Correspondence: James P. O'Gara, School of Biomolecular and Biomedical Science, Ardmore House, University College Dublin, Belfield, Dublin 4, Ireland. Tel.: +353 1 716 1263; fax: +353 1 716 1183; e-mail: email@example.com
Recent progress in elucidating the role of the icaADBC-encoded polysaccharide intercellular adhesin (PIA) or polymeric N-acetyl-glucosamine (PNAG) in staphylococcal biofilm development has in turn contributed significantly to our understanding of the pathogenesis of device-related infections. Nevertheless, our understanding of how the ica locus and PIA/PNAG biosynthesis are regulated is far from complete and many questions remain. Moreover, beyond ica, evidence is now emerging for the existence of ica-independent biofilm mechanisms in both Staphylococcus aureus and Staphylococcus epidermidis. Teichoic acids, which are a major carbohydrate component of the S. epidermidis biofilm matrix and the major cell wall autolysin, play an important role in the primary attachment phase of biofilm development, whereas the cell surface biofilm-associated protein and accumulation-associated protein are capable of mediating intercellular accumulation. These findings raise the exciting prospect that other surface proteins, which typically function as antigenic determinants or in binding to extracellular matrix proteins, may also act as biofilm adhesins. Given the impressive array of surface proteins expressed by S. aureus and S. epidermidis, future research into their potential role in biofilm development either independent of PIA/PNAG or in cooperation with PIA/PNAG will be of particular interest.
Bacterial biofilms are now known to play an important role in a range of chronic infections such as native valve endocarditis, otitis media, cystic fibrosis pneumonia and infections associated with implanted biomaterials (Costerton et al., 1999). The capacity of both Staphylococcus epidermidis and Staphylococcus aureus to form biofilm is an important virulence factor in the development of device-related infections. Device-related infections, in turn, represent a serious clinical problem, given that the majority of hospital patients undergo procedures for the insertion of foreign devices, from catheters to artificial heart valves, etc. Moreover, patients susceptible to device-related infections are often colonized with hospital-acquired, multiple antibiotic-resistant organisms and may be further compromised by serious underlying disease or trauma. Significantly, the majority of biofilm-mediated device-related infections are caused by either S. epidermidis or S. aureus (O'Gara & Humphreys, 2001).
Biomaterials are rapidly coated with a conditioning film composed primarily of host-derived extracellular matrix proteins following insertion into a patient. Some of these host proteins can act as receptors for bacterial attachment and can thus clearly play an important role in the early stages of biofilm development. Staphylococcal binding to extracellular matrix proteins is mediated by microbial surface components recognizing adhesive matrix molecules (MSCRAMMS) (Foster & Hook, 1998). Following initial attachment to an implanted device surface, biofilm formation involves a cellular accumulation process.
The icaA gene product is a transmembrane protein with homology to N-acetyl-glucosaminyltransferases, requiring the icaD gene product for optimal activity (Gerke et al., 1998). N-acetyl-glucosamine oligomers produced by icaAD reach a maximal length of 20 residues and it is only when icaAD is coexpressed with icaC, which encodes a putative membrane protein, that longer oligomer chains are synthesized (Gerke et al., 1998). IcaC is also likely to be involved in translocation of the growing polysaccharide to the cell surface. The surface-attached protein IcaB is then responsible for deacetylation of the poly-N-acetylglucosamine molecule (Vuong et al., 2004). Nondeacetylated poly-acetylglucosamine in an isogenic icaB mutant strain cannot attach to the bacterial cell surface or mediate biofilm development (Vuong et al., 2004).
The divergently transcribed icaR gene, which is located upstream of the icaADBC operon in both S. epidermidis and S. aureus, appears to be a member of the tetR family of transcriptional regulators. It has recently been reported that the icaR gene encodes a transcriptional repressor with a central role in the environmental regulation of ica operon expression in S. epidermidis (Conlon et al., 2002a). Modification of the bacterial environment by the addition of NaCl or ethanol to the growth medium can activate the ica operon via separate regulatory pathways in an icaR-dependent manner (Conlon et al., 2002a, b, 2004). Using DNase I footprint analysis, Jefferson et al. (2004) demonstrated that purified IcaR protein from S. aureus bound the ica operon promoter region close to the icaA start codon. Using DNA affinity chromatography to isolate proteins that bound to the ica promoter region, Jefferson et al. (2004) also identified TcaR as a weak repressor of ica operon transcription. TcaR is a member of the MarR family of transcriptional regulators and controls expression of the teicoplanin-associated locus. However, deletion of tcaR alone in S. aureus was not associated with any changes in PNAG or biofilm production, whereas deletion of both tcaR and icaR was found to have a synergistic effect on these phenotypes (Jefferson et al., 2004).
In addition to IcaR and TcaR, the insertion sequence element IS256 can also play a negative role in the control of S. epidermidis biofilm development. Ziebuhr et al. (1999) first demonstrated that reversible integration of the IS256 element into the ica locus results in biofilm-positive to biofilm-negative phenotypic switching. It was subsequently demonstrated that IS256 insertions in the rsbU gene of the sigB operon or the sarA locus, which both result in diminished ica operon expression, were also associated with phenotypic switching (Conlon et al., 2004; Handke et al., 2004). Clearly, IS256 mutations in loci encoding global regulators will affect multiple traits in staphylococci, suggesting that this genetic element may play a broader role in facilitating rapid phenotypic responses to specific external environmental signals.
An important component in bacterial biofilms is the capacity to communicate using quorum-sensing systems. The most important and best-characterized quorum-sensing system in staphylococci is the accessory gene regulator (agr) system, which does not seem to affect ica expression and the production of PIA/PNAG (discussed below). A second quorum-sensing system, the LuxS system has recently emerged as another negative regulator of biofilm formation in S. epidermidis (Xu et al., 2006). Mutation of the luxS gene, which is required for synthesis of the AI-2 autoinducer molecule, positively impacted on ica operon expression, PIA/PNAG production and biofilm development. Significantly, the luxS mutant also exhibited enhanced virulence in a rat model of biofilm-associated infection (Xu et al., 2006). Interestingly, Doherty et al. (2006) concluded that unlike S. epidermidis, mutation of luxS in S. aureus was not associated with altered biofilm-forming capacity.
The global stress response regulator σB and RsbU (a positive regulator of σB) control biofilm development in S. epidermidis (Knobloch et al., 2001) but not S. aureus (Valle et al., 2003). σB controls ica operon expression in S. epidermidis by repressing transcription of icaR via an unidentified regulatory intermediate (Knobloch et al., 2004). Interestingly, biofilm induction by NaCl is impaired in S. epidermidis rsbU mutants, whereas ethanol-mediated biofilm activation is unaffected (Knobloch et al., 2001; Conlon et al., 2004), further suggesting the existence of two regulatory pathways controlling biofilm expression. Consistent with this ethanol-mediated icaR repression is σB-independent (Knobloch et al., 2004). The differential effect of σB on ica operon expression in S. epidermidis and S. aureus explains, at least in part, differences in the regulation of biofilm development in these two species. Interestingly, σB also appears to be important for the long-term stability of S. epidermidis biofilms (Jager et al., 2005). Using ethanol to induce biofilm formation in σB mutants, it was observed that, over time, biofilms formed by sigB mutants were more rapidly dispersed than wild-type biofilms. Although this further highlights the importance of σB for both biofilm development and stability, it is interesting to note that biofilm dispersal may also be beneficial for the pathogen in terms of dissemination. In addition, although σB does not appear to be required for biofilm development in S. aureus, it would be interesting to examine the stability of biofilms formed by S. aureus sigB mutants.
In S. aureus, the Spx protein, which interacts directly with the α subunit of RNA polymerase to regulate transcription initiation, also contributes to stress resistance and biofilm regulation. Specifically, Spx appears to activate icaR transcription, leading to decreased ica operon expression (Frees et al., 2004; Pamp et al., 2006). Interestingly, Spx is normally degraded by the ClpXP protease. Consistent with this, biofilm formation was impaired in a clpX mutant (Frees et al., 2004). However, the absence of ClpP was associated with enhanced biofilm formation (Frees et al., 2004), leading the authors to speculate that although ClpP may positively contribute to PIA/PNAG production, it may also play a negative role in an ica-independent biofilm mechanism (Pamp et al., 2006).
Two studies published in 2003 reported that the sarA locus was required for ica operon transcription, PIA/PNAG production and biofilm formation in S. aureus (Beenken et al., 2003; Valle et al., 2003). SarA is a major global regulator of a diverse range of virulence determinants in S. aureus (Cheung & Zhang, 2002), with gene chip analysis revealing that this protein can affect the transcription of at least 120 genes in S. aureus (Dunman et al., 2001). Given that SarA also regulates agr (Cheung & Projan, 1994), the impact of agr mutations on biofilm development was also examined. These agr mutants had similar biofilm-forming capacity compared with the wild-type strains, indicating that SarA affects biofilm formation by an agr-independent pathway (Beenken et al., 2003; Valle et al., 2003). Because σB regulates sarA expression in S. aureus and ica operon transcription is decreased in S. epidermidisσB mutants, Valle et al. (2003) also examined the role of σB in the biofilm-negative phenotype of the S. aureus sarA mutants. Mutants lacking σB alone and sarA–σB double mutants were constructed. Interestingly, these σB mutants were able to form a biofilm (Valle et al., 2003), which contradicted the earlier findings of Rachid et al. (2000b), who reported that σB positively regulates biofilm formation in strain M12. This may suggest that σB regulates S. aureus biofilm in a strain-dependent manner.
The SarA protein of S. epidermidis is 84% identical to SarA of S. aureus despite significant divergence at the nucleotide level outside the coding region. The three S. aureus sarA promoters P1, P2 and P3 are present in S. epidermidis but the organization and spacing between these promoters are different from S. aureus (Fluckiger et al., 1998), suggesting that SarA regulation and activity may differ significantly between the two species. As described above and elsewhere, sarA mutations in S. epidermidis are also associated with a biofilm-negative phenotype (Conlon et al., 2004; Tormo et al., 2005b). Purified SarA protein from S. aureus binds to the ica operon promoter of both S. epidermidis and S. aureus with high affinity (Tormo et al., 2005b) and positively regulates ica operon expression via an IcaR-independent mechanism (Conlon et al., 2004; Tormo et al., 2005b). A Tn917 insertion in the S. epidermidis purR gene, whose homologue regulates purine biosynthesis in other gram-positive organisms, negatively impacts ica expression and biofilm development (Mack et al., 2007). However, the absence of identifiable PurR recognition sequences upstream of the icaA or icaR genes or an obvious correlation between purine biosynthesis and the biofilm phenotype suggests that PurR plays an indirect role in ica regulation (Mack et al., 2007). Nevertheless, although connections between biosynthetic pathways and PIA/PNAG production may not be immediately obvious, external stresses that increase PIA/PNAG production and activate biofilm formation also repress central metabolism (Vuong et al., 2005). Furthermore, Vuong et al. (2005) reported that the growth of S. epidermidis in the presence of a subinhibitory concentration of the TCA cycle inhibitor fluorocitrate dramatically increased PIA/PNAG production, perhaps leading the authors to make the interesting suggestion that alterations in the intracellular levels of biosynthetic intermediates, ATP or the redox status of the cell in response to external signals may influence PIA/PNAG biosynthesis (Vuong et al., 2005).
Interestingly, a recent screen of a Mu transposition insertion library in the S. aureus paediatric isolate S30 revealed that the majority of insertion mutations outside the icaADBC locus resulting in a biofilm-impaired phenotype were associated with diminished PIA/PNAG production (Tu Quoc et al., 2007). Clearly, additional regulators of ica operon expression and PIA/PNAG production remain to be described.
The first well-characterized icaADBC-independent biofilm mechanism was described by the research groups of J. Penades and I. Lasa, who demonstrated that the biofilm-associated protein (Bap) was essential for both initial adherence and intercellular accumulation during biofilm development by the S. aureus bovine mastitis isolate V329 (Cucarella et al., 2001). Mutation of the ica locus in this isolate had no impact on the biofilm phenotype (Cucarella et al., 2004). However, SarA acts as a transcriptional activator of the bap gene and therefore a positive regulator of Bap-mediated biofilm development (Trotonda et al., 2005). A Bap homologue protein, which can also mediate ica-independent biofilm, has been identified in some S. epidermidis and other coagulase-negative staphylococcal species (Tormo et al., 2005a). Although the bap gene is present in only 5% of bovine isolates and appears to be absent in human clinical isolates of S. aureus (Cucarella et al., 2001), the Bap protein has now emerged as a member of a group of over 100 surface proteins with conserved structural and functional characteristics from diverse bacterial species, which play an important role in biofilm development [reviewed by Lasa & Penades (2006)].
The research group of M. Smeltzer subsequently demonstrated that deletion of the ica locus in a biofilm-positive S. aureus human clinical isolate UAMS-1 had no impact on biofilm development (Beenken et al., 2004). In contrast, mutation of the sarA gene in S. aureus UAMS-1 resulted in impaired biofilm forming capacity. We recently reported that icaADBC-independent biofilm formation was possible in four clinical methicillin-resistant S. aureus (MRSA) isolates (Fitzpatrick et al., 2005). Interestingly, by examining the environmental regulation of biofilm development and using phage transduction to construct ica deletion mutations, evidence for different mechanisms of biofilm in MRSA and methicillin-sensitive S. aureus (MSSA) isolates was obtained (Fitzpatrick et al., 2006). Specifically, NaCl was significantly more likely to induce biofilm formation in MSSA isolates than MRSA isolates. NaCl is a known activator of ica operon transcription (Rachid et al., 2000a; Fitzpatrick et al., 2005) and, consistent with this, biofilm development in MSSA isolates is apparently icaADBC-dependent and involves the production of PIA/PNAG. Biofilm development in MRSA isolates is primarily glucose-induced, is ica-independent and apparently involves a protein adhesin (Fitzpatrick et al., 2005; O'Neill et al., 2007). These data reveal a previously unknown level of complexity in terms of both biofilm mechanisms and regulation in S. aureus. Nevertheless, consistent with the findings of Beenken et al. (2004) and Trotonda et al. (2005), we recently demonstrated that mutation of the sarA locus abolished biofilm development in eight MRSA and seven MSSA strains (O'Neill et al., 2007). Thus, given that considerable progress has already been made in understanding the role of SarA in controlling ica-mediated biofilm (Valle et al., 2003; Tormo et al., 2005b), these data strongly suggest that SarA may act as a master regulator controlling both ica-dependent and ica-independent biofilm mechanisms.
In contrast to the role of SarA, mutation of the global regulator agr has a neutral effect on biofilm development (Beenken et al., 2003) or can lead to increased biofilm formation (Vuong et al., 2000; Toledo-Arana et al., 2005). Moreover, under different growth conditions, the influence of agr on biofilm formation can vary from positive to neutral to negative, perhaps reflecting the responsiveness of the Agr system to external environmental cues (Yarwood et al., 2004). We recently observed that mutation of the agr locus in 21 S. aureus isolates had a variable (either neutral or positive) effect on biofilm development (O'Neill et al., 2007). In S. aureus, Vuong et al. (2000) observed that expression of the major cell wall autolysin atlE, which has previously been implicated in the early stages of S. epidermidis adherence by directly interacting with hydrophobic surfaces (such as medical devices) (Heilmann et al., 1997) or PIA/PNAG, was not affected by agr mutation. This led them to hypothesize that Agr may influence biofilm development, at least in part, because of the surfactant properties of the agr-encoded δ-toxin, which might prevent hydrophobic interactions between the cell and material surfaces. Consistent with this possibility, when δ-toxin was added to the medium of an S. aureus agr mutant strain, biofilm-forming capacity decreased (Vuong et al., 2000). Similar to S. aureus, mutation of the agr system in S. epidermidis was also associated with increased biofilm formation, at least in part through the δ-toxin, and independent of changes in PIA/PNAG production (Vuong et al., 2003). However, unlike S. aureus, Agr also regulates AtlE production in S. epidermidis. The role of the agr system in the Staphylococcus biofilm phenotype has been covered in depth in a recent review (Kong et al., 2006).
Like Atl, the S. epidermidis surface proteins SSP-1 and SSP-2 have also been implicated in the primary attachment phase of biofilm development (Veenstra et al., 1996). In contrast, the accumulation-associated protein (Aap) has been implicated in biofilm accumulation in S. epidermidis. Aap was originally identified by Hussain et al. (1997), who used mitomycin mutagenesis to generate a mutant that lacked a 140 kDa protein required for biofilm accumulation (Hussain et al., 1997). Interestingly, the observation that anti-Aap antibodies could inhibit PIA/PNAG-dependent biofilm suggested that Aap might somehow anchor PIA/PNAG to the cell surface (Hussain et al., 1997). Consistent with this possibility, a novel repeat domain termed the G5 domain (after conserved glycine residues) implicated in N-acetylglucosamine binding has also been identified in AAP and a number of other Staphylococcus surface proteins (Bateman et al., 2005). However, Rohde et al. (2005) recently demonstrated that proteolytic cleavage of Aap by staphylococcal or host proteases, which converts the protein into an adhesin, is sufficient to mediate biofilm in a PIA/PNAG-independent mechanism. The Aap homologue in S. aureus, designated SasG, is known to mediate binding to the surface of nasal epithelial cells (Roche et al., 2003) but its potential role in biofilm has not yet been investigated.
These data naturally highlight the importance of cell surface molecules in mediating adherence and biofilm development. Teichoic acids, which are now known to be an important component of the staphylococcal biofilm extracellular matrix (Sadovskaya et al., 2004, 2005), have also been shown to play a role in these phenotypes. A mutation in the dlt operon of S. aureus, which mediates d-alanine incorporation into teichoic acids, was biofilm-negative even though the production of PIA/PNAG was unaffected (Gross et al., 2001). However, the primary attachment of the dlt mutant was decreased compared with wild type. This phenotype, which was attributed to the repulsive forces generated by the increased negative charge of the cell wall in the dltA mutant, appeared to prevent bacterial adherence to polystyrene or glass. These data suggest that alterations in the electrostatic forces between the bacterial cell and the adhering surface may influence biofilm-forming capacity (Gross et al., 2001).
A number of transcriptional regulators have recently been implicated in ica-independent biofilm development. Lim et al. (2004) reported that biofilm accumulation by S. aureus is also controlled by an araC-type transcriptional regulator termed regulator of biofilm formation (rbf). Rbf is more than twice the size of average AraC family proteins and may control the expression of a ∼190-kDa protein, which was absent in an rbf mutant. Interestingly, transcription of the ica operon was not affected by the rbf mutation (Lim et al., 2004), suggesting that rbf controls biofilm development by a novel regulatory pathway.
The research group of Inigo Lasa used systematic mutagenesis of two-component systems and random transposon mutagenesis to screen for biofilm-positive mutants of S. aureus strain 15981 (Toledo-Arana et al., 2005). Although capable of biofilm formation, this strain fails to form biofilm in a chemically defined medium. This novel approach revealed that mutation of the autolysis related locus (ArlRS) two-component signal transduction system resulted in increased initial attachment and PIA/PNAG production. Nevertheless, biofilm activation in the arlRS mutant was unaffected by an icaADBC deletion, indicating that PIA/PNAG is not an essential component of the biofilm matrix produced by the 15981 arlRS mutant in chemically defined media. Similarly, disruption of the major autolysin gene, atl, had no effect on the biofilm phenotype of the arlRS mutant. On the other hand, one of the most striking phenotypes associated with a previously constructed arlS mutation was the dramatic decrease in extracellular proteolytic activity (Fournier & Hooper, 2000), which may facilitate the involvement of a putative protein adhesin(s) in attachment and biofilm accumulation. With this in mind, it is particularly interesting to note that biofilm activation in the arlRS mutant was sarA-dependent, which, given that SarA negatively regulates extracellular protease production, may also imply the involvement of a cell wall-anchored or extracellular protein in the arlRS-regulated biofilm phenotype. In contrast to a sarA mutation, deletion of agr enhanced biofilm activation in the arlRS mutant. Interestingly, the ArlRS system has recently been shown to negatively regulate transcription of the dlt operon required for teichoic acid biosynthesis (Koprivnjak et al., 2006). Given the positive role of teichoic acids in primary attachment (Gross et al., 2001), this finding further implicates ArlRS in the regulation of the S. aureus biofilm phenotype.
Further insights into the role of SarA in ica-independent biofilm mechanisms await the identification of other biofilm mediators. However, given that Aap- and Bap-mediated biofilm phenotypes obviously involve protein adhesins, that an arlS mutation is associated with decreased protease production and that glucose-induced MRSA biofilm is mediated by a protein adhesin (O'Neill et al., 2007), it is seems likely that the influence of SarA on biofilm regulation may also relate to its activity as a repressor of extracellular protease production. SarA is known to repress the production of four major extracellular proteases in S. aureus (SspA, SspB, Aur and ScpA), which results in reduced levels of cell-bound fibronectin-binding proteins and protein A (Karlsson et al., 2001). Hence, putative protein adhesin(s) mediating ica-independent biofilm may be modified or degraded in a SarA mutant.
Conclusions and future outlook
The intensification of interest in the Staphylococcus biofilm phenotype has yielded significant insights into the mechanisms and regulation of this important virulence determinant. Considerable progress has been made in profiling the role of the icaADBC-encoded exopolysaccharide PIA/PNAG in biofilm development and the pathogenesis of device-related infections. Nevertheless, our understanding of how the ica locus and PIA/PNAG biosynthesis are regulated is far from complete and many questions remain, such as the identity of the regulatory intermediates controlled by σB and ethanol, which repress icaR transcription.
Beyond ica, recent developments have now revealed that surface proteins such as Aap and Bap can mediate PIA/PNAG-independent intercellular accumulation during biofilm development by ica-deficient strains of S. aureus and S. epidermidis. These findings open up the intriguing prospect that other Staphylococcus surface proteins may also be involved in biofilm development, a possibility that is perhaps predictable given the numbers of surface proteins in staphylococci; S. aureus is predicted to encode 28 surface proteins (Roche et al., 2003; Gill et al., 2005), while 18 have been identified in S. epidermidis (Bowden et al., 2005; Gill et al., 2005). Among the 28 surface proteins identified in S. aureus, 21 are predicted to contain LPXTG motifs required for surface display. Anchoring of LPXTG proteins to the Gram-positive cell wall is catalyzed by an extracellular transpeptidase, Sortase, which is required for virulence (Mazmanian et al., 2000, 2002). Staphylococcus epidermidis possesses 11 LPXTG-motif containing surface proteins and apart from Aap, their role in biofilm development is largely unknown. The remaining seven non-LPXTG surface proteins in S. aureus and S. epidermidis include Atl and AtlE, respectively. The role of surface proteins as biofilm adhesins (as opposed to their role as MSCRAMMs) needs to be further investigated in order to advance our understanding of the pathogenesis of device-related infections. Of particular importance in this context is the observation that almost all S. aureus clinical isolates possess and express the ica operon under various growth conditions, but many are incapable of biofilm development under the same conditions (Fitzpatrick et al., 2005, 2006). These findings represent something of a paradox; if the ica operon is conserved and expressed in most S. aureus clinical strains, then why does it not always mediate biofilm development? Thus, progress in understanding the role of surface proteins as biofilm adhesins may help in answering this question. Alternatively, under appropriate conditions, proteins and other surface molecules may cooperate with PIA/PNAG to mediate intercellular adhesion. Such a scenario would also explain why the ica locus is maintained, expressed and regulated in many Staphylococcus clinical isolates capable of ica-independent biofilm formation.
Useful discussions and input from past and present members of the O'Gara laboratory are gratefully acknowledged. Research in the laboratory is funded by grants from the Health Research Board (Ireland), the Irish Research Council for Science, Engineering and Technology and the Hospital Infection Society (UK).