SarA and not σB is essential for biofilm development by Staphylococcus aureus

Authors


Summary

Staphylococcus aureus biofilm formation is associated with the production of the polysaccharide intercellular adhesin (PIA/PNAG), the product of the ica operon. The staphylococcal accessory regulator, SarA, is a central regulatory element that controls the production of S. aureus virulence factors. By screening a library of Tn917 insertions in a clinical S. aureus strain, we identified SarA as being essential for biofilm development. Non-polar mutations of sarA in four genetically unrelated S. aureus strains decreased PIA/PNAG production and completely impaired biofilm development, both in steady state and flow conditions via an agr-independent mechanism. Accordingly, real-time PCR showed that the mutation in the sarA gene resulted in downregulation of the ica operon transcription. We also demonstrated that complete deletion of σB did not affect PIA/PNAG production and biofilm formation, although it slightly decreased ica operon transcription. Furthermore, the sarA-σB double mutant showed a significant decrease of ica expression but an increase of PIA/PNAG production and biofilm formation compared to the sarA single mutant. We propose that SarA activates S. aureus development of biofilm by both enhancing the ica operon transcription and suppressing the transcription of either a protein involved in the turnover of PIA/PNAG or a repressor of its synthesis, whose expression would be σB-dependent.

Introduction

Staphylococcus aureus is one of the most frequent bacterial causes of community-acquired and hospital-acquired infections. One reason for the ubiquitous nature of infections caused by this pathogen is the easy transmission from the nasal membranes and skin to wounds where bacteria produce numerous extracellular proteins and toxins. In addition, S. aureus has the capacity to adhere to catheters and other indwelling devices and form a multicellular community, known as a biofilm, that is then difficult to combat with host defences or antibiotics [for reviews, see Götz and Peters (2000) and Costerton and Stewart (2000)].

Staphylococcus aureus biofilm formation is mediated by the production of the extracellular polysaccharide adhesin PIA/PNAG, whose synthesis depends on the expression of icaADBC-encoded enzymes (Cramton et al., 1999; Mckenney et al., 1999; Gotz, 2002; Maira-Litran et al., 2002). Most of the S. aureus strains analysed so far contain the entire ica gene cluster (Cramton et al., 1999; Arciola et al., 2001; Martin-Lopez et al., 2002), but only a few express the ica operon and produce biofilms in microtitre dish-based assays. These differences could be due to the fact that ica expression is subject to environmental regulation. Stimuli such as high osmolarity (3% NaCl), growth in anaerobic conditions, high temperature and subinhibitory concentrations of certain antibiotics are known to enhance ica transcription and biofilm formation (Rachid et al., 2000a; Cramton et al., 2001). Moreover, ica operon expression in the closely related S. epidermidis can be turned on and off by the insertion and excision of the insertion sequence IS256 at specific hot spots of the icaA and icaC genes (Ziebuhr et al., 1999). By this mechanism, PIA/PNAG production and biofilm formation phenotypes may be phase variable, allowing individual S. epidermidis cells to leave the biofilm and colonize new surfaces. However, whether a similar mechanism could explain phase variation in S. aureus (Baselga et al., 1993) is unknown.

Furthermore, DtlA and Bap proteins have been identified as contributors to S. aureus formation of biofilm. The dtlABCD operon is responsible for the d-alanine incorporation in the teichoic acids. The lack of d-alanine esters causes a strongly negative net charge on the bacterial surface that affects primary attachment to polystyrene or glass surfaces (Gross et al., 2001). On the other hand, the surface protein Bap (Cucarella et al., 2001) interferes with initial bacterial attachment of other surface molecules (MSCRAMM) to host receptors and cellular internalization (Cucarella et al., 2002). In contrast to the dtlA and ica operons, all the staphylococcal isolates harbouring bap are highly adherent and strong biofilm producers, indicating a strong correlation between the presence of the protein and biofilm formation ability on abiotic surfaces (Cucarella et al., 2001).

The regulatory mechanisms involved in the S. aureus biofilm development remain poorly understood. A recent report shows that icaR, located adjacent to the ica operon, encodes a transcriptional repressor involved in environmental regulation of the ica operon expression in S. epidermidis (Conlon et al., 2002). In addition, insertional mutagenesis and complementation experiments have demonstrated that transcription of the ica operon is at least partially controlled by the alternative transcription factor σB (Rachid et al., 2000b). The absence of a nucleotide sequence immediately upstream of the ica operon, resembling the consensus sequence recognized by σB, suggests the existence of additional regulatory factors. Global quorum sensing regulators have been shown to regulate the biofilm formation of many Gram-negative bacteria, including Pseudomonas aeruginosa (Davies et al., 1998), Burkholderia cepacia (Huber et al., 2001) and Pantoea stewartii (von Bodman et al., 1998). Accordingly, it has been proposed that biofilm formation by S. aureus may be regulated in a density-dependent manner by loci such as the accessory gene regulator (agr ) and the staphylococcal accessory regulator (sarA). The agr regulator, a two-component quorum sensing system activated during the transition from the exponential to the stationary growth phase, is known to mediate the expression conversion from genes encoding cell-surface proteins to genes encoding exoproteins. The presence of agr seems to be negatively correlated with the ability to adhere to polystyrene (Vuong et al., 2000). With respect to the sar regulator, the sar locus encodes a 14.5 kDa protein that binds to the agr promoter to stimulate RNAIII transcription, resulting in the modulation of target genes downstream on the agr regulatory cascade (agr-dependent pathway). Additionally, SarA can interact directly with target gene promoters to control gene expression (agr-independent pathway). Previous work has established that mutation in sarA may confer an enhanced adherence to inert surfaces compared to wild-type or agr mutant strains, suggesting that SarA might be a negative regulator of biofilm development (Pratten et al., 2001).

In this study, we have carried out a random transposon mutagenesis to further investigate regulatory genes involved in the process of biofilm formation by S. aureus. One of the isolated mutants was shown to carry the transposon within the sarA gene. Analysis using four genetically non-related S. aureus strains revealed that SarA deficiency decreased ica operon transcription as well as PIA/PNAG accumulation, and impaired biofilm development both in steady state and flow conditions. We have also demonstrated that neither σB nor agr are essential for biofilm development. A simple regulatory model on the role of SarA in the ica regulatory circuitry is presented and discussed.

Results

Identification of genes required for biofilm formation

Repeated passages of bacterial isolates in liquid culture may select for fast growth and loss of multicellular behaviour attributes. Thus, we decided to investigate biofilm formation processes of a clinical isolate of S. aureus 15981. For that, we constructed a collection of approximately 10000 transposon mutations from three independent mutagenesis and screened for mutants that were deficient in biofilm formation on microtitre polystyrene plates in B2 and TSB-gluc media. Sixteen transposon mutants, exhibiting growth rates indistinguishable from the wild type, were found to be deficient in biofilm formation in TSB-gluc, but only five were also unable to produce a biofilm in B2. Interestingly, these five mutants produced smooth colonies in Congo red agar plates and were subsequently selected for further analysis. Sequencing of regions adjacent to the transposon insertion site and homology searches using the blast 2.0 program at the NCBI server showed that insertion sites were localized in three loci: one affected the expression of SarA protein (M6); one involved a locus called pnpA (M3) and three affected different genes in the icaADBC operon already known to be implicated in the biofilm formation process (M1, 4, 5). The finding that three independent mutants affected the icaADBC operon suggested that we had performed a saturating mutagenesis screening (see Supplementary material, Table S1 for the identification of the mutants deficient in biofilm development). Further analysis of the role of sarA in the biofilm formation process is presented below. Analysis of the role of pnpA will be presented elsewhere.

Table 1. . Strains and plasmids used in this study.
Strain and plasmidRelevant characteristicsSource and reference
S. aureus
 15981Clinical strain. Biofilm positive. rsbU+This study
 M615981 sarA::Tn917This study
 M315981 pnpA::Tn917This study
 ΔsarA15981 ΔsarAThis study
 ΔsarA-c15981 ΔsarA complemented with sarA geneThis study
 ΔσB15981 ΔσBThis study
 ΔsarA-ΔσB15981 ΔsarA-ΔσBThis study
 ΔsarA-ΔσB-sarA15981 ΔsarA-ΔσB complemented with sarA geneThis study
 ΔsarA-ΔσBB15981 ΔsarA-ΔσB complemented with pSK9This study
 ΔsarA-ΔσB-ica15981 ΔsarA-ΔσB complemented with pSC18This study
 Δagr15981 ΔagrThis study
 8325–4Wild-type strain 8325 cured of phages φ11, φ12 and φ 13 Novick (1990)
 PC18398325–4 sarA::km Chan and Foster (1998)
 RN69118325–4 agr::tet Novick et al. (1993)
 PC4008325–4 sigB::tet Chan and Foster (1998)
 RN4220A mutant of 8325–4 that accepts foreign DNA Novick (1990)
 AK18325–4 aur::ermB Karlsson et al. (2001)
 AK28325–4 ssp.: ermB Karlsson et al. (2001)
 ISP479cISP479 cured of plasmid. rsbU- Pattee (1981)
 ISP479 sarA-ISP479c sarA::kmThis study
 ISP479 agr-ISP479c agr::tetThis study
 ISP479 σB-ISP479c sigB::tetThis study
 ISP479 sarA-σB-ISP479c sarA::km sigB::tetThis study
 ISP479 sarA- aur-ISP479 sarA::km aur::ermBThis study
 ISP479 sarA- ssp-ISP479 sarA::km ssp: ermBThis study
 V329Bovine subclinical mastitis isolate. Biofilm positive. rsbU + Cucarella et al. (2001)
 V329 ΔsarAV329 ΔsarAThis study
 V329 ΔagrV329 ΔagrThis study
 MA12Mucosal isolate. Biofilm positive. rsbU + Rachid et al. (2000b)
 MA12 ΔsarAMA12 ΔsarAThis study
Plasmid
 pID408 E. coli – S. aureus shuttle vector with a termosensitive origin of replication for Gram-positive bacteria that contains the transposon Tn917 Mei et al. (1997)
 pMAD E. coli – S. aureus shuttle vector with a termosensitive origin of replication for Gram-positive bacteriaM. Arnaud and M. Debarbouille, unpublished
 pCU1Vector for complementation experiments Augustin et al. (1992)
 pSK9Plasmid that carries σB gene. Rachid et al. (2000b)
 pSC18Plasmid that carries icaADBC genes Cramton et al. (1999)

Effect of sarA deletion among unrelated S. aureus isolates

To further support the role of the sarA gene in biofilm formation and to exclude spontaneous mutations or polar effects of the transposon insertion on downstream genes, a strain carrying a complete deletion of the sarA gene (ΔsarA) was constructed by allelic exchange. As shown in Fig. 1, the deletion mutant, akin to the sarA transposon mutant, showed a smooth colony morphology phenotype in Congo Red agar (Fig. 1A), lost the ability to produce a ring of cells adhered to the glass wall at the air–liquid interface (Fig. 1B) and was unable to form a biofilm on polystyrene and PVC microtitre plates (Fig. 1C and D). To confirm that the biofilm-deficient phenotype was due to the deletion of sarA, ΔsarA mutant was complemented with plasmid pCU1 carrying a PCR amplified 1349 bp fragment containing the sarA gene under the control of its own promoter. The complemented strain (ΔsarA-c) displayed a biofilm phenotype similar to that of the wild-type strain (Fig. 1A–D). In order to show the independence of the biofilm deficient phenotype with respect to the strain background, we also produced a non-polar sarA deletion in the genetically unrelated S. aureus MA12 and V329 strains and an insertional mutation in the ISP479c strain. The link between the sarA mutation and the absence of the biofilm formation ability was confirmed in all the strains (Fig. 2).

Figure 1.

Characterization of various biofilm phenotypes of S. aureus 15981, ΔsarA and ΔsarA-c.
A. Comparison of colony morphology on Congo red agar plates at 37°C for 24 h.
B. Biofilm formation on a glass surface of an overnight culture incubated with shaking.
C. Biofilm formation in PVC microtitre dishes of cells grown for 24 h in TSB-gluc at 37°C without shaking.
D. Crystal violet-stained, surface-attached cells were quantified by solubilizing the dye in ethanol/acetone and determining the absorbance at 595 nm.

Figure 2.

Biofilm formation phenotype of four genetically unrelated S. aureus strains and their respective sarA mutants. Biofilm formation capacity of S. aureus 15981, V329, MA12 and ISP479c and their corresponding sarA mutants on polystyrene microtitre plates after 24 h hours in TSB-gluc medium at 37°C. The bacterial cells were stained with crystal violet and were quantified by solubilizing the dye in ethanol/acetone and determining the absorbance at 595 nm.

Biofilm formation ability of the sarA mutant in continuous flow culture microfermenters

In order to rule out the possibility that inhibition of biofilm formation in the ΔsarA mutant was a result of the accumulation of products during the steady state conditions in the microtitre plates, we analysed the biofilm formation capacity in a continuous flow culture on pyrex slides submerged in microfermenters (Ghigo, 2001). The flow rate of fresh medium (40 ml h−1) imposed in the process was high enough to avoid any significant planktonic growth. As shown in Fig. 3, strains 15981 and the complemented ΔsarA-c adhered abundantly to the glass substratum and formed a thick biofilm in 8 h. In contrast, after 24 h the ΔsarA strain only formed microcolonies on the surface of the slide without much biofilm development thereafter. This result demonstrated that SarA affects biofilm formation under flow conditions.

Figure 3.

Biofilm formation phenotype of S. aureus 15981, ΔsarA and ΔsarA-c in microfermenters. Left, biofilm development of bacteria grown in TSB-gluc at 37°C for 8 h and 24 h in microfermenters. Right, biofilm phenotype on the corresponding Pyrex slides removed from the microfermenters after 24 h of incubation in TSB-gluc at 37°C.

Regulation of biofilm formation by agr

SarA is a positive regulator of the agr operon and therefore influences the regulation of various virulence factors in an agr-dependent pathway. We therefore speculated whether SarA might affect biofilm formation indirectly, by acting through agr. To test this hypothesis, we created a complete deletion of the agr operon in the 15981 and V329 strains and an insertional mutation in the ISP479c strain. Interestingly, the biofilm formation capacity on microtitre plates or in continuous flow cultures of the resulting agr deficient strains was similar to the corresponding wild-type strains, clearly showing that SarA affects biofilm formation via an agr-independent pathway (data not shown).

Relationship between the extracellular proteolytic activity and the biofilm formation capacity of the sarA mutant

It is well established that production of proteases is up-regulated in sarA mutants (Chan and Foster, 1998). It seems therefore reasonable to suppose that inhibition of biofilm formation in sarA mutants could be due to the degradation of a surface protein required for biofilm formation. To investigate this possibility we first verified the increased protease production of sarA mutants in milk agar plates. Figure 4A shows that, as expected, sarA mutants overproduces proteases. We then performed primary attachment assays on microtitre plates with wild-type and sarA mutants grown in the presence of α2-macroglobulin, a universal protease inhibitor that inhibits the activity of all major staphylococcal proteases (McGavin et al., 1997; Karlsson et al., 2001) and E64, a cysteine protease inhibitor, that inhibits two of the major S. aureus proteases (SspB and Scp). No significant differences were found on the primary attachment ability of ΔsarA cells grown in the presence or absence of α2-macroglobulin or E64 (Fig. 4B), strongly suggesting that the biofilm formation deficiency of ΔsarA was not due to the increased production of extracellular proteases. In addition, wild-type 15981 and ISP479c cells were incubated with a concentrated supernatant from a stationary phase culture of their corresponding sarA mutant strain. Prolonged incubation of the cells with the supernatant did not affect biofilm formation capacity (data not shown). To prove that the major proteases were not responsible for biofilm deficiency, we constructed a sarA mutant harbouring a deletion of either aureolysin gene (aur) or the serine protease gene (ssp) and determined the ability of the double mutants to produce a biofilm. Because of the facility in genetically manipulating ISP479c strain we constructed these mutations only in this strain. The sarA- aur- and sarA- ssp- double mutants displayed a similar biofilm deficiency to that of the sarA mutant (Fig. 4C). Taken together, these data strongly suggest that extracellular proteases were not responsible for the biofilm deficiency of sarA mutant cells.

Figure 4. Effect of protease inhibition on the biofilm formation phenotype. A.

Figure 4. Effect of protease inhibition on the biofilm formation phenotype. A.

Protease production by S. aureus 15981 and ISP479c and their corresponding sarA mutants grown in skimmed milk agar plates.
B. Comparison of the primary attachment ability of S. aureus 15981 and ISP479c and their corresponding sarA mutant strains cultured in TSB-gluc (open bar), TSB-gluc with 0.4 U of α2-macroglobulin ml−1 (filled bar), or TSB-gluc with 10 µM of E64 (hatched bar).
C. Upper part, zones of proteolysis around S. aureus ISP479c, sarA–, sarA–ssp– and sarA–aur– mutant strains grown on skimmed milk agar plates. Lower part, comparison of the ability to produce a biofilm on polystyrene microtitre dishes of S. aureus ISP479c, sarA–, sarA–ssp– and sarA–aur– mutant strains grown in TSB-gluc at 37°C without shaking.

Effect of SarA on production of PIA/PNAG

We hypothesized that the role of SarA in biofilm formation could be the activation of the expression of PIA/PNAG. To investigate whether PIA/PNAG production was altered in the sarA mutant strains, the production of PIA was monitored by dot-blot using specific anti-PIA/PNAG polyclonal antisera. Our results showed that whilst significant amounts of PIA/PNAG could be detected in young cultures (OD650nm = 0.2) in the wild-type strains, PIA/PNAG production was substantially reduced in the ΔsarA mutants as compared to the parental strains at all phases of growth (Fig. 5). The reduction level was particularly less pronounced in the ISP479 sarA mutant strain. The fact that sarA mutants are still able to produce detectable amounts of PIA/PNAG indicates that SarA is relevant but not essential for PIA/PNAG production. Altogether, these results strongly suggest that the biofilm deficient phenotype of the ΔsarA mutants may be at least partially attributable to a decrease in PIA/PNAG levels.

Figure 5.

Dot blot analysis of PIA-PNAG accumulation in S. aureus wild-type strains and their corresponding sarA mutants at different points of the growth curve. Cell surface extracts at different points of the growth curve of S. aureus 15981 and ISP479c and their corresponding sarA mutants were treated with proteinase K and spotted onto nitrocellulose filters. PIA/PNAG production was detected with an anti-S. aureus PIA/PNAG antiserum. The sarA mutant strains produced lower levels of PIA/PNAG product.

Transcriptional analysis of the ica operon expression in sarA-deficient strains

We used real-time quantitative PCR to investigate whether the decrease in PIA/PNAG production observed in ΔsarA mutant was caused by a reduction of the ica operon expression. Total RNA of strains 15981 and ISP479c and their corresponding sarA mutants was isolated at early exponential and mid-log exponential phases, as the expression of ica operon is maximum during this period. After treatment with DNAase to remove contaminant DNA, RNA was reverse transcribed in the presence and absence of reverse transcriptase. The level of expression of icaA was normalized on gyrB expression (Wolz et al., 2002). Only samples with no amplification of gyrB in the minus reverse transcriptase aliquot were included in the study. The results showed that the sarA mutation resulted in a significant (P < 0.05) decrease of ica operon transcription compared to that of the wild-type strains at exponential and mid-log exponential phases (Fig. 6A). Similar results were obtained with RNA purified at early stationary phase (OD650 = 2) (data not shown). These results were confirmed by multiplex PCR with simultaneous amplification of icaC and gyrB (Fig. 6B). Because icaADBC operon is still transcribed, we theorize that SarA is an activator of the ica operon expression but that SarA activity alone cannot account for the total ica operon expression, thus suggesting that an additional factor(s) may be involved.

Figure 6.

Analysis of ica expression on S. aureus wild-type strains and their corresponding sarA mutants at different points of the growth curve.
A. Real time quantification of icaA expression. The gyrase B transcript was used as endogenous control and results are expressed as n-fold difference relative to the control gene. The figure represents the mean and standard deviation of five independent experiments. Asterisks denote P < 0.05.
B. Multiplex PCR of icaC and gyr B on S. aureus wild-type and sarA mutant strains at mid-log exponential phase (OD650 = 0.8). An aliquot of cDNA prepared in the absence of reverse transcriptase (–RT) (see Experimental procedures) is included to confirm the absence of genomic DNA.

Effect of σB mutation in a sarA-deficient background

It has been reported that a mutation of either σB or rsbU, an activator of σB, results in a complete abolition of salt-induced biofilm formation and a drastical decrease of ica transcription (Rachid et al., 2000b; Knobloch et al., 2001). It has also been proven that one of the sarA promoters (P3) is σB-dependent (Deora et al., 1997; Manna et al., 1998; Bischoff et al., 2001). We therefore speculated whether σB may modulate sarA expression and the ensuing ica transcription. If such an association existed, double mutants carrying ΔsarA and ΔσB mutations and single mutants in any of these regulators would display the same decrease in ica transcription. We therefore produced a complete deletion of σB (ΔσB) by allelic exchange in 15981 and ΔsarA and insertional mutation in the ISP479c and ISP479 sarA-. Remarkably, σB mutants: (i) formed large clumps in broth cultures and displayed a rough colony morphology in Congo Red agar plates (data not shown); (ii) retained the ability to form a biofilm on microtitre plates (Fig. 7A and B) and microfermenters; and (iii) produced PIA/PNAG levels similar to those of the wild-type strains (Fig. 7C). Interestingly, the sarA-σB double mutant displayed a higher capacity to produce a biofilm and an enhanced PIA/PNAG production compared with the sarA mutants, though never reached parental strain levels. Analysis of the transcriptional activity of the ica operon by real time PCR showed that the level of icaA transcripts was slightly reduced in σB mutants (Fig. 7D). In contrast to the results of PIA/PNAG production, double sarA-σB mutants showed a significant reduction (P < 0.05) in the ica operon transcription compared to that of the sarA mutants (Fig. 7D). These results were confirmed by multiplex PCR with simultaneous amplification of icaC and gyrB (Fig. 7E). Complementation experiments of the ΔsarA and sarA-σB mutants with a recombinant shuttle plasmid carrying either sarA, σB or icaADBC operon demonstrated that only SarA was able to restore biofilm formation in both strains. Interestingly, complementation in trans with a multicopy plasmid containing the complete ica operon slightly increased biofilm development in sarA-σB mutant and not in ΔsarA strain (data not shown).

Figure 7.

Analysis of biofilm formation, PIA/PNAG production and ica expression on S. aureus wild-type, sarA, σB and sarA-σB mutant strains.
A. Biofilm formation in polystyrene microtitre dishes of S. aureus 15981 and their corresponding ΔsarA, ΔσB and ΔsarA-ΔσB mutant strains grown for 24 h in TSB-gluc at 37°C without shaking.
B. Crystal violet-stained, surface-attached cells were quantified by solubilizing the dye in ethanol/acetone and determining the absorbance at 595 nm.
C. Dot blot analysis of PIA-PNAG accumulation on S. aureus 15981 and their corresponding ΔsarA, ΔσB and ΔsarA-ΔσB mutant strains at different points of the growth curve.
D. Real time quantification of icaA expression on S. aureus 15981 and ISP479c strains and their corresponding sarA, σB and sarA-σB mutant strains at mid-log exponential phase (OD650 = 0.8).
E. Quantification of icaC expression by multiplex PCR of icaC and gyrB on S. aureus 15981 and ISP479c strains and their corresponding sarA, σB and sarA-σB mutant strains at mid-log exponential phase (OD650 = 0.8). An aliquot of cDNA prepared in the absence of reverse transcriptase (– RT) (see Experimental procedures) is included to confirm the absence of genomic DNA.

Discussion

Bacteria seem to initiate biofilm development in response to a variety of environmental signals, such as nutrient and oxygen availability, osmolarity, temperature or pH. Obviously, the transition from planktonic to multicellular lifestyle requires the co-ordinated expression of a variety of specialized extracellular and cellular components. With the aim of identifying gene regulators that co-ordinate the biofilm formation process of S. aureus, we used the strategy of studying mutants defective in biofilm formation in two laboratory media (TSB-gluc and B2). With a similar strategy we were able to identify in a previous work two classes of genes involved in Salmonella enteritidis biofilm formation (Solano et al., 2002). The first class included genes involved in the synthesis of the exopolysaccharidic matrix that was required for biofilm formation in different nutritional conditions. The second class included genes involved in many different pathways that were required to form a biofilm in some media but not in others. Similarly, two classes of genes involved in the biofilm formation process of E. coli have been identified using different culture media (Danese et al., 2000). Also in this study, genes associated with S. aureus biofilm production were classified in two groups, as 11 of the 16 mutants detected produced a biofilm in B2 but not in TSB-gluc media. The most striking difference between B2 and TSB-gluc media is the NaCl content (2.5% in B2 versus 0.5% in TSB-gluc). Osmotic stress conditions similar to those of B2 (3% NaCl) significatively stimulated S. aureus biofilm formation (Rachid et al., 2000b). This suggests that in an environment favourable for biofilm formation, such as B2 for S. aureus, only indispensable genes are required for biofilm development. In contrast, bacteria in a less favourable environment would additionally require a specific set of genes for sensing and responding to the particular nutritional characteristics. Therefore, we propose that sarA and ica genes represent indispensable common genes required for S. aureus biofilm formation in different environmental conditions.

Previous work established that sarA was a negative regulator of biofilm development, as sarA mutation increased adherence to glass (Pratten et al., 2001). In contrast to these results, we have demonstrated that complete deletion of the sarA gene in four genetically non-related S. aureus strains completely abolished biofilm formation capacity by an agr-independent pathway. An explanation to this discrepancy could be that Pratten et al. (2001) compared the biofilm formation ability of mutants produced apparently on the same strain, but generated in different laboratories instead of comparing mutants created in the same 8325–4 isolate. Biofilm formation of isolate 8325–4 varies according to the laboratory source (our unpublished results). Therefore, we surmise that their conclusions could be influenced by the fact that isolates of 8325–4 from different laboratories could have different abilities to produce a biofilm.

In the present work, an examination of ica mRNA production in the wild-type strain and the sarA mutant at various growth phases by real-time and multiplex PCR showed that the ica operon transcription decreased in sarA mutant cells. A consensus-directed search strategy for identifying the conserved motif essential for sarA-dependent gene regulation (Chien et al., 1999) in the intergenic region between icaADBC and icaR revealed the presence of a sequence 70 nt upstream the start codon with 58% homology (15 nucleotides identity along 26 overlap) to the predicted sarA-like recognition sequence. This finding suggests that SarA may interact directly with the ica promoter and induce ica transcription. Alternatively, SarA might affect the expression of a regulatory protein that induces the transcription of the ica operon downstream of SarA. One such candidate is IcaR (Conlon et al., 2002). IcaR appears to be a member of the tetR family of transcriptional regulators and encodes for a repressor of ica operon transcription under certain environmental conditions in S. epidermidis. Specifically, IcaR is required for ethanol activation of the ica operon, but does not affect the activation of ica operon expression induced by NaCl-glucose in S. epidermidis (Conlon et al., 2002). It seemed reasonable to predict that sarA up-regulates ica transcription by repressing icaR expression. However, preliminary studies on icaR expression in the sarA mutant strain (our unpublished results) did not support this hypothesis and indicated that other regulatory elements may be involved in the ica operon repression. Other candidate regulators are SarH1 (also called SarS) (Tegmark et al., 2000; Cheung et al., 2001), SarT (Schmidt et al., 2001) and Rot (McNamara et al., 2000). Both SarH1 and SarT belong to a family of SarA homologues whose expression is regulated by both sarA and agr. Thus, sarA mutants constitutively express high levels of SarH1 and SarT, which could affect the expression of target genes either as activators or as repressors. However, SarH1 and SarT are also overproduced in an agr mutant and therefore we would expect in this mutant a similar decrease in the biofilm phenotype. Because the agr mutants that we have produced are not affected in biofilm formation, it seems unlikely that either SarH1 or SarT would be the repressors of the ica operon in the sarA mutant. The other candidate regulator, rot, encodes a repressor that affects the transcription of extracellular virulence factors during the lag and exponential phase of bacterial growth (McNamara et al., 2000). Intringuingly, we have observed that insertional mutation of rot impaired biofilm formation in TSB-gluc but did not affect biofilm formation in B2 (Supplementary material, Table S1), strongly suggesting that under certain environmental conditions the Rot protein could be one of the factor(s) involved in the downstream control of biofilm development.

The alternative transcription factor σB controls, in conjunction with additional regulators, the differentiation pattern associated with starvation, several stress factors and cell entry into stationary phase. It has been shown that σB, among others, transcriptionally regulates sarA (Deora et al., 1997; Kullik et al., 1998; Manna et al., 1998; Bischoff et al., 2001) and ica operon transcription (Rachid et al., 2000b). In this respect, and based on our results on ica operon transcription in sarA mutants, we performed additional studies to establish the relationship between σB, sarA and ica operon. Our results showed that ica transcription was slightly decreased in the σB mutant. However, it did not affect either PIA/PNAG production or biofilm development. This finding apparently contrasts with a previous report (Rachid et al., 2000b) that suggested that σB was required for biofilm formation. In the article of Rachid et al. (2000b) the strain MA12 and its corresponding σB mutant (MA12.2) produced similar low levels of biofilm in TSB media, whereas in the presence of 3% NaCl the biofilm production was only increased in strain MA12. On the other hand, the levels of PIA/PNAG were not measured and nothing was concluded on this matter. The strains used in our study were able to produce biofilm without being subjected to osmotic stress. In this situation σB did not significatively affect PIA/PNAG production and biofilm formation although a sigB mutation caused a slightly decrease in ica operon transcription.

Aditionally, we observed that ica expression was significantly decreased in the sarA-σB double mutant compared to ΔσB mutant. Remarkably, the levels of PIA/PNAG and biofilm formation capacity increased in this double mutant compared to those of the ΔsarA strain. Overall, our findings on ica transcription (wild-type> σB> ΔsarA > sarA-σB–), PIA/PNAG production (σB≥ wild-type>sarA-σB–> ΔsarA) and biofilm formation indicate that an increase in ica transcription does not necessarily lead to an increase in PIA/PNAG and biofilm production. Both SarA and σB appear to promote ica operon transcription but may have a reverse effect in PIA/PNAG and biofilm production: σB mutation leads to a decrease in ica transcription but to maintenance or increase of PIA/PNAG and biofilm production. On the other hand, SarA mutation leads to a decrease in ica transcription and PIA/PNAG production which is at least partially counteracted by a σB mutation.

Thus, emerging from this work and aligned with previous findings (Conlon et al., 2002), a possible regulatory mechanism of PIA/PNAG production and biofilm formation could be proposed (Fig. 8). Both SarA and σB upregulate ica operon expression at the transcriptional level; whether this regulation is direct or indirect remains unclear. In addition, SarA may affect PIA/PNAG production upon suppression of an unknown element that would either degrade PIA/PNAG product or repress the PIA/PNAG synthesis. σB would upregulate the expression of the unknown element. IcaR downregulates ica transcription in response to specific environmental conditions probably in concert with additional factor(s).

Figure 8.

Summary illustrating the role of SarA in the regulation pathway leading to the biofilm formation. SarA and σB both activate ica transcription either directly or indirectly. In addition, SarA may affect PIA/PNAG production upon suppression of an unknown element that would either degrade PIA/PNAG product or repress the PIA/PNAG synthesis. σB would upregulate the expression of the unknown element. IcaR represses ica transcription possibly in concert with an additional factor(s) probably located downstream the sarA pathway

To the well-characterized roles of the SarA protein in the regulation of many virulence determinants, we have now added a novel one in the upregulation of biofilm development. The co-regulation of both processes suggests that multicellular behaviour may play a role in several previously identified infection models. SarA may thus be a promising target to simultaneously control the biofilm development on indwelling devices and virulence mediated by extracellular toxins.

Experimental procedures

Bacterial strains, culture conditions and plasmids

Staphylococcus aureus 15981, isolated at the Microbiology Department of the University Clinics of Navarra, was selected because of its strong biofilm production phenotype, antibiotic susceptibility profile and availability to accept recombinant DNA by electroporation. The most relevant bacterial strains and plasmids used and constructed in this study are listed in Table 1. Escherichia coli XL1-Blue cells were grown in Luria–Bertani (LB) broth or on LB agar (Pronadisa) with appropriate antibiotics. Staphylococcal strains were cultured on Trypticase soy agar (TSA), in trypticase soy broth supplemented with glucose (0.25% w/v) when indicated (TSB-gluc), in TSA supplemented with skimmed milk (5% w/v) when indicated (milk agar plates), in B2 broth (Schenk and Laddaga, 1992) and in Congo red agar (Baselga et al., 1993). Media were supplemented with appropiate antibiotics at the following concentrations: erythromycin (Er) 20 µg ml−1 or 1.5 µg ml−1, ampicillin (Am) 100 µg ml−1, choramphenicol (Cm) 20 µg ml−1, kanamycin (Km) 50 µg ml−1 and tetracyclin (Tet) 10 µg ml−1. The correspondence of the OD650 values with the growth curve is: early exponential phase (OD650 = 0.4); mid-log exponential phase (OD650 = 0.8); early stationary phase (OD650 = 2) and stationary phase (OD650 = 5, overnight culture).

DNA manipulations

DNA plasmids were isolated from E. coli strain using the Qiagen plasmid miniprep Kit, according to the manufacture's protocol. Plasmids were transformed into staphylococci by electroporation, using a previously described protocol (Cucarella et al., 2001). Restriction enzymes were purchased from Boehringer Mannheim and used according to the manufacturer's instructions. Oligonucleotides were obtained from Life Technologies (Table 2). sarA, aur, ssp, agr and σB genes were inactivated in S. aureus ISP479c transferring of the sar-, aur-, ssp-, agr and σB by phage transduction using Φ85 (Novick, 1991). For Southern hybridization, chromosomal DNA was purified as previously described (Marmur, 1961), digested and analysed by agarose gel electrophoresis. DNA fragments were transferred by alkaline capillary blotting onto nylon membranes (Schleicher and Schuell) using standard methods (Ausubel et al., 1990). Labelling of the probe and DNA hybridization were performed according to the protocol supplied with PCR-DIG DNA-labelling and chemiluminiscence detection kit (Boehringer Mannheim).

Table 2. . Oligonucleotides used in this study.
Oligonucleotide 
pseqAGAGAGATGTCACCGTCAAGT
sarB1GGGGATCCAAAGCGTTGATTTGGGTA
sarB2GGGGATCCAGTGCCATTAGTGCAAAA
sar-AGATGATAGTGCGGTGG
sar-BGTTTAGCGCAATTTGG
sar-CTTCACCAAATTGCGCTAAACCCTCCCTATTTGATGCATCTTGCT
sar-DACCCGTTATCAATCGG
sigB-AGCAGTGTTAATACTGCTTC
sigB-BAGCACATAAATAGAATTTGC
sigB-CTAATAAGCAAATTCTATTTATGTGCTGCCATTGGTTAATTTGCTCAG
sigB-DGTTAATGAAGGAACGGAGG
agr-AAGCACTGAGTCCAAGG
agr-BTTTTACACCACTCTCC
agr-CGTGAGGAGAGTGGTGTAAAAAAGATAATAAAGTCAGTTAACGGC
agr-DCAGTTATTAGCAGGAT
aur-1CGATGGTGACAGTAATAAA
aur-2CGTTAATGCTCGGTAGTA
ssp-1CCCTTAATGTATTTGTAAA
ssp-2CCATGACGTATACAAATT
IcaAligth-1CTTGGATGCAGATACTATCG
IcaAligth-2GCGTTGCTTCCAAAGACCTC
IcaCligth-1ACACAGCGTTTCACGATACCG
IcaCligth-2CCAATAGTCTCCATTTGCTAACGC
GyrUTTATGGTGCTGGGCAAATACA
GyrLCACCATGTAAACCACCAGATA

Transposon mutagenesis and gene identification

Staphylococcus aureus 15981 was transformed with plasmid pID408 (Mei et al., 1997) by electroporation and random Tn917 insertions were carried out as described (Cucarella et al., 2001). Mutants deficient in biofilm formation were screened on microtitre polystyrene plates using two rich media, B2 and TSB-gluc. B2 broth is a rich medium with high salt content, in which we and others (Lee, 1995) noted that different S. aureus strains clumped and formed extremely tenacious biofilm on the surface of the glass container during preparation of electrocompetent cells (Schenk and Laddaga, 1992). To identify genes containing Tn917 insertions, chromosomal DNA of each mutant was digested with EcoRI and religated in 200 µl of ligation buffer (Promega) for 12 h at 14°C. The ligation products were transformed into E. coli XL1-Blue and selected onto LB agar containing Am. Plasmid DNA was extracted using a Qiagen plasmid miniprep kit. Chromosomal DNA sequences flanking the transposon were obtained using primer pseq (Table 2) which corresponds to the inverted repeat region located 70 bp from the erm-proximal end of Tn917. Nucleotide sequences were determined at the DNA Sequencing Service of the University of Navarra (Spain). Homology searches were carried out using blast 2.0 program (Altschul et al., 1997) at the NCBI server.

Allelic exchange of chromosomal genes

To construct the deletions, we amplified by PCR two fragments of 500 bp that flanked the left (oligonucleotides A and B) and the right (oligonucleotides C and D) of the sequence targeted for deletion (Table 2). The oligonucleotides B and C have at least a 16-base complementary region (underlined in the oligonucleotide sequence) to allow the products of the first PCR to anneal at their overlapping region. A second PCR was performed with primers A and D to obtain a single fragment. Specifically, 1 µl of each of the first PCR was mixed with 10 pM of the outside primers and PCR amplified. The fusion products were purified and cloned in the pGEM-T easy vector (Promega). The fragment was then cloned into the EcoRI site of the shuttle plasmid pMAD (M. Arnaud and M. Debarbouille, unpublished data) and the resulting plasmid was transformed into S. aureus by electroporation. pMAD contains a temperature-sensitive origin of replication and an erythromycin resistance gene. The plasmid was integrated into the chromosome through homologous recombination at non-permissive temperature (43.5°C). From the 43.5°C plate, one to five colonies were picked into 10 ml of TSB-gluc and incubated for 24 h at 30°C. Ten-fold serial dilutions of this culture in sterile TSB-gluc were plated on TSA contain-ing X-gal(5-bromo-4-chloro-3-indolyl-B-d-galactopyranoside) (150 µg ml−1). White colonies, which no longer contained the pMAD plasmid, were tested to confirm the replacement by PCR using the oligonucleotides A and D and Southern blotting.

Complementation of the mutants

The sarA gene from S. aureus 15981 was amplified with high-fidelity thermophilic DNA polymerase (Dynazyme Ext, Finnzymes) with primers sarB1-sarB2 (Table 2). The PCR products were cloned into pCU1 (Augustin et al., 1992) and the resulting plasmid pCU1sar was transformed by electroporation into S. aureusΔsarA. Complementation experiments with wild-type σB and ica operon were performed using plasmids pSK9 and pSC18 respectively (Table 1).

Biofilm formation and primary attachment assays

Biofilm formation assay in microtitre wells was performed as described (Heilmann et al., 1996). For adherence assays to glass tubes, a single colony was transferred to 5 ml of TSB-gluc and incubated at 37°C in an orbital shaker (250 r.p.m) for 12 h. Primary attachment assays in the presence of protease inhibitors were performed as follows: S. aureus strains were grown overnight in TSB-gluc and diluted 1:100 in TSB-gluc containing either 0.4 U ml−1 of α2-macroglobulin (Sigma) or 10 µM of E64 {[l-trans-epoxysuccinyl-leucyl-amido-(4-guanidino)butane], Sigma}. Bacteria were incubated in the presence of the inhibitors until mid-log exponential phase (OD650 = 0.8). The culture was diluted to OD650 = 0.1 and 200 µl were used to inoculate sterile 96-well polystyrene microtitre plates (Iwaki). After 1 h at 37°C the wells were gently rinsed at least five times with phosphate-buffered saline (PBS), dried in an inverted position and stained with 0.1% of crystal violet for 15 min. The wells were rinsed again, and the crystal violet solubilized in 200 µl of ethanol-acetone (80 : 20 v/v). The optical density at 595 nm (OD595) was determined using a microplate reader (Multiskan EX; Labsystems). Each assay was performed in triplicate and repeated three times.

To analyse the biofilm formation under flow conditions we used 60-ml microfermenters (Pasteur Institute's Laboratory of Fermentation) with a continuous flow of 40 ml h−1 of TSB-gluc and constant aeration with sterile pressed air (0.3 bar). Submerged pyrex slides served as growth substratum. 108 bacteria from an overnight preculture grown in TSB-gluc of each strain were used to inoculate microfermenters and were cultivated 24 h at 37°C. Biofilm development was recorded with a Nikon Coolpix 950 digital camera.

PIA/PNAG detection

PIA/PNAG production in S. aureus 15981 and mutants was detected as described (Cramton et al., 1999). Overnight cultures of S. aureus strains were diluted 1:100 and grown to early exponential phase (OD650 = 0.4); mid-log exponential phase (OD650 = 0.8); early stationary phase (OD650 = 2) and stationary phase (OD650 = 5, overnight culture). Cells were grown TSB-gluc until the appropiate optical density. The same number of cells of each culture was resuspended in 50 µl of 0.5 M EDTA (pH 8.0), incubated for 5 min at 100°C and centrifuged to pellet them. 40 µl of the supernatant was incubated with 10 µl of proteinase K (20 mg ml−1; Sigma) for 30 min at 37°C. After addition of 10 µl of Tris-buffered saline [20 mM Tris-HCl, 150 mM NaCl (pH 7.4)] containing 0.01% bromophenol blue, 5 µl was spotted on a nitrocellulose filter using a Bio-Dot Microfiltration Apparatus (Bio-Rad), blocked overnight with 5% skimmed milk in phosphate-buffered saline (PBS) with 0.1% Tween 20 and incubated for 2 h with an anti-S. aureus PIA/PNAG antibody diluted 1:10 000 (Mckenney et al., 1999). Bound antibodies were detected with a peroxidase-conjugated goat anti-rabbit immunoglobulin G (IgG) antibody (Jackson ImmunoResearch Laboratories) diluted 1:10 000, and the Amersham ECL Western blotting system.

Real-time quantitative PCR

Total S. aureus RNA was prepared using the Fast RNA-Blue kit (Bio101) according to the manufacturer's instructions. Two micrograms of each RNA were subjected in duplicate to DNase I (Gibco-BRL) treatment for 30 min at 37°C. The enzyme was inactivated at 65°C in the presence of EDTA. To verify the absence of genomic DNA in every sample, the RNA duplicates were reverse transcribed in the presence and absence of M-MLV Reverse Transcriptase (Gibco-BRL) respectively. All preparations were purified using CentriSep spin columns (Princeton Separations). One twentieth of each reaction was used for real-time quantitative PCR using a LightCycler and the LC-DNA Master SYBR Green I mix (Roche Diagnostics). The IcaA transcripts were amplified using primers IcaAlight-1 and IcaAlight-2 (Table 2). The gyrB transcripts that are constitutively expressed were amplified as endogenous control using primers GyrU and GyrL (Wolz et al., 2002). To monitor the specificity, final PCR products were analysed by melting curves and electrophoresis. Only samples with no gyrB amplification of the minus reverse transcriptase aliquot were considered in the study. The amount of icaA transcript was expressed as the n-fold difference relative to the control gene (2–ΔCT, where ΔCT represents the difference in threshold cycle between the target and control genes).

Multiplex PCR

One-tenth of each reverse transcriptase reaction was used for the simultaneous amplification of gyr B and icaC transcripts during 30 cycles at 55°C. The primers used for the amplification of icaC were IcaClight-1 and IcaClight-2 (Table 2).

Statistical analysis

The data corresponding to gene expression were compared using the Kruskal–Wallis and the Mann–Whitney tests. All the tests were two-sided and the significance level was 5%. The statistical analysis was performed with the spss program.

Acknowledgements

We express our gratitude to D. Mckenney for providing the anti-S. aureus PIA/PNAG antiserum; F. Götz for plasmid pCU1, D. W. Holden for plasmid pID408, M. Arnaud and M. Debarbouille for pMAD plasmid, R. Nagel for S. aureus ISP479c strain, A. Karlsson for AK1 and AK2 strains, S. E. Cramton for plasmid pSC18, W. Ziebuhr for MA12 strain and J. Leiva for S. aureus 15981 strain. We thank C. Solano for critical reading the manuscript. Jaione Valle and Alejandro Toledo-Arana are predoctoral fellows from the Ministerio de Ciencia y Tecnologia and Ministerio de Educación, Cultura y Deporte (FPI and FPU, respectively), Spain. This work was supported by the BIO99-0285 and HF2000-0057 grants from the Comisión Interministerial de Ciencia y Tecnología and the ‘Beca Ortiz de Landazuri’ grant from the Departamento de Salud del Gobierno de Navarra, Spain.

Supplementary material

The following material is available from http:www.blackwellpublishing.comproductsjournalssuppmatmolemole3493mmi3493sm.htm

Table S1 . Sequence analysis of transposon-tagged genes of selected biofilm deficient mutants.

Ancillary