Characteristics of the biofilm matrix and its role as a possible target for the detection and eradication of Staphylococcus epidermidis associated with medical implant infections


  • Said Jabbouri,

    1. Laboratoire de Recherches sur les Biomatériaux et Biotechnologies, Université du Littoral-Côte d'Opale, Boulogne-sur-mer, France
    2. Institut de Recherche pour le Developpement, UMR 180, Laboratoire de Microbiologie et Biotechnologie des Environnements Chauds, Marseille, France
    Search for more papers by this author
  • Irina Sadovskaya

    1. Laboratoire de Recherches sur les Biomatériaux et Biotechnologies, Université du Littoral-Côte d'Opale, Boulogne-sur-mer, France
    Search for more papers by this author

  • Editor: Roger Bayston

Correspondence: Present address: Said Jabbouri, Water Research and Pollution Control Department, National Research Center, Environmental Research Division, Street El-Bohous, Dokki, PO Box 12622, Cairo, Egypt. Tel.: + 201 866 45041; fax: + 202 237 83308; e-mail:


The virulence of Staphylococcus epidermidis is related to its capacity to form biofilms. Such biofilm-related infections are extremely difficult to treat and to detect in early stages by the traditional microbiological analyses. The determination of the chemical composition of the extracellular polymeric substances (EPS) of the biofilm matrix, as well as the elucidation of the sensitivity of biofilms to enzymatic degradation should facilitate the development of new therapies against biofilm-related infections. The chemical analyses of EPS had shown qualitative and quantitative variations of their nature, depending on the strains and culture conditions. The poly-N-acetylglucosamine (PNAG) is considered the main component of staphylococcal biofilms. However, certain strains form biofilms without PNAG. In addition to PNAG and proteins, extracellular teichoic acid was identified as a new component of the staphylococcal biofilms. The sensitivity of staphylococcal biofilms to enzymatic treatments depended on their relative chemical composition, and a PNAG-degrading enzyme, in conjunction with proteases, could be an efficient solution to eliminate the staphylococcal biofilms. A detection of specific ‘antibiofilm’ antibodies in the blood serum of patients could serve as a convenient noninvasive and inexpensive diagnostic tool for the detection of foreign body-associated staphylococcal infections. Used as a coating antigen in the enzyme-linked immunosorbent assay test, PNAG did not sufficiently discriminate healthy individuals from the infected patients.


While Staphylococcus aureus is known as a pathogen with a number of virulence factors (e.g. exotoxins and enzymes), Staphylococcus epidermidis is mainly a normal inhabitant of the healthy human skin and mucosal microbial communities. As a commensal bacterium, it has a low pathogenic potential. In recent decades, however, S. epidermidis and other coagulase-negative staphylococci (CoNS) have emerged as a common cause of numerous nosocomial infections, mostly occurring in immunocompromised hosts or patients with implanted medical devices, such as intravascular and peritoneal dialysis catheters, prosthetic heart valves, or orthopaedic implants (Ziebuhr et al., 2006). These infections can be described as ‘chronic polymer-associated infections’ (Götz, 2002). A characteristic feature of this kind of infection is the ability of the causative microorganisms to colonize surfaces of biomaterials in multilayered biofilm-structured communities of cells enclosed in a self-produced polymeric matrix, an amorphous slimy material, which is loosely bound to staphylococcal cells. This ability to form biofilms is believed to make the microorganisms more resistant to administered antibiotics and to the defence mechanisms of host immunity (von Eiff et al., 1999). Evidence suggests that biofilm formation also plays a role in S. aureus wound infections (Akiyama et al., 1996) and osteomyeltis (Buxton et al., 1987).

To date, no efficient treatment or early diagnostics of implant-associated infections has been proposed. Direct microbiological analysis is frequently negative, and does not rule out infections. Often, when cultures taken at the infected site become positive, the infection is already at an advanced stage and removal of the prosthesis in order to increase the efficiency of the antibiotic therapy becomes unavoidable.

To develop efficient tools that would improve the medical decision making and help to combat the infections related to medical implants, two strategies can be proposed: the first is preventive and the second is curative. The preventive strategy consists of inhibiting the bacterial adhesion on implant surfaces, and in detecting bacteria in blood circulation in early stages of infection, in order to eliminate them using the conventional antibiotics. The curative method also consists of enhancing the action of antibiotics by dissolution of the biofilm and dispersal of sessile bacteria into their sensitive planktonic state. These two strategies could be accomplished using tools of molecular genetics and/or biochemistry. The genetic approach, at the preventive level, may enable the control the expression of genes involved in the early stages of adhesion and biofilm formation. The curative aspect should be able to control the expression of genes involved in bacterial detachment and dispersal. The genetic aspect will not be discussed in this Minireview.

The biochemical approaches of both strategies (preventive and curative) may consist of acting on the extracellular polymeric substances (EPS) of the biofilm matrix, by blocking their biosynthesis or by enzymatically degrading them. EPS antigenic properties may be explored for the early detection of antibodies directed against the biofilm EPS in the early stages of the biofilm formation. In the present Minireview, we discuss some aspects of the biochemical approach to the eradication and detection of staphylococcal biofilm-associated infection, developed by our research group. We mainly focused on the chemical characterization of biofilm EPS of S. epidermidis and other CoNS. We also studied the sensitivity of the biofilm to different degrading enzymes, taking into account their composition and attempting to specifically target the biofilm constituents. Poly-β(1,6)-N-acetylglucosamine (PNAG), a characteristic component of staphylococcal biofilms with a well-established chemical structure, was tested as a coating agent in enzyme-linked immunosorbent assay (ELISA) tests for potential serodiagnostics.

Chemical composition of the EPS matrix of staphylococcal biofilms

PNAG: an overview

Staphylococcus epidermidis RP62A (ATCC 35984) has been used as a preferential model biofilm-forming strain by a number of authors. Its extracellular polysaccharide antigens were isolated and studied independently by several different research groups (for a recent review, see Otto, 2009). An extracellular capsular polysaccharide adhesin (PS/A) was first isolated by the group of G. Pier (Boston, MA) (Tojo et al., 1988) from the culture supernatant of S. epidermidis strain RP62A. PS/A has been reported to be the component of the bacterial cell surface and biofilm layer, which mediates cell adherence to biomaterials and protects the bacterial cells from host defences. Over the years, this polysaccharide has been referred to as a PS/A by this research group. Later, Christensen et al. (1990) described a slime-associated antigen (SAA) isolated from the same strain and having a similar function. SAA was claimed to be different from PS/A. However, Baldassari et al. (1996) suggested that SAA and a hexosamine-containing polysaccharide intercellular adhesin (PIA) of S. epidermidis strains RP62A and 1457, described at that time by Mack et al. (1994), could be the same antigenic molecule.

Mack et al. (1996) were the first to elucidate the chemical structure of PNAG (called, according to its biological properties, PIA). Using a combination of analytical methods and nuclear magnetic resonance (NMR), PIA was identified as a linear β(1,6)-linked N-acetylglucosaminoglycan containing c. 130 N-acetylglucosamine (GlcNAc) residues, partially substituted with O-succinyl groups, partially de-N-acetylated and apparently phosphorylated (Mack et al., 1996). The genes encoding PNAG biosynthesis are organized in the icaADBC (intercellular adhesion) operon, which consists of four ORFs (Heilmann et al., 1996; Gerke et al., 1998) with a transcriptional repressor gene, icaR, located upstream and transcribed in the opposite orientation (Conlon et al., 2002). The ica locus was later found in a number of S. aureus strains, and its presence was related to the ability to form a biofilm in vitro (Cramton et al., 1999). A recombinant strain of Staphylococcus carnosus (pCN27), containing icaABC of S. epidermidis RP62A, unlike the parent S. carnosus strain, which is biofilm-negative, was adherent to glass and revealed the ability to form intercellular aggregates as well as to produce PNAG (Heilmann et al., 1996). McKenney et al. (1998) subsequently demonstrated that the recombinant S. carnosus (pCN27) antigen was identical to PS/A and ‘chemically related’, but distinct from PIA in molecular size, solubility, and substitution of the majority of the amino groups of the glucosamine residues with succinate. This polymer, named poly-N-succinyl-β-(1,6)-glucosamine (PNSG), was suggested as a potential vaccine candidate against staphylococcal infections (McKenney et al., 1999, 2000). However, subsequent studies carried out by the same group showed that the presence of the N-succinyl substitution was an analytical artefact (Joyce et al., 2003).

These authors used a ‘PS/A overproducing strain’S. aureus MN8m, and the corresponding polysaccharide was named SAE (S. aureus exopolysaccharide). Detailed NMR studies, in combination with the chemical modifications, allowed a complete assignment of NMR spectra of SAE. According to Joyce et al. (2003), the main differences between SAE and PIA were phosphorylation (absence of a phosphate substitution in SAE) and molecular mass [>300 kDa for SAE and ∼30 kDa for PIA (Mack et al., 1996)].

It seemed that the confusion could arise from the variety of growth conditions and purification methods used by different research groups working mainly with two model strains: S. epidermidis RP62A and S. aureus MN8m. In order to clarify this ambiguity, a direct comparative study of ‘PS/A’ and PIA has been carried out in our group. As a first step, we established a simple protocol for a large-scale biofilm culture and a mild method of extraction and separation of components of the biofilm matrix for a model biofilm-forming strain S. epidermidis RP62A (Sadovskaya et al., 2005).

We then compared the chromatographic elution profiles and the chemical structure of PNAG, prepared from two model strains, S. epidermidis RP62A and S. aureus MN8m, grown under identical conditions and using the same method of extraction and purification as the GlcNAc-containing polysaccharides. In agreement with the literature data (Mack et al., 1996; Joyce et al., 2003), the PNAG obtained of both strains represented a β(1,6)-linked N-acetylglucosaminoglycan, with a part of the GlcNAc residues deacetylated and partially O-succinylated. The molecular weights (MWs) of the two polymers were close, and their chemical structure was identical, except for the degree of partial N-deacetylation and O-succinylation (Sadovskaya et al., 2005). The PNAG from S. epidermidis RP62A did not contain any phosphate substitution; the presence of phosphate demonstrated by Mack et al. (1996) was probably due to the contamination by the phosphate buffer used during purification. Therefore, our data confirmed that, as stated in Maira-Litran et al. (2004), ‘PIA and PS/A are the same chemical entity – PNAG’.

The chemical structure of PNAG from a number of strains of CoNS from our collection was also investigated. We have shown that the PNAG of all strains studied had the same structural features as the one from model staphylococcal strains, with the difference in the quantities produced and the degree in substitution with charged groups (Sadovskaya et al., 2006).

A genetic locus pgaABCD, promoting surface binding, intercellular adhesion, and biofilm formation, has been identified recently in a number of Gram-negative bacteria. Genetic and biochemical studies demonstrated that, despite a very limited homology of pga and ica at the nucleotide or the amino acid level, a pga-dependent polysaccharide in Escherichia coli was a poly-β-(1,6)-GlcNAc (PGA), a polymer with a structure close to staphylococcal PNAG (Wang et al., 2004). Later, we have isolated a pga-dependent polysaccharide from the biofilms of a swine pathogen Actinobacillus pleuropneumoniae (Izano et al., 2007) and a human periodontal pathogen Aggregatibacter actinomycetemcomitans (Izano et al., 2008). We have shown that polysaccharides of the two strains were β(1,6)-linked poly-GlcNAc. Depending on the strain and the preparation, some of the GlcNAc residues (1–15%) were N-deacetylated. Thus, for the first time, the PGA was isolated directly from a biofilm of clinical Gram-negative strains, and the first chemical and NMR evidence for the de-N-acetylation of PGA from a Gram-negative bacterium was presented (Izano et al., 2007, 2008). Homologues of the pgaABCD locus were found in the genomes of several pathogenic Gram-negative bacteria, such as Actinobacillus actinomycetemcomitans, Actinobacillus pleuropneumoniae, Bordetella pertussis, Burkholderia cepacia, Pseudomonas fluorescens, Yersinia pestis, etc. These pathogens could synthesize hexosamine-containing exopolysaccharides that stabilize biofilms of these species (Kaplan et al., 2004; Wang et al., 2004).

Thus, PNAG appears to be an antigen that may play an important role in biofilm formation in a number of bacterial species, both Gram-positive and Gram-negative.

Extracellular teichoic acid (EC-TA) as the component of the EPS of a biofilm matrix

Teichoic acid (TA) is another extracellular carbohydrate-containing polymer known to be produced by S. epidermidis RP62A (Tojo et al., 1988; Hussain et al., 1991, 1992). While cell-wall TA (CW-TA) is a common component of all Gram-positive bacteria, EC-TA has been discovered only in a limited number of species (Jacques et al., 1979; de Boer et al., 1981). Studying the ‘slime’ produced by S. epidermidis in a chemically defined medium, Hussain et al. (1992) characterized an extracellular high MW carbohydrate polymer with a composition similar to the S. epidermidis CW-TA. Both polymers contained glycerol, phosphate, glucose, glucosamine, and d-alanine (d-Ala).

The importance of CW-TA and particularly the presence of d-Ala substitution in the CW-TA, in the biofilm formation of S. aureus, was demonstrated (Gross et al., 2001). In S. epidermidis, the CW-TA significantly enhances adhesion of the bacterial cells to fibronectin-coated surfaces, which suggested its possible role as a bridging molecule between microorganisms and immobilized fibronectin in the early stages of S. epidermidis pathogenesis (Hussain et al., 2001).

However, a certain controversy existed regarding the composition of biofilm, or ‘slime’, of S. epidermidis, and the role that EC-TA may play as its constituent. Until recently, the staphylococcal ‘slime’ has been mainly associated with PIA (Götz, 2002). On the other hand, earlier literature data indicated that S. epidermidis‘slime’ consisted mostly of TA and protein (Hussain et al., 1993).

The chemical composition of the extracellular biofilm matrix of S. epidermidis RP62A, grown under previously established conditions favourable for the formation of biofilm, was studied in our group.

A simple extraction and purification procedure allowed us to obtain the total extract of extracellular biofilm polymers, minimizing the contaminations with macromolecules from culture media and cellular polymers. After the fractionation of the crude biofilm extract we isolated, along with PNAG and protein components, another carbohydrate-containing polymer with a lower MW. This polymer contained glycerol, phosphate, Glc, and GlcNAc. After further purification, we identified it as an EC-TA (Sadovskaya et al., 2005).

Similar analysis has been performed on another model biofilm-producing strain, S. aureus MN8m (Vinogradov et al., 2006). There, the EC-TA was found to be composed of phosphate, ribitol, glycerol, GlcNAc, and Ala. Likewise, for several clinical strains studied, the TA was always present in their extracellular biofilm matrix (Kogan et al., 2006; Sadovskaya et al., 2006). All these data confirmed that the EC-TA was an important and permanent component of the staphylococcal biofilm matrix. It could be suggested that, because, in a number of Gram-positive strains, part of the CW-TA is located in a ‘fluffy’-layer region beyond the cell wall (Neuhaus & Baddiley, 2003), some of the TA would be released from the cell surface into the extracellular space and thus becomes a part of the extracellular matrix (Kogan et al., 2006).

Surprisingly, the chemical structures of the staphylococcal TAs, especially the pattern of d-Ala substitution – an important element for the pathogenecity of this microorganism – have not been elucidated in detail. As a subsequent step of the investigation, we elucidated the chemical structures of the TAs of two model biofilm-producing strains –S. epidermidis RP62A and S. aureus MN8m.

The chemical structure of CW-TAs of S. aureus and S. epidermidis is known thanks to the pioneer studies of Baddiley and colleagues in the 1960s and 1970s. These studies have shown that the TA of S. aureus was a (1,5)-linked poly(ribitol phosphate), substituted in position 4 of the ribitol residue with a β-GlcNAc (Fig. 1a; Baddiley et al., 1961, 1962a, b). The lipoteichoic acid of S. aureus was a (1,3)-linked poly(glycerol phosphate), attached to the diacylglycerol lipid anchor via a diglucosyl (gentobiosyl) unit (Fig. 1b; Duckworth et al., 1975). The CW-TA of S. epidermidis I2 was also a (1,3)-linked poly(glycerol phosphate), containing β-Glc and d-Ala residues (Fig. 1c; Archibald et al., 1968). Later, Endl et al. (1983) analysed the composition of the CW-TAs of several strains of S. aureus and CoNS; however, the detailed structures or the pattern of d-Ala substitution have not been studied.

Figure 1.

 Structural representation of the TA and lipoteichoic acid of Staphylococcus aureus and the TA of Staphylococcus epidermidis (a–c; literature data) and the schematic structure of the CW and EC-TA of model strains S. epidermidis RP62A (d; Sadovskaya et al., 2004), and S. aureus MN8m (e; Vinogradov et al., 2006). GlcNAc, N-acetylglucosamine; Glc, glucose. In (d): D-Ala, substituent may be present at C2 of glycerol and C6 of Glc, but not on C6 of GlcNAc.

The structures of TAs of both model strains were elucidated using chemical methods, MS, and NMR spectroscopy. It was found that EC and CW TAs of S. epidermidis RP62A had the same structure of (1,3)-linked poly(glycerol phosphate), substituted at the 2-position of glycerol residues with α-Glc, α-GlcNAc, d-Ala, and most interestingly, α-Glc6Ala (Fig. 1d; Sadovskaya et al., 2004).

Both EC and CW TAs from S. areus MN8M were composed of two different polymeric chains: a poly(ribitol phosphate) and poly(glycerol phosphate). In the poly(ribitol phosphate) chain, nearly 100% of ribitol was substituted with β-GlcNAc at position 4, and the structure corresponded to the one described in the literature for S. aureus H (Baddiley et al., 1961). Glycerol residues were (1,3)-linked. Most of these residues were unsubstituted, with ∼20% of the residues acylated with d-Ala at position 2 (Fig. 1e; Vinogradov et al., 2006).

Traditionally, the TA of S. aureus is considered as a sole poly(ribitol phosphate); mixtures of both poly(ribtol phosphate) and poly(glycerol phosphate) were reported previously only for Staphylococcus xylosus and Staphylococcus saprophyticus (Endl et al., 1983). However, our results on the analysis of TAs of several clinical strains of CoNS (Kogan et al., 2006) and S. aureus SA113 (unpublished data) indicate that the presence of two poly(polyol phosphates) TAs in S. aureus MN8m is not an exception.

To summarize, it was shown that along with proteins, the biofilm formed by two model biofilm-forming staphylococcal strains contained two carbohydrate-containing polymers: a homo-polysaccharide PNAG and poly(polyol phosphate) EC-TA. EC-TA is a highly polar and hydrophilic molecule, while PNAG is rich in relatively hydrophobic NAc groups. Both macromolecules possess positive and negative charges due to substitution with charged groups (free amino-groups and O-succinyl substituent in PNAG, d-alanyl esterification and phosphate in EC-TA), the amount of which may vary and may also be influenced by the conditions of growth (Sadovskaya et al., 2005). It can be suggested that the capacity to regulate positive and negative charges, as well as the hydrophilic properties of its biofilm constituents, should increase the ability of staphylococci to form biofilm on surfaces with different physico-chemical properties and to survive and proliferate under varying environmental conditions.

The presence of the d-Ala on C6 of glucose or the C2 of glycerol must be under the control of two distinct d-alanyl-transferases, probably encoded by two different genes. Their respective mutations should inform about the role played by the alanine group at each position, in biofilm formation and S. epidermidis virulence, and their likely role in staphylococcal defensive mechanisms such as resistance to antimicrobial peptides (Peschel et al., 1999; Weidenmaier & Peschel, 2008).

Diversity of clinical Staphylococcus strains

Because the ability to form a biofilm is traditionally considered as the main virulence factor of CoNS, and PNAG was regarded as the most characteristic biofilm component, the staphylococcal strains isolated from infected sites and particularly from prosthetic devices should be able (1) to form a biofilm (B+), (2) to possess the icaADBC operon (I+), and (3) to produce PNAG (P+). To verify the validity of this concept, Chokr et al. (2006) analysed the B+/−, P+/−, and I+/− criteria in 66 potentially virulent CoNS strains, collected from patients with infected implanted devices, undergoing treatment at the Mignot Hospital of Versailles, France. The ability to produce PNAG was tested by an immuno dot-blot using an anti-PNAG rabbit antiserum.

The results are summarized in Table 1. They indicated a significant implication of CoNS other than S. epidermidis, to which not much attention has been paid as yet, in the infections of medical implants.

Table 1.   CoNS species used in our studies and their in vitro characteristics
Staphylococcus speciesB−,P−,I+B−,P−,I−B+,P+,I+B+,P−,I+B+,P−,I−B−,P+,I+B−,P+,I−
  1. B+/−, ability to produce biofilm; P+/−, ability to produce PNAG; I+/−, presence of the icaADBC locus.

pasteuri0200 00

According to their characteristics based in the three criteria B+/−, P+/−, and I+/−, the strains were divided into seven groups (Table 1). Only 12 strains of 66 corresponded to the ‘classical’ B+P+I+ type. The prevalent type was B−, P−, I+, and it included 24 CoNS of the 66 studied strains. Despite the presence of ica genes in several species, no PNAG was detected in vitro. The inactivation of the ica operon could be attributed to several factors such as the insertion of the IS256 element (Ziebuhr et al., 1999), the action of the IcaR repressor (Conlon et al., 2002), and post-transcriptional regulation (Knobloch et al., 2002). Factually, the maximum transcription of icaADBC can be obtained with a persistence of PNAG and a biofilm-negative phenotype (Dobinsky et al., 2003). The reason for the absence of biofilm production despite the presence on the entire ica operon remains unclear. Similar results were obtained in the ica operon expression studies on 10 strains of S. epidermidis (seven biofilm-positive and three biofilm-negative strains) (Cafiso et al., 2004). Because the strains were isolated from patients with infected implanted devices, PNAG and biofilm may be formed in vivo, but not in vitro.

The two types of strains B+, P−, I+ (eight of 66 CoNS strains) and B+, P−, I− (two Staphylococcus lugdunensis of 66 strains) are very interesting, because they imply a possibility that different CoNS species could form a biofilm in vitro not containing PNAG. Selected biofilm-positive strains of this collection were then used for a detailed chemical analysis of their EPS.

Biofilms that do not contain PNAG

Having established the reliable method of analysis of the extracellular matrix of a staphylococcal biofilm (Sadovskaya et al., 2005), our group investigated the chemical composition of carbohydrate-containing polymers of a number of biofilm-positive staphylococcal strains associated with the infections of orthopaedic implants (Kogan et al., 2006; Sadovskaya et al., 2006). Of the 15 biofilm-producing clinical staphylococcal strains studied, three produced high amounts of PNAG in vitro. The production of PNAG by one of them, S. epidermidis 5 (CIP 109562), was higher than that of the model strain S. epidermidis RP62A, and therefore, this strain may be considered as a PNAG overproducer (Fig. 2a and b). Three strains (two S. epidermidis and one S. lugdunensis) were found to produce a small, but detectable amount of PNAG (Fig. 2c). Nine other strains (six S. epidermidis and one of each S. aureus, Staphylococcus warneri, and S. lugdunensis) did not produce in vitro PNAG in an amount that could be detected using direct chemical methods (Fig. 2d). While the presence of trace amounts of PNAG cannot be excluded, we suggested that biofilms of these strains contain mainly TA and protein components, which could be easily isolated from their extracellular extracts.

Figure 2.

 Typical chromatographic elution profiles of the EPS of different strains of Staphylococcus epidermidis on a Sephacryl S-300 column. Aliquots of each fraction were assayed for neutral sugars (circles) and aminosugars (squares). Staphylococcus epidermidis 5 (CIP 109562, b) produces PNAG in higher amounts then the model strain S. epidermidis RP62A (a). Staphylococcus epidermidis 341 (c) produces a small, but detectable amount of PNAG. EPS of S. epidermidis 455a (d) do not contain PNAG. EC-TC acid is always present among the EPS of the biofilm matrix (a–d).

Such a high proportion of strains that do not contain PNAG in their extracellular matrix is in good agreement with the number of recent reports pointing out the importance of an alternative, icaADBC and PNAG-independent mechanism of biofilm formation by staphylococci (Fitzpatrick et al., 2005; Rohde et al., 2005; Toledo-Arana et al., 2005).

Enzymatic treatment of biofilm matrix

In orthopaedic surgery, bacterial biofilm-related infections represent one of the most serious complications and have a huge impact in terms of morbidity, mortality, and medical costs (Campoccia et al., 2006). The treatment of these infections usually requires an appropriate surgical intervention, combined with a prolonged course of antimicrobial therapy (Trampuz & Zimmerli, 2005). In certain cases of infection, washing–draining procedures of the infected device with solutions containing antibiotics are used, in order to maintain the implant if possible. The use of an agent that would disintegrate the bacterial biofilm, release the planktonic cells into the environment, and therefore allow the appropriate antibiotic to eliminate infection would considerably improve the efficiency of this medical procedure. Complete elimination of the biofilm could thus help to avoid the removal of the orthopaedic implant. The enzymes capable of specifically degrading the constituents of the extracellular staphylococcal matrix could be further used in clinical procedures for the treatment of orthopaedic implant-associated infections.

We tested different enzymes and enzyme preparations for their capacity to disintegrate biofilms formed by staphylococcal strains related to orthopaedic prosthesis infections. The chemical composition of the biofilm of these strains from our collection was studied earlier. Unlike most of the previous studies, we attempted to specifically target the biofilm constituents. For this purpose, we have tested the activities of dispersin B (enzyme specifically degrading PNAG, Kaplan et al., 2003, 2004), proteases (proteinase K, trypsin), pancreatin, and Pectinex Ultra SP preparation (PUS, Novozyme) on the biofilms formed by different staphylococcal strains of our collection (Chokr et al., 2006; Chaignon et al. 2007). We compared the efficiency of different biofilm-degrading agents with the chemical composition of the biofilms. We have also examined the effect of some of these agents on the purified carbohydrate components of staphylococcal biofilms, PNAG and TA, and tested the proteolytic activities on crude biofilm extracts (Chaignon et al., 2007).

According to the chemical compositions of their in vitro grown biofilms, 15 clinical isolates were separated into two major groups: strains producing biofilms with a significant amount of PNAG and a larger group of strains producing biofilms containing a small amount or not containing PNAG. Biofilms of all the strains studied contained proteins and TAs (Kogan et al., 2006; Sadovskaya et al., 2006).

Kaplan et al. (2004) showed the ability of dispersin B to detach a preformed biofilm of four S. epidermidis strains isolated from the surfaces of infected intravenous catheters. In agreement with this finding, dispersin B could also efficiently remove biofilms of PNAG-producing strains from our collection (Sadovskaya et al., 2006).

Despite the fact that all biofilms contain proteins, the three proteases tested efficiently degraded only biofilms of strains that do not produce PNAG, demonstrating that, in this case, protein components of the biofilm played an important role in stabilizing its intercellular structure. The hydrolytic activity of the dispersin B and proteinase K on biofilm components was confirmed by their direct action on PNAG and the protein fraction of biofilms, respectively (Chaignon et al., 2007).

The heterogeneity of the biofilm matrix limits the potential of the monocompound enzyme, and the use of two or several successive treatments may be necessary for sufficient degradation of biofilms produced by clinical staphylococcal strains. Thus, a treatment with dispersin B, followed by a protease (proteinase K or trypsin), may facilitate eradication of biofilms of a variety of staphylococcal strains on inert surfaces. Unfortunately, none of the enzymes tested in this study was able to depolymerize the EC-TA, an important and recurrent component of staphylococcal biofilms. Finding an enzyme capable of specifically degrading this phosphor-diester polymer could favourably complement the action of the dispersin B and a protease.

Is the ability to form a biofilm necessary to maintain chronic infections in vivo?

We attempted to better understand whether the ability to form a biofilm in vitro was a sufficient and important virulence factor in the development of S. epidermidis infections in vivo. Earlier results of in vivo studies using a tissue cage guinea-pig (TC-GP) animal model concluded that inactivation of the ica locus by mutation did not affect the ability of the mutant to cause a persistent in vivo infection (Fluckiger et al., 2005).

Additionally, a number of studies have demonstrated that S. epidermidis and S. aureus ica mutants were still capable of colonizing in a tissue cage animal model of infection (Francois et al., 2003; Kristian et al., 2004; Fluckiger et al., 2005), suggesting that biofilm is not an important virulence factor in this model. To further address this question, we chose a selection of previously characterized clinical isolates of S. epidermidis (Table 1) in a TC-GP animal model (Chokr et al., 2007). Our study showed that the (B+, I+, P+) model strain S. epidermidis RP62A develops and maintains an infection in vivo, while the negative (B−, I−, P−) strain S. carnosus TM300 does not. Then, these results were checked with clinical isolates of S. epidermidis, possessing, respectively, both types: (B+, I+, P+) and (B−, I−, P−). Those with the positive type (B+, I+, P+) were shown to cause a persistent infection that might be attributed to their ability to form a biofilm, as demonstrated previously in vitro (Chokr et al., 2006). Surprisingly, negative strains (B−, I−, P−) recovered from two origins, skin (commensal) and infected sites (infecting), also developed and maintained an infection in the animal model (Chokr et al., 2007). Such strains may possibly be able to form a biofilm in vivo without PNAG.

Testing of other S. epidermidis from the same collection (Table 1) indicates the presence of two B+, I+, P+ strains that are completely unable to develop an infection in spite of possessing the ica locus and forming a biofilm in vitro. This result indicates that in the TC-GP model, not all the clinical strains are able to develop and maintain an infection.

Three negative B−, I−, P− clinical and commensal strains showed, to some extent, a capacity to develop and maintain an infection. Such strains may form a biofilm in vivo without PIA. The presence of a significant amount of bacteria after sonication in the implants infected by these strains could indicate their presence in a biofilm form. It is also conceivable that these negative strains may develop and maintain an infection without a biofilm. Further experiments are needed to evaluate the capacity of the different strains to form a biofilm in vivo. However, the fact that the strains belonging to the ‘B+, I+, P+’ type showed a high capacity to cause persistent infections, compared with the opposite ‘B−, I−, P−’ type, emphasized the potential role of PNAG and the ica locus in the pathophysiology of strains. Whatever the strains, the exact mechanism responsible for virulence remains to be determined, and it can be assumed that subspecies-specific differences exist in the abilities of S. epidermidis isolates to form a biofilm and to cause infection in vivo.

PNAG is not a suitable antigen to diagnose infections related to medical implants

The early detection of the medical device-related staphylococcal infections is difficult using the classical tools of microbiological analyses. During an implant-related biofilm infection, the quantity of bacteria in the bloodstream is very low, and their direct detection is nearly impossible. The diagnosis is often made only at advanced stages of infection, when severe complications occur: formation of abscesses, pain, and unsealing of the prosthetic devices. Specific and noninvasive laboratory tests to diagnose these infections are not yet available.

Because the pathogenicity of S. epidermidis is mostly due to its ability to colonize indwelling polymeric devices and form a biofilm, a diagnostic test could be based on the detection of antibodies specific for biofilm components of CoNS, particularly S. epidermidis.

A detection of specific ‘antibiofilm’ antibodies in the blood serum of patients could serve as a convenient noninvasive and inexpensive diagnostic tool for the detection of foreign body-associated infections. However, no antigens specific for staphylococcal infection have been identified.

Different extracellular antigenic preparations have been proposed by different authors as candidates for immunological tests: an extracellular extract of a clinical S. epidermidis strain (staphylococcal slime polysaccharide antigen, Selan et al., 2002), a ‘20-kDa sulphated polysaccharide’, an ‘80-kDa peptidoclycan’ (Karamanos et al., 1997; Georgakopoulos et al., 2002; Lamari et al., 2004), or an extracellular ‘lipid S’ of S. epidermidis (Elliott et al., 2000). In most cases, the chemical structure of the antigens has not been determined. To date, none of these antigens have led to the development of a commercialized diagnostic test.

We have chosen to test, as an antigen for a serodiagnostic, the PNAG, a characteristic and well-characterized component of staphylococcal biofilms (Sadovskaya et al., 2007).

As a first step of our study, we investigated cases of chronic infections caused by the strains known as PNAG producers. This problem could be addressed thanks to a TC-GP animal model, mimicking an implant-related infection (Chokr et al., 2007), and a collection of staphylococcal strains with a well-characterized biofilm composition (Sadovskaya et al., 2005, 2006). We developed a sensible ELISA essay, which included coating the Microlon 600 plates with the preparations of purified PNAG, incubation with the animal or human sera, and detection of the bound anti-PNAG antibodies with the appropriate HRP- or AP-conjugated secondary antibodies (Sadovskaya et al., 2007).

We have shown that in the chosen animal model, the levels of anti-PNAG antibodies were significantly higher in guinea-pigs infected with S. epidermidis RP62A compared with healthy animals (P>0.01). When the evolution of an antibody response to PNAG in individual guinea-pigs was studied, we observed an increase of the level of antibodies following the implant-related infection.

The results were more ambiguous with human sera. Screening of patients' sera and the sera of healthy individuals reveals a relatively high level of anti-PNAG immunoglobulin Gs (IgGs) in the sera of healthy controls. The level of these IgGs in patients' sera was very variable and overall higher, but the difference was insignificant (P>0.05).

If this result is rather disappointing, it is nevertheless interesting to try to understand the reason for this phenomenon. Despite the fact that the presence of the ica operon is considered as a marker discriminating between clinical device-associated strains and skin flora (Galdbart et al., 2000; Kozitskaya et al., 2005), a significant percentage of commensal CoNS strains in healthy individuals is ica-positive and potentially capable of producing PNAG. The presence of anti-PNAG IgGs in the sera of healthy individuals could thus be explained by their natural exposure to PNAG-producing CoNS and Gram-negative bacteria, the possible presence of these antigens in common vaccine preparations, as well as previous infections and nasal carriage of S. aureus.


Biofilm is considered as a main virulence factor of CoNS, a major cause of medical implant-associated infections. Targeting the bacterial biofilm state and particularly the EPS matrix might be a key for the development of therapeutic tools against these infections. We have particularly focused on the biofilm of S. epidermidis and its involvement in infections associated with orthopaedic prostheses. The construction of a collection of strains recovered from infected prostheses enabled us to confirm the predominance of S. epidermidis as the leading species related to this type of nosocomial infections. The majority (73%) of the CoNS isolated from clinically diagnosed infected patients possessed the ica locus, but only 26% produced PNAG and 33% formed a biofilm. Variabilities in the capacity to form a biofilm in vitro and maintain chronic infections in vivo in an animal model were observed.

A direct analytical approach and a detailed analysis of literature data allowed us to clarify some ambiguities and to conclude that PIA and PS/A (also referred to as SAA, PNSG, and SAE) have the same chemical structure – a PNAG, and differ only by the degree of a positive and a negative charge due to substitution. PNAG of several clinical strains associated with orthopaedic prosthesis infections were purified and analysed using chemical methods and NMR spectroscopy.

We have clearly established that the staphylococcal biofilm can be subdivided into two categories, based on the presence of the PNAG among the EPS of its biofilm matrix, with two recurring constituents that are TAs and proteins. Taking into account the versatility and genomic plasticity of staphylococci, it is not excluded that same bacteria should be able to develop a biofilm with or without PNAG depending on their surrounding environment. This is evidenced by the ability of S. epidermidis to switch to a protein-dependent biofilm when PNAG production is abolished (Hennig et al., 2007). In a strategy to combat the biofilm, this major result affects diagnosis and therapy approaches.

The detachment and dispersal of staphylococcal biofilms is not always efficient after enzymatic hydrolysis of PNAG. Hydrolysis of biofilm proteins with proteases or depolymerization of the EC-TA would be more efficient for dispersal of the staphylococcal biofilm. However, in situ treatment by the proteases unfortunately may have side-effects on patient tissues surrounding the infected prosthesis. Targeting the EC-TA as a biofilm constituent might be more specific.

PNAG does not seem to be a convenient antigen for serodiagnostics of implant-related staphylococcal infections, because it does not sufficiently discriminate patients and healthy individuals.

Our studies on the animal model showed that CoNS do not necessarily have the same properties in vitro and in vivo. To understand how biofilm development contributes to infectious disease, in vivo studies remain insufficiently developed and deserve more attention. It would also be useful to extend the bacterial models of S. epidermidis to more representative clinical specimens encountered in associated implant infections.