes (Vatanyoopaisarn et al. 2000; Lemon et al. 2007; Todhanakasem and Young 2008) and Yersinia enterocolitica (Kim et al. 2008). Although lack of flagella also affected initial attachment under flow conditions for Y. enterocolitica and L. monocytogenes, further maturation was unaffected for Y. enterocolitica (Kim et al. 2008), and the formation of high density biofilms was not suppressed for L. monocytogenes (Todhanakasem and Young 2008). For Pseudomonas fluorescens, mutants lacking flagella showed a decreased attachment to a variety of plant seeds and inert surfaces such as sand (Deflaun et al. 1990, 1994) and a decreased colonization of potato roots (De Weger et al. 1987). Finally, initial attachment of L. monocytogenes to stainless steel can also be affected by flagella per se (Vatanyoopaisarn et al. 2000). These observations indicate that flagella can affect adherence and biofilm formation via different mechanisms depending on the type of bacterium. First, motility can be necessary to reach the surface by allowing the cell to overcome the repulsive forces between the cell and the surface. This mechanism is possibly more important under stagnant than under flow conditions. In addition, motility can be required to move along the surface, thereby, facilitating growth and spread of a developing biofilm. Finally, flagella themselves (as surface appendages) can directly mediate attachment to surfaces.
Fimbriae, thread-like structures that protrude from the cell surface, are classified on the basis of their adhesive, antigenic or physical properties, or on the basis of similarities in the primary amino acid sequence of their major protein subunits (Low et al. 1996). Type 1 fimbriae, which are rod-shaped and approximately 7-nm wide and 1-μm long, are the most common adhesins found in both commensal and pathogenic E. coli as well as in other Enterobacteriaceae (Klemm and Krogfelt 1994). Their role in biofilm formation has been studied exhaustively, demonstrating a critical role in initial stable cell-to-surface attachment for numerous E. coli strains (Pratt and Kolter 1998; Beloin et al. 2004; Ren et al. 2004) including Shiga toxin-producing strains (Cookson et al. 2002), in adherence to Teflon and stainless steel for Salmonella enterica serovar Enteritidis (Austin et al. 1998), and in promoting biofilm formation on abiotic surfaces (polystyrene) for Klebsiella pneumoniae (Schembri et al. 2005).
Besides Type 1 fimbriae, other types of fimbriae have been shown to affect biofilm formation. For example, Di Martino et al. (2003) showed that for a Kl. pneumoniae strain, which produced both Type 1 and Type 3 fimbriae, the latter constituted the main factor facilitating adherence to both glass and polypropylene, and the formation of a full-grown biofilm on polystyrene. Type 4 fimbriae promoted the rapid formation of strongly adherent biofilms for the opportunistic pathogen Aeromonas caviae (Bechet and Blondeau 2003), commonly found in water and foods (Neyts et al. 2000), and affected the binding of Pseudomonas aeruginosa to stainless steel, polystyrene and polyvinylchloride (Giltner et al. 2006). Genes involved in the biogenesis, regulation and secretion of Type 4 fimbriae were found to be up-regulated within 6 h of attachment to silicone tubing for Pseudomonas putida (Sauer and Camper 2001), often associated with spoilage of fresh milk and vegetables (Ternstrom et al. 1993; Garcia-Gimeno and Zurera-Cosano 1997). Type 4 fimbriae also played a role in the colonization and persistence of Vibrio vulnificus in oysters (Paranjpye et al. 2007). Vibrio vulnificus is a pathogen associated with human infections caused by raw oyster consumption (Blake et al. 1979) and an important cause of reported deaths from food-borne illness in Florida (Hlady et al. 1993). Furthermore, for enterohemorrhagic E. coli O157:H7, these structures not only affected attachment and biofilm formation but have also been implicated in virulence and transmission (Xicohtencatl-Cortes et al. 2009).
Curli fimbriae (called thin aggregative fimbriae in Salmonella) are proteinaceous, coiled filamentous surface structures, which are assembled by an extracellular nucleation/precipitation pathway (Olsen et al. 1989). The effect of curli on attachment and biofilm formation of E. coli O157:H7 appears to be variable. In one study, curli production enhanced the biofilm-forming capacity of a particular strain to stainless steel (Ryu et al. 2004b), although initial attachment was unaffected (Ryu and Beuchat 2005). In another study, different Shiga toxin-producing and enterohaemorrhagic E. coli strains showed an enhanced attachment to abiotic surfaces such as polystyrene and stainless steel when curli were produced (Cookson et al. 2002; Pawar et al. 2005). Probably, this increased attachment is strain dependent as shown in a study comparing the attachment of curli-producing and noncurli-producing E. coli O157:H7 strains to lettuce (Boyer et al. 2007). Interestingly, it cannot be excluded that the observed differences are not only strain dependent, but are also induced by other (nonevaluated) mechanisms or by the occurrence of dissimilar environmental triggers in the experiments.
In addition to curli, cellulose is also usually associated with biofilms of various salmonellae, including strains of the serovar Typhimurium (Solano et al. 2002; Jain and Chen 2007). The simultaneous production of cellulose and curli leads to the formation of a highly inert, hydrophobic extracellular matrix in which the cells are embedded (Zogaj et al. 2001). However, other capsular polysaccharides can be present in the extracellular biofilm matrix of Salmonella strains (de Rezende et al. 2005), and the exact composition depends upon the environmental conditions in which the biofilms are formed (Prouty and Gunn 2003). A variety of environmental cues such as nutrients, oxygen tension, temperature, pH, ethanol and osmolarity can influence the expression of the transcriptional regulator CsgD, which regulates the production of both cellulose and curli (Gerstel and Romling 2003). In addition, a study of 122 Salmonella strains indicated that all had the ability to adhere to plastic microwell plates and that, generally, more biofilm was produced in low nutrient conditions, as can be found in specific food-processing environments, compared to high nutrient conditions (Stepanovic et al. 2004).
Pili are structurally similar to fimbriae and are involved in a process of horizontal gene transfer called conjugation. Mostly, the transferred DNA is a conjugative plasmid encoding the formation of the conjugative pilus itself, and thereby mediates an intimate cell-to-cell contact. This conjugation process can stimulate biofilm development, because the conjugative pilus can act as an adhesion factor allowing nonspecific cell-solid surface or cell–cell contacts (Ghigo 2001; Reisner et al. 2003). Vice versa, the high density of bacterial populations in biofilms can stimulate conjugation and plasmid dispersal (Hausner and Wuertz 1999; Molin and Tolker-Nielsen 2003) and can therefore contribute to the spread of resistance genes, which are often also carried on the plasmid (Bower and Daeschel 1999). Luo et al. (2005) have demonstrated that conjugation enhanced the expression of CluA, a surface-bound clumping protein encoded by the chromosomally embedded sex factor, and subsequently facilitated biofilm formation in Lactococcus lactis. Furthermore, this enhanced biofilm-forming trait is transmissible by conjugation.
In addition to proteinaceous organelle-type surface appendages, some Gram-negative bacteria can produce autotransporter proteins. These are secretory proteins that contain in their primary structure all the information necessary to direct their own secretion across the cytoplasmic and outer membrane to the bacterial cell surface. Adhesive phenotypes such as aggregation and biofilm formation have been attributed to a subfamily of E. coli autotransporters, including antigen 43 (Ag43) (Danese et al. 2000a; Kjaergaard et al. 2000), the AIDA adhesin associated with some diarrheagenic E. coli (Sherlock et al. 2004), and the TibA adhesin/invasin from enterotoxigenic E. coli (Sherlock et al. 2005).
The lipopolysaccharide (LPS) outer layer of Gram-negative bacteria typically consists of a surface exposed O-antigen, a core structure and a lipid A moiety that is embedded in the outer membrane lipid bilayer. The LPS layer not only affects the bacterium’s susceptibility to disinfectants, antibiotics and other toxic molecules (Russell and Furr 1986), it also plays a role in biofilm formation. For example, O-antigen mutants of Salmonella enterica serovar Typhimurium showed reduced capacities to attach and colonize alfalfa sprouts (Barak et al. 2007). Alterations in the LPS of Salm. Typhimurium had also osmolyte-dependent effects on biofilm formation (Anriany et al. 2006). For E. coli, truncation of LPS (deep-rough phenotype) did not affect adhesion per se, but had a pleiotropic effect on the biosynthesis of Type 1 fimbriae and flagella, resulting in a reduced adherence (Genevaux et al. 1999). Alterations in the peptidoglycan structure exposed at the surface of Gram-positive bacteria can also have an effect on attachment, as shown by analysis of L. monocytogenes rough colony variants. The latter, characterized by an impaired cellular localization of several peptidoglycan-degrading enzymes such as the cell wall hydrolase A (CwhA), showed enhanced attachment to stainless steel (Monk et al. 2004).
Many bacteria produce and secrete extracellular polysaccharides (EPS). The polysaccharide-containing layers outside the cell are collectively defined as glycocalyx, but when the layers are rigid and organized in a tight matrix that excludes particles, the term capsule is used. If the layers do not exclude particles and are more easily deformed and detached, the term slime is used. These EPS are an important constituent of the extracellular matrix characteristically produced by many biofilms. The matrix often contains additional constituents, such as nucleic acids, proteins, glycoproteins and lipoproteins.
For Kl. pneumoniae, the capsule is considered to be a dominant virulence factor, and its synthesis blocked Type 1 fimbriae-promoted biofilm formation on abiotic surfaces (see above), thereby, actually reducing the bacterial adhesion to such surfaces (Schembri et al. 2005). For V. vulnificus, expression of capsular polysaccharides also inhibited attachment and biofilm formation on abiotic surfaces (plastic) (Joseph and Wright 2004). The EPS colanic acid (or M antigen) produced by most E. coli strains as well as by other species of the Enterobacteriaceae appears to be important for establishing the complex structure and depth of E. coli biofilms, but not for initial attachment to abiotic surfaces (Danese et al. 2000b; Prigent-Combaret et al. 2000). Overproduction of EPS can even inhibit initial attachment of E. coli O157:H7 to stainless steel (Ryu et al. 2004a). The unbranched polysaccharide, β-1,6-poly-N-acetyl-d-glucosamine (PGA), is involved not only in adhesion by staphylococci, but also in attachment to abiotic surfaces, intercellular adhesion and biofilm formation of E. coli (Wang et al. 2004). Furthermore, depolymerization of PGA led to dispersal of biofilms (Itoh et al. 2005). Colanic acid, PGA and cellulose production, but not LPS production, affected binding of E. coli O157:H7 to alfalfa sprouts as shown by mutational analysis (Matthysse et al. 2008).
These observations indicate contrasting roles for EPS (and LPS) in biofilm formation of different bacteria. The particular function of EPS in biofilm formation may depend on its structure, relative quantity and charge and on the properties of the abiotic surface and surrounding environment. Furthermore, EPS play a role not only in biofilm formation but also in the increased resistance of biofilm bacteria to biocides as described in section Implications of biofilm formation.
Factors affecting the bacterial cell surface
The attachment and biofilm-forming capabilities of bacteria depend on multiple factors including the attachment surface (see below), the presence of other bacteria, the temperature, the availability of nutrients and pH. Although the mechanisms underlying these effects are not always explained, biofilm formation can in some cases be influenced through alterations of the bacterial cell surface. For instance, curli expression and attachment to plastic surfaces by enterotoxin-producing E. coli strains was found to be higher at 30°C than at 37°C (Szabo et al. 2005). Similarly, expression of thin aggregative fimbriae in Salm. Typhimurium and of fimbriae in Aeromonas veronii strains isolated from food was affected by temperature, with a lower temperature (28 and 20°C, respectively) favouring expression (Kirov et al. 1995; Romling et al. 1998). Production of these outer surface structures at low(er) temperatures could enhance the attachment to surfaces and hence facilitate persistence and survival in food-processing environments. The adhesion of L. monocytogenes to polystyrene after growth at pH 5 was lower than after growth at pH 7, and this could be attributed to the down-regulation of flagellin synthesis (Tresse et al. 2006).
The large cell densities existing in biofilms create a local environment suitable for cell density-dependent bacterial communication. Bacteria throughout the bacterial kingdom have evolved the ability to steer the behaviour of individual cells, populations or communities by using various modes of communication. One of the best studied communication mechanisms in bacteria is quorum sensing, which is based on the production of low-molecular-mass signalling molecules. When the bacterial cell density is low, the extracellular concentration of the signals will also be low and remain undetected. However, as the cell density increases in a growing (biofilm) population, a critical signal concentration will be reached, allowing the signalling molecule to be sensed and enabling the bacteria to respond. The nature of the signalling molecules is diverse. While most Gram-negative bacteria use N-acyl-homoserine lactones (AHL) as signalling molecules (Lazdunski et al. 2004), Gram-positive bacteria commonly use amino acids and short post-translationally processed peptides (Sturme et al. 2002). Additional families of bacterial signalling molecules have been identified such as Autoinducer-2 (AI-2) for both Gram-negative and Gram-positive bacteria (Schauder and Bassler 2001; Xavier and Bassler 2003).
These communication mechanisms control various functions such as virulence, biofilm development and the production of antimicrobial compounds and several other secondary metabolites. As such, quorum sensing can affect the establishment of bacteria in a mixed biofilm community (Moons et al. 2006), their food spoilage potential (Ammor et al. 2008; Wevers et al. 2009), or their survival in particular (food-processing related) stressful environments (Van Houdt et al. 2006, 2007a). Also, the production of surface appendages and motility, putatively affecting biofilm formation, can be regulated by quorum sensing (Daniels et al. 2004; Van Houdt et al. 2007b).
Although quorum sensing has been shown to play a role in biofilm formation for several bacteria, this is not always the case, and no consistent correlation was found between AHL or AI-2 production and biofilm-forming capacity of 68 Gram-negative strains isolated from an industrial kitchen (Van Houdt et al. 2004).
The attachment surface and environmental parameters
The properties of the attachment surface are important factors that affect and determine the biofilm formation potential together with the bacterial cells. The choice of material is therefore of great importance in designing food contact and processing surfaces. Properties such as surface roughness, cleanability, disinfectability, wettability (determined by hydrophobicity) and vulnerability to wear influence the ability of cells to adhere to a particular surface and thus determine the hygienic status of the material. In addition, materials in direct contact with foods have to meet certain specifications and are subject to official approval procedures before they can be used. Materials often used in the food industry include plastics, rubber, glass, cement and stainless steel. The degree of biofilm formation on different materials for Legionella pneumophila has been ranked by Rogers et al. (1994) and by Meyer (2001) with the capacity to support biofilm growth increasing from glass, stainless steel, polypropylene, chlorinated PVC, unplasticized PVC, mild steel, polyethylene, ethylene-propylene to latex.
However, general predictions for the degree of biofilm formation on a particular material cannot be made because the biofilm-supporting capacity of any material also depends on bacteria and on environmental factors. For instance, temperature and nutrient availability can influence the ability of L. monocytogenes to adhere to polyvinyl chloride, buna-n rubber and stainless steel, because of altered bacterial surface physicochemical properties like hydrophobicity/hydrophilicity and surface charge (Briandet et al. 1999; Norwood and Gilmour 1999; Moltz and Martin 2005).
In food-processing environments, bacterial attachment is additionally affected by food matrix constituents. Residues from ready-to-eat meat products such as small amounts of meat extract, frankfurters or pork fat, initially reduced biofilm formation of L. monocytogenes, but with time supported increased biofilm cell numbers and prolonged survival on a variety of materials including stainless steel, conveyor belt rubber, and wall and floor materials (Somers and Wong 2004). Skim milk and milk proteins such as casein and lactalbumin were found to significantly reduce the attachment of Staphylococcus aureus, Serratia marcescens, Pseudomonas fragi, Salm. Typhimurium, spores and vegetative cells of thermophilic bacilli, and L. monocytogenes to stainless steel (Helke et al. 1993; Wong 1998; Barnes et al. 1999; Parkar et al. 2001) and buna-n rubber gaskets (Helke et al. 1993; Wong 1998). Not only physicochemical, but also nutritional properties of the food matrix affect attachment and persistence. For instance, Allan et al. (2004a,b) showed that survival rates of L. monocytogenes on several surfaces, including stainless steel, acetal resin, mortar and fibreglass reinforced plastic, were higher in the presence of biological soil (porcine serum). Finally, the presence of a mixed microbial community adds additional complexity to attachment and biofilm formation under certain conditions. The presence of Staphylococcus xylosus and Ps. fragi affected the numbers of L. monocytogenes found in biofilms on stainless steel (Norwood and Gilmour 2001). Similarly, bacteriocin-producing L. lactis as well as several endogenous bacterial strains isolated from food-processing plants influenced the establishment of L. monocytogenes on stainless steel, suggesting that the ‘house microflora’ can have a strong suppressing effect on L. monocytogenes establishment in biofilms in a food-processing environment (Leriche et al. 1999; Carpentier and Chassaing 2004).
Stainless steels, in particular austenitic grades 304 and 316, are probably the most commonly used food contact surfaces because of their chemical and mechanical/physical stability at various food-processing temperatures, cleanability and high resistance to corrosion (Zottola and Sasahara 1994). The grade, which reflects the composition and to a lesser extent the finish (pickling, bright annealed), significantly affected the hygienic status of stainless steel as measured by the number of residual adhering Bacillus cereus spores after a complete run of soiling followed by a cleaning-in-place procedure (Jullien et al. 2003). Grade 316 has nearly the same mechanical and physical characteristics as 304 but has a higher resistance to corrosion by foods, detergents and disinfectants, because of the anticorrosive properties of the added molybdenum. Food contact surfaces are commonly treated with disinfectants and cleaning agents that contain peroxides, chloramines or hypochlorites. In particular, the latter can be very aggressive to stainless steels depending on the prevailing pH. The liberation of free chlorine can cause pitting, characterized by local breakdown of the protective ‘passive’ oxide surface layer and formation of local deep pits on these free surfaces, thereby facilitating bacterial adhesion and biofilm formation. Therefore, the duration and operating temperature of cleaning and disinfection treatments should be carefully controlled, and thorough rinsing with water should always be performed as a last step (BSSA 2001).