Induction of Staphylococcus epidermidis biofilm formation via proteolytic processing of the accumulation-associated protein by staphylococcal and host proteases


E-mail; Tel. (+49) 404 2803 3147; Fax (+49) 404 2803 4881.


Because of its biofilm forming potential Staphylococcus epidermidis has evolved as a leading cause of device-related infections. The polysaccharide intercellular adhesin (PIA) is significantly involved in biofilm accumulation. However, infections because of PIA-negative strains are not uncommon, suggesting the existence of PIA-independent biofilm accumulation mechanisms. Here we found that biofilm formation in the clinically significant S. epidermidis 5179 depended on the expression of a truncated 140 kDa isoform of the 220 kDa accumulation-associated protein Aap. As expression of the truncated Aap isoform leads to biofilm formation in aap-negative S. epidermidis 1585, this domain mediates intercellular adhesion in a polysaccharide-independent manner. In contrast, expression of full-length Aap did not lead to a biofilm-positive phenotype. Obviously, to gain adhesive function, full-length Aap has to be proteolytically processed through staphylococcal proteases as demonstrated by inhibition of biofilm formation by α2-macroglobulin. Importantly, also exogenously added granulocyte proteases activated Aap, thereby inducing biofilm formation in S. epidermidis 5179 and four additional, independent clinical S. epidermidis strains. It is therefore reasonable to assume that in vivo effector mechanisms of the innate immunity can directly induce protein-dependent S. epidermidis cell aggregation and biofilm formation, thereby enabling the pathogen to evade clearance by phagocytes.


Staphylococcus epidermidis is the leading cause of infections related to implanted medical devices, i.e. intravascular catheters, joint prosthesis, artificial heart valves, and cerebrospinal fluid (CSF)-shunts (Rupp and Archer, 1994), making this organism one of the most important pathogens in nosocomial infections (Wisplinghoff et al., 2003; Karlowsky et al., 2004). Failure of antibiotic therapy is common, making the removal of the infected device necessary. S. epidermidis persist through biofilm formation – multilayered bacterial communities stabilized by intercellular adhesive mechanisms (Costerton et al., 1999; Chicurel, 2000; O’Toole et al., 2000). Biofilm formation is a two-step process in which the bacteria first adhere to the surface to be colonized (primary attachment), and subsequently accumulate into a multilayered cell architecture (accumulative phase) (Mack, 1999; Götz, 2002; Mack et al., 2004). Primary attachment is mediated by physico-chemical cell surface properties as well as specific factors like the major autolysin AtlE (Heilmann et al., 1997), the fibrinogen-binding protein Fbe (Nilsson et al., 1998), and the fibronectin-binding protein Embp (Williams et al., 2002). Following the accumulative phase most of the bacteria have no direct contact to the surface but remain in the biofilm via mechanisms mediating cell-to-cell adhesion. The polysaccharide intercellular adhesin (PIA), a homoglycan consisting of β-1,6-linked N-acetylglucosamine (Mack et al., 1996), mediates intercellular adhesion (Mack, 1999) and the icaADBC locus encoding the PIA synthesis apparatus (Heilmann et al., 1996) is widespread in clinical isolates (Galdbart et al., 2000). Using a well characterized isogenic biofilm-negative transposon mutant 1457-M10 in two animal models, it was demonstrated that a functional ica gene locus and PIA expression is important for the pathogenesis of S. epidermidis biomaterial-related infections (Rupp et al., 1999a,b). It was therefore anticipated that icaADBC-negative S. epidermidis represent prototypic, less virulent commensal strains (Zhang et al., 2003). Surprisingly, in several large, well controlled epidemiological studies investigating S. epidermidis populations isolated from foreign-body infections, a significant percentage of icaADBC- and biofilm-negative strains were found (Ziebuhr et al., 1997; Frebourg et al., 2000; Klug et al., 2003; Vandecasteele et al., 2003). As it is well accepted that in these types of infection S. epidermidis persists through biofilm formation one can assume that in these strains other than PIA-dependent mechanisms of biofilm formation must exist. The aim of the present study was to elucidate PIA-independent factors mediating biofilm formation in clinical significant icaADBC-, and PIA-negative S. epidermidis 5179 isolated from a persistent CSF-shunt infection (Rohde et al., 2001a).


During the course of a CSF-shunt infection we isolated the biofilm- and PIA-negative S. epidermidis 5179 (Rohde et al., 2001a), S. epidermidis 6317 and several further isolates of clonal origin from blood cultures. They all carried insertion sequence IS257 at bp 1301 in icaA (data not shown), appearing as a plausible cause for their biofilm- and PIA-negative phenotypes (Table 1). Enrichment for biofilm producing cells obtained biofilm-positive revertants of S. epidermidis 5179 and 6317 (designated 5179-R1 and 6317-R1 respectively). As only a twofold, apparently not substantial difference in primary attachment was found between S. epidermidis 5179 and 5179-R1 but the revertants formed large, macroscopically (data not shown) and microscopically visible cell clusters (Fig. 1A), the biofilm-positive phenotype obviously resulted not from altered primary attachment characteristics but from expression of intercellular adhesive properties. However, as the revertants still carried IS257 at the same position in icaA as the parent strains, resulting in inhibition of icaADBC transcription (data not shown), no PIA synthesis was detected in a coagglutination assay using a PIA-specific antiserum (Table 1). The involvement of an alternative polysaccharide could be ruled out as periodate oxidation had no effect neither on established biofilms of S. epidermidis 5179-R1 (Fig. 1B) nor on microscopic cell clustering (Fig. 1A). Additionally, transmission electron micrographs of 5179-R1 showed no exopolysaccharide structures (data not shown). In contrast, upon treatment with proteinase K, biofilms (Fig. 1B) and microscopic cell clusters (Fig. 1A) of S. epidermidis 5179-R1 were completely disintegrated, whereas the reverse is true for PIA-dependent biofilms of reference strain S. epidermidis 1457 (Fig. 1B) (Mack et al., 1992). Thus in the revertants intercellular adhesion and biofilm accumulation obviously relies essentially on proteinaceous structures, functionally substituting PIA.

Table 1.  Biofilm production and PIA synthesis by clinical S. epidermidis isolates 5179, 6317 and their respective revertants.
StrainSourceBiofilm (A570)aPIA-production
(reciprocal titre)b
  • a

    . Quantitatively determined by the biofilm assay detecting stained adherent cells in 96-well microtitre plates at a wavelength of 570 nm in an ELISA reader.

  • b

    . As detected by coagglutination with a PIA-specific antiserum.

  • c

    . Biofilm-positive revertants of the respective parent strain.

1457Central venous catheter infection2.5128
5179CSF-Shunt infection0.04  0
5179-R1c 2.5  0
6317CSF-Shunt infection0.03  0
6317-R1c 2.26  0
Figure 1.

A. Gram-stain of (I) S. epidermidis 5179 and (II) 5179-R1 cells derived from a planctonic culture in TSB; (III) S. epidermidis 5179-R1 cells after sodium-meta-periodate and (IV) proteinase K treatment.
B. Biofilms of S. epidermidis 1457 and 5179-R1 after overnight exposure against proteinase K (1 mg ml−1) and sodium-meta-periodate (40 mM). Water served as a control.

Analysis of S. epidermidis 5179 and 5179-R1 cell surface proteins

SDS-PAGE revealed a differential expression pattern of  cell  wall  associated  proteins  in  S.  epidermidis 5179 and 5179-R1 (Fig. 2A). Whereas S. epidermidis 5179 expressed a 180 kDa and a 220 kDa protein that were present in only small amounts in S. epidermidis 5179-R1, the revertant expressed an additional protein with an apparent molecular weight of 140 kDa that was not found in protein preparations of S. epidermidis 5179. Analysis of the differentially expressed proteins by mass spectrometry revealed that all peptides obtained from the 180 kDa and the 220 kDa as well as the 140 kDa protein could unambiguously be assigned to the accumulation-associated protein (Aap; accession number CAB77251) (Hussain et al., 1997). These results were confirmed by Western blot analysis of surface proteins using an antiserum raised against affinity-purified Aap that detected a 180 kDa and a 220 kDa protein in S. epidermidis 5179. In contrast, apart from the minor amounts of the 180 kDa and the 220 kDa protein an additional 140 kDa protein reacted with this antiserum in S. epidermidis 5179-R1 (Fig. 2B). To clarify the relation between distinct Aap species N-terminal sequencing was carried out, revealing in the 140 kDa protein an N-terminus that could be mapped to amino acid (aa) residue 596 of the amino acid sequence of S. epidermidis 5179 (Fig. 2C). These results demonstrate that Aap can be found in different forms on the S. epidermidis cell surface: as the full-length protein or in a truncated form (AapT; aa 596–1507), mainly consisting of the repetitive domain B (Fig. 2C). Importantly, the published N-terminal sequence of Aap (Hussain et al., 1997) does not represent the amino acid sequence deposit in the GenBank (accession number CAB77251) but that of AtlE (Heilmann et al., 1997) contaminating the Aap preparation. Failure to establish the sequence of the Aap N-terminus failed most probably because of amino-terminal blockade (H. Rohde, F. Buck, and D. Mack, unpubl.).

Figure 2.

A. Separation of cell surface-associated proteins from S. epidermidis 5179 and 5179-R1 by SDS-PAGE. S. epidermidis 5179-R1 expressed an additional 140 kDa protein (bold arrow) identified as Aap by N-terminal sequencing and mass spectrometry. Arrows indicate additional bands in 5179 and 5179-R1 also identified as Aap.
B. Immunoblot analysis of cell surface proteins from S. epidermidis 5179 and 5179-R1 using an antiserum raised against affinity-purified Aap. Double arrowhead, 220 kDa and 180 kDa band respectively; arrowhead, 140 kDa band.
C. Schematic representation of Aap from S. epidermidis 5179 as deduced from the aap nucleotide sequence (GenBank accession number AY359815). The N-terminus of the 140 kDa protein of S. epidermidis 5179-R1 is aligned to the predicted amino acid sequence of Aap from S. epidermidis 5179, identifying aa 596–609 as the N-terminal sequence (X, indeterminate amino acid). The position of truncation is located more C-terminal than a 30 aa deletion (indicated by an inverse arrow head) found in the aap allele of 5179-R1 (aapR1; GenBank accession number AY359816). Length polymorphism of Aap in different S. epidermidis strains results from variation of B repeat numbers (Hussain et al., 1997) (H. Rohde, M. Kalitzky, D. Mack, unpubl. results). Recombinantly expressed proteins are indicated by a black (domain A) and a grey (domain B) line. E, export signal; A, 564 aa domain; B, domain consisting of five (B1–B5) complete 128 aa and one (Bp) 68 aa partial repeat units; C, collagen triple helix motifs; L, LPXTG-motif containing cell wall anchor.

Analysis of aap in S. epidermidis 5179 and 5179-R1

Sequencing of aap from S. epidermidis 5179 (aap5179; GenBank accession number AY359815) revealed an open reading frame of 4521 nucleotides. The deduced amino acid sequence is characterized by an N-terminal export signal, a 564 aa domain A, sharing significant homology to Staphylococcus aureus surface protein SasG, and a domain B comprised by five repeats of 128 aa length, followed by an incomplete repeat of 68 aa. The C-terminus consists of two potential collagen triple helix repeats (aa 1076–1135 and aa 1136–1195) and an LPXTG motif containing Gram-positive cell wall anchor (Fig. 2C). Comparison of the aap nucleotide sequence of S. epidermidis 5179-R1 (aapR1; GenBank accession number AY359816) with aap5179 identified as the only difference an in-frame deletion ranging from bp 1202 to bp 1291, leading to loss of 30 amino acid residues (data not shown). As the deletion is located outside the truncated Aap (Fig. 2C) and sequencing of aap of an independent biofilm-positive revertant (6317-R1; Table 1) exhibiting the identical phenotypic characteristics as S. epidermidis 5179-R1, revealed no sequence alterations, the deletion obviously has no functional consequences on biofilm formation and a direct impact of the deletion on the function of the processed protein can be excluded.

Specific inhibition of S. epidermidis 5179-R1 biofilm formation

Evidence for the direct functional involvement of Aap in biofilm formation was gained by specific inhibition of S. epidermidis 5179-R1 biofilm formation using the antiserum raised against affinity-purified Aap which interfered with biofilm formation in dilutions up to 1:800 (Fig. 3A). In contrast, anti-Aap antiserum absorbed against S. epidermidis 5179 and 5179-R1 cells had no effect on biofilm formation respectively (Fig. 3A). This demonstrates that the relevant epitopes mediating biofilm formation in the revertant are already present on the cell surface of the biofilm-negative parent strain, whereas the N-terminal region of Aap missing in S. epidermidis 5179-R1 is obviously not functionally involved in biofilm formation. To further elucidate the functional role of distinct Aap regions we cloned domain A and B into pDEST17 and expressed them as His6-tagged proteins in Escherichia coli. Cultivation of S. epidermidis 5179-R1 in the presence of purified domain B inhibited biofilm formation in the biofilm assay in a concentration-dependent manner (Fig. 3B). The observed inhibition was significantly (P < 0.001; multiple-comparison test with a two-way analysis of variance) stronger than with purified domain A, which had little or no effect on the biofilm phenotype (Fig. 3B). Thus, domain B is obviously active in mediating intercellular adhesion and biofilm formation.

Figure 3.

A. Inhibition of S. epidermidis 5179-R1 biofilm formation using an antiserum raised against affinity-purified Aap and anti-Aap antiserum absorbed against S. epidermidis 5179 and 5179-R1 respectively. Relative inhibition was calculated by using the formula (1 − A570 with antiserum/A570 without antiserum) × 100. Serum from a rabbit before immunization served as a control.
B. Influence of recombinantly expressed Aap domain A and B on S. epidermidis 5179-R1 biofilm formation. Domain B inhibited biofilm formation significantly better than domain A (P < 0.001; multiple-comparison test with a two-way analysis of variance). Bovine serum albumin served as a negative control to exclude unspecific blocking of cell attachment. Relative inhibition of biofilm formation was calculated by using the formula (1 − A570 with recombinant protein/A570 without recombinant protein) × 100.
C. Biofilm phenotype of S. epidermidis 1585 and 1585 carrying pRBaap and pRBaapT respectively. Whereas S. epidermidis 1585 transformed with pRBaap remained biofilm-negative, the strain became biofilm-positive after transformation with pRBaapT, resulting in constitutive expression of AapT. Immunoblot analysis of cell surface proteins using an Aap-specific antiserum proved the expression of the different Aap species (180 kDa full-length Aap, double arrowhead; 140 kDa truncated Aap, arrow). In 1585 × pRBaap only the 180 kDa band is visualized. The diffuse banding of the 140 kDa protein detected in 1585 × pRBaapT probably results from adhesive properties of the protein, leading to aggregate formation.
D. Biofilm formation of S. carnosus TM300 carrying pASaapT. When grown in TSB without glucose supplemented with 0.4% (w/v) xylose the strain becomes moderately biofilm-positive. In contrast, when grown in TSB without glucose alone, no biofilm is produced.

Expression of aap in S. epidermidis 1585 and Staphylococcus carnosus TM300

To further characterize the role of Aap domain B in S. epidermidis 5179-R1 biofilm formation aap and a truncated isoform comprising the repetitive domain B (nucleotides 1786–4521; aapT) were cloned into the expression vector pRB474 (Brückner, 1992). The resulting constructs were introduced into biofilm-, icaADBC- and aap-negative S. epidermidis 1585. Whereas expression of the complete aap did not affect the biofilm phenotype of this strain, the expression of a truncated aap led to a biofilm-positive phenotype (Fig. 3C). This demonstrates that the repetitive domain B mediates intercellular adhesion and biofilm formation and gives additional evidence that domain A of the protein is not involved. Furthermore, this demonstrates that the biofilm-positive phenotype of S. epidermidis 5179-R1 is not due to unmasking of cell surface structures hidden by full-length Aap. AapT was additionally cloned into pAS1 under the control of a xylose-inducible promoter, resulting in pASaapT. After transformation with pASaapT the biofilm-negative S. carnosus TM300 became biofilm-positive when grown in TSB without glucose (Dobinsky et al., 2003) containing 0.4% (w/v) xylose. Obviously, domain B mediates intercellular adhesion not only in S. epidermidis but also in heterologous genetic backgrounds.

Aap mediates biofilm formation after limited proteolysis

Addition of α2-macroglobulin to the growth medium specifically led to the loss of cell cluster formation (data not shown) and inhibited biofilm formation of S. epidermidis 5179-R1 in a dose-dependent manner (Fig. 4A). Additionally, the differences in the cell surface protein patterns of S. epidermidis 5179 and the corresponding revertant were also abolished after addition of α2-macroglobulin to the growth medium (Fig. 4B). Thus, Aap is subject to stepwise, limited, endogenous proteolysis, most probably through activity of a serine or metalloprotease, as the cysteine protease inhibitor E64 had no effect on biofilm formation (Fig. 4A). Enhanced proteolysis in the revertant, carrying genes encoding the serine proteases GluSE (Moon et al., 2001; Dubin, 2002), ClpX and sspP1 (Teufel and Götz, 1993), encoding a metalloprotease, may result from differences in protease activity level compared with S. epidermidis 5179. However, no differential expression of the respective proteases was found by Northern analysis nor was a difference in protease activity detected by zymographic assays (data not shown), indicating that these underlying differences are very subtle. Nevertheless, they finally result in a quantitative shift between the different Aap species from the 220 kDa towards the 180 kDa Aap and 140 kDa Aap isoforms with different functional properties.

Figure 4.

A. Concentration-dependent influence of broad range protease inhibitor α2-macroglobulin (α2-M), and cysteine protease inhibitor E64 on S. epidermidis 5179-R1 biofilm formation as determined by the biofilm assay. Heat-inactivated α2-macroglobulin was used to control the specificity of biofilm inhibition.
B. Immunoblot analysis of cell surface proteins from S. epidermidis 5179 and 5179-R1 using an antiserum raised against affinity-purified Aap. Proteins were isolated from cells grown in TSB containing 0.25 U ml−1α2-macroglobulin, leading to the loss of the 140 kDa and the 180 kDa band in S. epidermidis 5179-R1 and the 180 kDa protein in S. epidermidis 5179. In contrast, the 220 kDa band was more pronounced in both strains.

Importantly, not only endogenous, but also exogenous, host-derived proteases can induce biofilm formation via proteolytic Aap processing. Trypsin, elastase or cathepsin G all induced biofilm formation in S. epidermidis 5179 in a dose-dependent manner: whereas in high and very low protease concentrations this strain remained biofilm-negative, in a certain concentration range biofilm formation can be detected (Fig. 5A). In parallel, the biofilm-positive phenotype of S. epidermidis 5179 was accompanied by the detection of a truncated 140 kDa Aap by immunoblotting using the specific anti-Aap antiserum (Fig. 5B). As in addition S. epidermidis 5179 biofilm formation under biofilm-inducing conditions could be specifically inhibited by the specific anti-Aap antiserum (Fig. 5C), this confirms that proteolytic processing of Aap is responsible for the expression of a biofilm-positive phenotype. Importantly, protease-mediated induction of biofilm formation through Aap processing appears to be a general mechanism, as it could also be observed in additional clinically significant, biofilm- and icaADBC-negative but aap-positive S. epidermidis strains isolated from prosthetic joint infections. In presence of trypsin in the growth medium these strains formed biofilms that could be inhibited by anti-Aap antiserum (Fig. 5D), indicating that in these strains Aap is also essentially involved in biofilm formation.

Figure 5.

A. Concentration-dependent induction of S. epidermidis 5179 biofilm formation by addition of trypsin, cathepsin G and elastase to the growth medium. The cut-off at A570 0.1, differentiating between a biofilm-positive and a biofilm-negative phenotype, is indicated by a line. The biofilm formed by S. epidermidis 5179 is weaker than that of the revertant (see Table 1), mirroring the necessity of temporally or spatial protease activity for full Aap activation.
B. Immunoblot analysis of cell surface proteins from S. epidermidis 5179 obtained from cells grown in TSB and TSB containing trypsin (2 µg ml−1). Using an antiserum raised against affinity-purified Aap, an additional 140 kDa (arrow) and a pronounced 180 kDa band (double arrowhead) is detected under biofilm-inducing growth conditions, while the 220 kDa band is lost (arrowhead).
C. Concentration-dependent inhibition of biofilm formation of S. epidermidis 5179 grown under biofilm-inducing conditions (TSB + 2 µg ml−1 trypsin) by an Aap-specific antiserum. Anti-Aap antiserum absorbed against S. epidermidis 5179, anti-Aap antiserum absorbed against 5179-R1 cells and serum from a rabbit before immunization served as controls. Relative inhibition was calculated by the formula (1 − A570 with antiserum/A570 without antiserum).
D. Biofilm formation of clinical significant S. epidermidis 43 under different growth conditions in the biofilm assay. Whereas the isolate is biofilm-negative when grown in TSB, it becomes biofilm-positive in the presence of trypsin (2 µg ml−1). As biofilm formation can be specifically inhibited by anti-Aap antiserum (diluted 1:400 in TSB), Aap apparently mediates biofilm formation in this strain. A normal rabbit serum from an animal before immunization was used as a negative control. Similar results were obtained for three independent, clinical significant S. epidermidis strains (S. epidermidis 45, 57 and 68).

In addition to the altered biofilm phenotype proteolytic induction of Aap-mediated cell aggregation may have important implications with regard to pathogen clearance by the immune system, as cluster-forming cells of S. epidermidis 5179-R1 were about 10-fold less susceptible to killing by whole blood than were cells of S. epidermidis 5179 (Table 2).

Figure 2.

Killing of S. epidermidis 5179 and 5179-R1 by whole blood.


Biofilm formation has been shown to be a major feature determining S. epidermidis virulence in device-related infections (Rupp et al., 1999a,b; 2001). Several epidemiological studies found an association between the isolation of a clinically relevant S. epidermidis and a biofilm-positive phenotype (Christensen et al., 1994) and detection of icaADBC (Ziebuhr et al., 1997; Frebourg et al., 2000; Galdbart et al., 2000), respectively. Therefore, icaADBC-negative strains are currently regarded as prototype avirulent strains (Zhang et al., 2003). However, our results demonstrate that certain S. epidermidis strains cause foreign body infections despite inactivation of icaADBC and a biofilm- and PIA-negative phenotype in vitro. We have identified for the first time a polysaccharide independent mechanism of S. epidermidis biofilm accumulation, mediated by Aap. Aap has been previously implicated in S. epidermidis biofilm formation as a putative cell wall receptor for PIA (Hussain et al., 1997; Mack, 1999). However, we showed that a proteolytically processed 140 kDa derivative of the wild-type protein alone mediates intercellular adhesion leading to biofilm accumulation also in a completely PIA-negative background, attributing a bimodal function in S. epidermidis biofilm accumulation to Aap. As biofilms have been demonstrated in vivo in other icaADBC- and PIA-negative S. epidermidis strains like KH11 (Peters et al., 1981; 1982; Rohde et al., 2001a), and independent clinical S. epidermidis isolates also formed Aap-dependent biofilms in the presence of exogenous proteases, homologous mechanisms appear to be widespread in clinical S. epidermidis isolates. Taken into account that many clinically significant strains carry both icaADBC and aap (Klug et al., 2003; Vandecasteele et al., 2003), PIA and Aap may act cooperatively in S. epidermidis biofilm formation. Importantly, two Aap homologues SasG (Roche et al., 2003a,b) and Pls (Savolainen et al., 2001; Huesca et al., 2002) have recently been described in S. aureus where PIA/PNAG-mediated biofilm formation has also been observed (Cramton et al., 1999; McKenney et al., 1999; Rohde et al., 2001b; Knobloch et al., 2002). SasG and Pls function as surface adhesins involved in adherence to desquamated nasal epithelial cells, cellular lipids and glycolipids respectively. In SasG, adhesive properties are located in the N-terminal domain A. The presence of a repetitive domain sharing significant homology with domain B of Aap suggests that SasG and Pls proteins may additionally also function as PIA-independent, alternative intercellular adhesins in this species.

Two S. epidermidis cell surface proteins, the Staphylococcal Surface Protein (SSP-1 and SSP-2) (Veenstra et al., 1996) and the autolysin AtlE (Heilmann et al., 1997; Rupp et al., 2001) have already been implicated in S. epidermidis biomaterial colonization, a third protein designated Bhp, a homologue to the biofilm-associated protein (Bap) of S. aureus, is only hypothetically involved (Cucarella et al., 2001). However, SSP and AtlE act during the phase of primary attachment to the surface to be colonized, whereas the clinical isolate S. epidermidis 5179 and the corresponding revertant 5179-R1 displayed nearly identical attachment characteristics to the native polymer surface studied, demonstrating that Aap is involved in the second, accumulative phase of biofilm formation. The aggregative properties of Aap are determined by domain B, as S. epidermidis 5179-R1 biofilm formation can be inhibited by purified domain B but not domain A and expression of the repeat region alone in an aap-negative genetic background  leads to a biofilm-positive phenotype. However, this domain becomes only active after cleavage of the native protein, for example, because of conformational changes. Cell aggregation may then result from AapT–AapT interactions. Such homophilic protein interactions leading to auto-aggregation have been described for protein H-mediated S. pyogenes cell aggregation (Frick et al., 2000), the surface-exposed haemagglutinin mediated cell aggregation in B. pertussis (Menozzi et al., 1994) and antigen 43, involved in E. coli biofilm formation (Danese et al., 2000). However, the involvement of other, also non-proteinaceous cell wall structures serving as AapT ligands in the aggregation process cannot be excluded.

Proteolysis plays a pivotal role in the pathogenicity of many microorganisms. Especially in S. aureus the differential expression of protease activity contributes to the adaptation of the pathogen to a changing environment by altering the adhesive properties to extracellular matrix proteins (McGavin et al., 1997; Karlsson et al., 2001; McAleese et al., 2001; Savolainen et al., 2001). In S. epidermidis, limited proteolysis may allow the bacterium to rapidly optimize intercellular adhesive cell surface properties via modulation of endogenous protease expression by yet unknown mechanisms which determine via activation or degradation of Aap the persistence in or release from the staphylococcal biofilm and subsequent spreading of the infection. The expression of staphylococcal proteases is under environmental control of quorum sensing circuits like agr and sar (Lindsay and Foster, 1999), potentially linking Aap processing with these global regulators of staphylococcal virulence. Importantly, in addition to this concept of biofilm induction through endogenous staphylococcal proteases, during infection induction of cell aggregation and biofilm formation after Aap processing may result from host-derived proteases. As these cell aggregates of S. epidermidis 5179-R1 are more difficult to phagocytose this strategy provides an additional, PIA-independent mechanism (Vuong et al., 2004) for S. epidermidis to evade effector mechanisms of the innate immune response during their ‘race for the surface’ (Gristina, 1987) for establishment of biomaterial-related infections without being yet organized in a biofilm architecture, which itself prevents bacteria from phagocytosis (Johnson et al., 1986).

Experimental procedures

Bacterial strains

Clinical S. epidermidis isolates were obtained from a patient of the University Hospital Hamburg-Eppendorf (S. epidermidis 5179 and S. epidermidis 6317). In addition, four S. epidermidis strains with proven clinical significance were isolated from patients with infected joint prostheses (S. epidermidis 43, 45, 57 and 68). Biofilm-negative S. epidermidis 1585 was isolated from a port-catheter infection. Using specific primers (see below) neither aap nor icaADBC were detected. All strains were identified as S. epidermidis by a negative clumping factor and biochemically using the ID32Staph system (bioMerieux, Marcy l’Etoile, France). S. epidermidis 1457, the biofilm-negative transposon mutant 1457-M10, and S. carnosus TM300 have been described elsewhere (Mack et al., 1992; 2001; Table 3)

Table 3.  Bacterial strains and plasmids used in this study.
S. epidermidis 1457Biofilm-, icaADBC-, and PIA-positive strain Mack et al. (2001)
S. epidermidis 5179Biofilm-, and PIA negative strain; icaA::IS257 Rohde et al. (2001a)
S. epidermidis 1585Isolated from port-catheter infection; biofilm-, aap-, icaADBC- and PIA negativeThis work
S. carnosus TM300Cloning host for protoplast transformation Mack et al. (2000)
S. aureus RN42208325–4r (restriction-deficient cloning host) Novick (1991)
E. coli TOP10Cloning host, genotype FmcrΑΔ (mrr-hsdRMS-mcrBC) Φ80lacZΔM15 ΔlacX74recA1 deoR araD139 Δ(ara-leu)7697 galU galK rpsL (StrR) endA1 nupGInvitrogen
E. coli BL21AIExpression host; genotype FompT hsdSB(rB-mB-) gal dcm araB::T7RNAP-tetAInvitrogen
pRB474Shuttle vector for cloning in E. coli and staphylococci; constitutive gene expression in staphylococci via vegII promoter Brückner (1992)
pRBaappRB474 containing aap (nt 1–4521)This work
pRBaapTpRB474 containing aap domain B (nt 1816–4521) fused with an export signal (nt 1–189)This work
pAS1Plasmid for gene expression under the control of the xylA promoter Tegmark et al. (2000)
pASaapTpAS1 containing aap domain B (nt 1816–4521) fused with an export signal (nt 1–189)This work
pCX15Plasmid for gene expression under the control of the xylA promoter Wieland et al. (1995)
pCXaappCX15 containing aap from S. epidermidis RP62AThis work
pBluescript II SK E. coli cloning vectorStratagene
pCR2.1Vector for cloning of PCR ampliconsInvitrogen
pENTR/D-TopoEntry vector for Gateway® technologyInvitrogen
pDEST17Expression vector; N-terminal His6-tag fusionInvitrogen
pDESTdomApDEST17 containing aap domain A (nt 190–1854)This work
pDESTdomBpDEST17 containing aap domain B (nt 1816–4521)This work

Adherence assay, in vitro biofilm formation, PIA detection

Primary attachment to native plastic surfaces was assessed by using a specific ELISA (Mack et al., 2001). Biofilm formation was tested in TSB (Trypticase soy broth, Becton Dickinson, Cockeysville, USA) using the semiquantitative microtitre plate test (biofilm assay) as described (Mack et al., 2001). Biofilm-positive S. epidermidis 1457 (Mack et al., 1992) and its corresponding, biofilm-negative icaADBC transposon mutant 1457-M10 (Mack et al., 1994) served as positive and negative controls respectively. α2-Macroglobulin (Roche, Mannheim, Germany), l-trans-epoxysuccinylleucylamido-(4-guanidino) butane (E64; Sigma, Munich, Germany), trypsin (Roche, Mannheim, Germany), elastase (Roche, Mannheim, Germany) or cathepsin G (Sigma, Munich, Germany) were added to the medium in concentrations as indicated in the results section. Biofilm stability against protease or sodium-meta-periodate treatment was tested as published elsewhere (Mack et al., 1992). Biofilm inhibition experiments using specific rabbit antisera and estimation of relative inhibition was performed as described (Hussain et al., 1997). Inhibition experiments using recombinant proteins were carried out by adding serial dilutions of the respective protein into the growth medium. Buffer (0.05 M NaPO4, 0.5 M NaCl, pH 7.4) containing no protein served as a negative control. To qualitatively and semiquantitatively detect PIA synthesis a coagglutination assay (Mack et al., 1992) using a PIA-specific antiserum was performed (Mack et al., 2001).

Enrichment of biofilm producing cells

A fresh colony of the respective strain was inoculated into 5 ml of TSB and incubated in cell culture bottles (Nunc, Roskilde, Denmark) under static conditions. After 22 h, the bottles were heavily shaken and the medium was changed. After establishment of a visible biofilm on the bottom of the bottle (usually after 3–5 days), the adherent cells were washed twice with 5 ml of phosphate-buffered saline. Then cells were scraped off from the surface and plated onto Columbia blood agar plates. Subsequently, single colonies were picked and tested for biofilm formation in the biofilm assay. Subcultivation of the isolate on Columbia blood agar plates revealed the stability of the biofilm-positive phenotype for at least 15 passages.

Preparation of cell wall proteins, SDS-PAGE, immunoblotting, N-terminal sequencing, mass spectrometry

For preparation of cell wall anchored proteins a protocol from McCrea et al. (McCrea et al., 2000) was followed except that cells from a static overnight culture were used. The total protein concentration was estimated using the Bradford-Assay as recommended by the manufacturer (Bio-Rad, Munich, Germany). SDS-PAGE, blotting of cell wall-associated proteins, immunoblot analysis, N-terminal sequencing and mass spectrometry have been described elsewhere (Stürenburg et al., 2002).

Anti-Aap antiserum

For immunization aap from strain RP62A (GenBank accession number AJ249487) was cloned into pCX15 under the control of a xylose-inducible promoter. The resulting construct pCXaap was introduced into S. carnosusTM300 via protoplast transformation (Mack et al., 2000). For Aap purification cells were grown in B2 medium (1% casein hydrolysate, 2.5% yeast extract, 0.1% K2HPO4, 0.5% glucose, 2.5% NaCl) to an OD578 of 0.5, induced by xylose (0.25% w/v) and the culture was grown for an additional 5 h. Cells were lysed with lysostaphin and the lysate was analysed by SDS-PAGE and Western immunoblots with anti-Aap antibodies. Subsequently Aap was purified using affinity chromatography. Anti-Aap antibodies (Hussain et al., 1997) were pre-absorbed against uninduced S. carnosus × pCXaap and coupled to cyanogen bromide-activated Sepharose 4-B (Amersham, Freiburg, Germany) according to instructions of the supplier. The coupled sepharose beads were shaken with lysostaphin lysate of xylose-induced S. carnosus × pCXaap for 30 min at room temperature. The mixture was poured into a column and washed with 50 mM phosphate buffer pH 7.5. The bound Aap was eluted with 100 mM glycine buffer pH 3.5 and immediately neutralized with 1 M Tris-HCl pH 8.0. The obtained fractions were analysed by SDS-PAGE and those containing the Aap were pooled, dialysed against 10 mM phosphate buffer pH 7.0 and freeze-dried. Eurogentec (Seraing, Belgium) performed the immunization of rabbits with recombinant, affinity purified Aap according to the standard immunization program of the company. The resulting antiserum was used in this study.

Detection and sequencing of icaADBC and aap

Detection of icaADBC was carried out as described (Mack et al., 2001; Rohde et al., 2001a). For sequence analysis full-length aap was amplified using primers aapfor1 and aaprev3 using the Expand Long Template PCR System (Roche, Mannheim, Germany) under conditions as recommended by the manufacturer. After cloning of the resulting amplicon into pCR2.1 sequencing and data analysis was performed as published previously (Knobloch et al., 2001; Rohde et al., 2001a). Screening for aap was carried out using primers aapfor1–3 and aaprev1 and 2 in different combinations (Table 4). Detection of icaADBC mRNA was performed as described (Dobinsky et al., 2003).

Table 4.  Primers used in this study.
Primer namePrimer sequencePrimer
position (nt)a
  • a

    . Primer positions are given according to the aap sequence of S. epidermidis 5179 (GenBank accession number AY359815). Positions indicated with a minus are located 5′, primers with positions higher than nt 4521 are located 3′ outside the coding sequence.

  • n.a., not applicable.

aap rev15′-AGTTGGCGGTATATCTATTGTA-3′1098n.a.
aap rev35′-AGTTTTTATATGAAATTATTTTTCATTACCT-3′4539Contains aap stop codon
aap rev domA 5′-TTATGCTTTAGGAGTGTATGTCAATG-3′1854Contains additional 5′ stop codon
aap for5′HindIII5′-AACGTTTTAAATACATGGGAGGTATAATATGGGCA-3′ −22 HindIII restriction site (bold)
aap TforstartR15′-CTGCAGATTTAGATGGTGCAACATTGACAT-3′1816 PstI restriction site (bold)
aap for5′BamHI5′-GGATCCTTAAATACATGGGAGGTATAATATGGGCA-3′ −22 BamHI restriction site (bold)
aap revAXA3′5′-CTGCAGCTTTCGCTTCATGGCTACTACTT-3′ 190 PstI restriction site (bold)
aap Trev3′KpnI5′-GGTACCAGTTTTTATATGAAATTATTTTTCATTACCT-3′4539 KpnI (bold) restriction site
aap Trev3′HindIII5′-AACGTTAGTTTTTATATGAAATTATTTTTCATTACCT-3′4539 HindIII (bold) restriction site
aap Trev3′SacI5′-GAGCTCAGTTTTTATATGAAATTATTTTTCATTACCT-3′4539 SacI (bold) restriction site
AXAmotivfor5′-CACCGAAGAAAAACAAGTTGATC-3′ 190Entry site for directional cloning in pENTR/D-Topo (bold)
StartaapTfor5′-CACCGATTTAGATGGTGCAACATTGACAT-3′1816Entry site for directional cloning in pENTR/D-Topo (bold)

Expression of aap and aapT in S. epidermidis and S. carnosus TM300

Full-length aap5179 (nt 1–4521) was amplified using primer aapfor5′HindIII and aapT rev3′KpnI, introducing a HindIII restriction site at the 5′ end and a KpnI restriction site at the 3′ end. These were used to clone the resulting amplicon into pRB474 (Brückner, 1992; Table 3), resulting in pRBaap. For fusion of the truncated aap to the export signal, nucleotides 1–190 were amplified with primers aapfor5′BamHI and aaprevAXA3′. The BamHI and PstI restriction sites were then used to clone the amplicon into pBluescript II SK. aap5179 nucleotides 1816–4521, containing the repetitive domain B and the C-terminal end of Aap, were amplified with primers aapTforstartR1 and aapTrev3′HindIII and cloned behind the export signal. Finally, the fused sequences were amplified with aapfor5′HindIII and aapTrev3′KpnI. This amplicon was directionally cloned into pRB474, giving pRBaapT. All amplifications were done using the Expand Long Template PCR System (Roche, Mannheim, Germany) under conditions as recommended by the manufacturer. All cloning procedures were performed using E. coli TOP10 as a host. pRBaap and pRBaapT were then transformed into the restriction-deficient S. aureus RN4220, introduced into S. epidermidis 1457-M12 via protoplast transformation and transduced into the icaADBC- and aap-negative S. epidermidis 1585 using phage A6C (Mack et al., 2001). In order to clone aapT into the expression vector pAS1 under the control of a xylose-inducible promoter (Tegmark et al., 2000) the truncated aap was amplified from pRBaapT using primers aapfor5′BamHI and aapTrev3′SacI. The amplicon was cloned via the BamHI and the KpnI restriction sites into pAS1, resulting in pASaapT. After introduction into S. aureus RN4220 by electroporation S. carnosus TM300 was transformed via protoplast transformation (Mack et al., 2000). Expression of the respective Aap species in the different hosts was confirmed by immunoblot analysis of cell wall-associated proteins using an Aap-specific antiserum.

Expression and purification of recombinant Aap domains A and B in E. coli

For recombinant expression of domain A and B the Gateway technology (Invitrogen, Karlsruhe, Germany) was used following essentially the manufacturer's recommendations. After amplification of domain A and B (Fig. 2C) using the Platinum Pfx DNA polymerase (Invitrogen, Karlsruhe, Germany) (domain A: forward primer AXAmotivfor, reverse primer: aaprevdomA; domain B: forward primer StartaapTfor, reverse primer aaprev3) the resulting amplicons were cloned into pENTR/D-TOPO and subcloned into pDEST17 (Invitrogen, Karlsruhe, Germany), fusing domain A and B with an N-terminal His6 tag. Expression was performed in E. coli BL21AI. The recombinant protein was affinity-purified using a HiTrap chelating HP column (Amersham, Freiburg, Germany) and finally eluted with sodium-phosphate buffer (0.05 M NaPO4, 0.5 M NaCl, pH 7.4) via a HiTrap desalting column (Amersham, Freiburg, Germany) to remove imidazole. SDS-PAGE and subsequent Coomassie staining tested integrity of the respective protein preparations.

Bactericidal test

The ability of S. epidermidis 5179 and 5179-R1 to survive in human blood was tested essentially as described previously (Frick et al., 2000) with the exception that cells from stationary phase cultures were used. Briefly, 100 µl of the bacterial suspension containing approximately 5 × 105 (range 1.8–7.3 × 105) cells were mixed with 1 ml of heparin-treated blood and gently agitated at 37°C. After 3 and 5 h appropriate dilutions were plated onto TSB-agar plates.


We thank Rainer Laufs for his continuous support. The constructive discussion with G. Peters is gratefully acknowledged. The contribution of N. Siemssen and L. Frommelt, ENDO-Klinik, Hamburg, Germany, by providing clinical S. epidermidis 43, 45, 57 and 68 from prosthetic joint infections is gratefully acknowledged. Phage A6C was kindly provided by V.T. Rosdahl, Statens Serum Institute, Copenhagen, Denmark. pRB474 was a gift of R. Brückner, Abteilung Mikrobiologie, Universität Kaiserslautern, Germany. pAS1 was a kind gift of S. Arvidson, Microbiology and Tumorbiology Center, Karolinska Institute, Stockholm, Sweden. We thank E. Bock for performance of electron microscopy. This work is supported by grants of the Deutsche Forschungsgemeinschaft (SPP 1047 Ma 1522/4-3 given to H. R., J. K.-M. K. and D. M.; SFB 470, Teilprojekt C10 given to D.M.; SPP 1130 He 1850/6-1, SFB 492, Teilprojekt B9, given to M.H., and SPP1047 He 1850/2-3 given to M.H.). C.B. is recipient of a fellowship of the Werner Otto-Stiftung Hamburg.