OUP user menu

Lack of biofilm contribution to bacterial colonisation in an experimental model of foreign body infection by Staphylococcus aureus and Staphylococcus epidermidis

Patrice Francois, Patrick H. Tu Quoc, Carmelo Bisognano, William L. Kelley, Daniel P. Lew, Jacques Schrenzel, Sarah E. Cramton, Friedrich Götz, Pierre Vaudaux
DOI: http://dx.doi.org/10.1016/S0928-8244(02)00463-7 135-140 First published online: 1 March 2003


The contribution of in vivo biofilm-forming potential of Staphylococcus aureus and Staphylococcus epidermidis was studied in an experimental model of foreign body infections. Increasing inocula (from 102 to 107 organisms) of ica-positive strains of S. aureus and S. epidermidis and their ica-negative isogenic mutants (the ica locus codes for a major polysaccharide component of biofilm) were injected into subcutaneously implanted tissue cages in guinea pigs. Surprisingly, bacterial counts and time-course of tissue cage infection by ica-positive strains of S. aureus or S. epidermidis were equivalent to those of their respective ica-negative mutants, in the locally infected fluids and on tissue-cage-inserted plastic coverslips.

  • Foreign body infection
  • Staphylococcus aureus
  • Staphylococcus epidermidis
  • Biofilm
  • ica gene

1 Introduction

Staphylococcal infections are a major source of patient morbidity and implant failure [1]. Animal models have confirmed the high susceptibility of artificial implants to low inocula of staphylococci [2]. We previously developed an experimental model of foreign body infections by implanting polymethylmethacrylate (PMMA) subcutaneously into guinea pigs and then infecting animals with low inocula (102–103 CFU) of Staphylococcus aureus or Staphylococcus epidermidis[3,4]. In subsequent studies, PMMA coverslips were inserted inside tissue cages to study the characteristics of S. aureus and S. epidermidis host-modulated attachment on such implanted materials [57].

Numerous studies in the past two decades have shown that S. epidermidis and other pathogenically relevant coagulase-negative staphylococcal (CNS) species can produce conspicuous bacterial biofilms on the surface of biomedical devices [810]. Biofilm production, which is a leading cause of implant colonisation and sepsis by S. epidermidis, is thought to require at least two sequential steps: (i) attachment of individual cells to a solid substrate; (ii) elaboration of an intercellular biofilm interconnecting the cells and leading to multiple layers of bacterial cells. The major extracellular component of biofilm has been identified as the polysaccharide intercellular adhesin (PIA) [11]. An almost identical polysaccharide was also described as a capsular surface component of S. epidermidis[12]. PIA is composed of linear β-1,6-linked glucosaminylglycans and can be synthesised in vitro from UDP-N-acetylglucosamine [13] by products of the intercellular adhesion (ica) locus [14]. The ica locus is an operon composed of icaR (regulator) and icaADBC (biosynthesis genes), first described as highly prevalent in S. epidermidis strains responsible for device-related infections [15,16], but subsequently detected in all strains of S. aureus regardless of their ability to form a biofilm in vitro [17]. Sequence comparison of S. aureus and S. epidermidis ica genes revealed a quite good homology, leading to 59–78% amino acid identity. Deletion of the ica locus was found to result in a loss of the ability to form biofilms, produce PIA or mediate N-acetylglucosaminyl-transferase activity in vitro. Cross-species hybridisation experiments revealed the presence of icaA in several other staphylococcal species, suggesting that cell–cell adhesion and the potential to form biofilms is conserved within this genus [17].

The virulence of biofilm-negative mutants compared to their biofilm-positive parental strains of S. epidermidis of bacterial biofilm production has been evaluated in several animal models [18,19]. These studies yielded contradictory findings which were both model- and strain-dependent [1825].

To evaluate the potential contribution of biofilm production by S. aureus and S. epidermidis in the aforementioned guinea pig model of foreign body infections, increasing inocula of biofilm-producing wild-type bacteria compared to their biofilm-defective mutants were injected into subcutaneously implanted tissue cages in guinea pigs. The time-course of tissue-cage infection was followed by bacterial counts performed in tissue-cage fluids and on coverslips explanted from sacrificed animals. A similar approach was taken to evaluate the infectivity and biofilm production contribution of recently developed ica-negative isogenic mutants of either S. aureus or S. epidermidis.

2 Materials and methods

2.1 Bacterial strains

The bacterial strains used for these studies were (i) S. epidermidis 1457 and its PIA-negative isogenic mutant 1457-M11, that was produced by insertion of transposon Tn917 at nucleotide 1031 of the icaA gene [11,26]; (ii) S. aureus strain SA113 (ATCC 35556) and its ica knockout mutant SA113Δica:tet[17]; (iii) S. epidermidis O-47 and its ica knockout mutant O-47Δica::tet[27].

Bacterial cultures prepared for experimental tissue-cage infections were first performed overnight in tryptic soy broth (TSB) without glucose (Difco), followed by washing and resuspension in 0.9% NaCl. For experimental infections, each bacterial strain was grown for 4 h in TSB, followed by washing and resuspension in 0.9% NaCl. Just before inoculation of calibrated bacterial suspensions into tissue cages, the washed bacterial cultures were briefly (60 W, 1 min) sonicated (model 2200; Brandson Ultrasonics, Branburry, CT, USA) to remove any biofilm residue and minimise bacterial clumping as described [28].

2.2 In vivo experiments

All animal experiments were approved by the Ethics Committee of the Faculty of Medicine of the University of Geneva and by the Veterinary Office of the State of Geneva. Four polytetrafluorethylene (Teflon) multiperforated tissue cages, each containing one PMMA coverslip (7×7 mm), were implanted subcutaneously in guinea pigs under aseptic conditions as described in detail previously [3,4]. The tissue-cage-inserted coverslips were used to quantify in vivo bacterial adhesion onto well defined flat surfaces. An important characteristic of the tissue-cage model is that the perforations permit the influx of inflammatory cells into tissue-cage fluid, in particular polymorphonuclear neutrophils and monocytes, as previously described [3]. At 3 weeks after implantation, tissue-cage fluids were aseptically aspirated and were checked for sterility.

Experimental tissue infections were induced by inoculating each tissue cage with 0.1 ml saline containing either 102, 103, or 104 CFU, prepared by serial 10-fold dilutions of sonicated bacterial suspensions of ica-positive or ica-negative strains. When indicated, tissue cages were also challenged with higher bacterial inocula, ranging from 105 to 107 CFU.

At 1, 2 or 3, and 7 days (when indicated) after the local injection of each strain, quantitative cultures were performed by plating 0.1 ml of serially, 10-fold diluted, tissue-cage fluid on Mueller–Hinton agar (MHA), with a detection limit of 102 CFU ml−1.

On the last day of the experimental protocol, all tissue cages and inserted coverslips were aseptically removed from sacrificed animals. One coverslip from each cage was used to quantify the number of surface-attached viable bacteria. These were removed from explanted coverslips by sonication (60 W, 1 min), performed as described above in sterile tubes containing 1 ml of Mueller–Hinton broth (MHB). One 0.1-ml portion of suspended organisms, either undiluted or serially 10-fold diluted, was immediately plated on MHA, with a detection limit of 10 CFU per coverslip [28]. Each culture of sonicated coverslip was further incubated in MHB at 37°C for 4 days, to detect lower amounts of surface-attached bacteria (1–9 CFU). To verify that the in vitro biofilm-forming potential of parental strains of S. epidermidis or S. aureus removed from explanted coverslips was intact, biofilm production was assessed by the tube test as described by Christensen et al. [23] following the 4-day culture in MHB.

2.3 Statistics

The results are expressed as mean values±standard errors of the means (S.E.M.). Statistical analysis was performed by Student's two-tailed test and defined as P value. A P<0.05 was considered significant.

3 Results

3.1 Time course and dose-response of tissue-cage infections

Fig. 1 summarises the results of three experiments designed to explore the time course of tissue-cage infections and the extent of coverslip colonisation by the biofilm-forming strain 1457 of S. epidermidis compared to its isogenic biofilm-negative mutant 1457-M11. The first experiment (panel A) shows average bacterial counts reached by the parental strain and its PIA-defective mutant in tissue-cage fluids, at 1, 2 and 7 days after tissue-cage injection of equivalent inocula of 103 CFU per cage. This inoculum was sufficient to produce permanent but low-grade tissue-cage infections with either strain, yielding viable counts ranging from 3.17 to 4.23 log10 CFU ml−1 of tissue-cage fluid. While strain 1457 showed somewhat higher counts in tissue-cage fluids at days 1 and 2 compared to the PIA-defective mutant 1457-M11, CFU counts of the parental strain unexpectedly declined by 1 log10 from day 2 to day 7, which turned out to be equivalent to those of strain 1457-M11. These marginal viable counts of strain 1457 or 1457-M11 scored not only in tissue-cage fluids but also on tissue-cage-inserted coverslips (data not shown) after 7 days of infection were not suitable for performing biochemical studies of biofilm production during infection.

Figure 1

Experimental tissue-cage infections in guinea pigs by either S. epidermidis strain 1457 or its Tn917-inserted ica mutant strain 1457-M11. A: Numbers of viable counts (CFU ml−1) in tissue-cage fluid (n=8) scored at 1, 2, and 7 days after being locally challenged with 3 log10 CFU of either strain. Each value is the mean (±S.E.M.) of eight cages for strain 1457 and 1457-M11, respectively. B: Numbers of viable counts in tissue-cage fluid (CFU ml−1) or on tissue-cage-inserted coverslips (CFU per coverslip) scored at 1 day after being locally challenged with increasing inocula, ranging from 4.45 to 7.45 log10 CFU for strain 1457 or from 4.60 to 7.60 log10 CFU for strain 1457-M11. Each value is the mean (±S.E.M.) of four cage fluids or coverslips for strain 1457 and 1457-M11, respectively. C: Mean (±S.E.M.) numbers of viable counts in tissue-cage fluid (CFU ml−1) or on tissue-cage-inserted coverslips (CFU per coverslip) scored 1 or 2 days after being locally challenged with a high inoculum of either strain 1457 (7.15 log10 CFU) or 1457-M11 (7.28 log10 CFU). Each value is the mean (±S.E.M.) of four cage fluids or coverslips for either strain. The broken lines indicate the limit of detection of CFU counts.

To improve experimental conditions for studying in vivo biofilm production, the dose-responses of tissue-cage infections were monitored using increasing bacterial inocula of strains 1457 and 1457-M11. Fig. 1B shows that viable counts scored at 24 h after the bacterial challenge remained constant in either infected tissue-cage fluids or inserted coverslips despite 10-fold and 100-fold increases in the challenging inoculum (from 4.45 to 6.45 log10 CFU per cage) of strain 1457. A significant increase in viable counts of the parental strain in tissue-cage fluid and on explanted coverslips was obtained after inoculating the tissue cages with 7.45 log10 CFU. In contrast, the mutant 1457-M11 showed no significant difference in both tissue-cage fluids and coverslip-associated viable counts, despite 1000-fold differences (from 4.60 to 7.60 log10 CFU per cage) in the challenging inocula. Together, these data indicate that the relatively low counts of strain 1457 and its PIA-defective mutant 1457-M11 in tissue-cage fluids and inserted coverslips were not caused by inadequately low inocula.

To further explore the time-course of tissue-cage infections induced by high-challenging inocula (>107 CFU per cage) of strains 1457 and 1457-M11, we assessed their respective viable counts after 24 and 48 h in infected fluids and on inserted coverslips. Fig. 1C indeed confirmed that strain 1457 yielded transiently higher CFU counts in tissue-cage fluids at 24 h compared to strain 1457-M11, but that this situation was reversed at 48 h. This change reflected a significant decline in viable counts of strain 1457 which did not occur in the mutant 1457-M11. Furthermore, average viable counts of bacteria removed from explanted coverslips were quite low (ranging from 3.30 to 3.96 log10 CFU per coverslip) and failed to increase from 24 to 48 h for either strain 1457 or its mutant 1457-M11.

Collectively, the data shown in Fig. 1 indicated that local conditions in the tissue-cage infection model failed to confer a significant advantage to the biofilm-producing strain 1457 over its biofilm-negative mutant 1457-M11 in terms of infection rate and quantitative biomass accumulation on the surface of tissue-cage-inserted coverslips. In all cages infected with strain 1457, the in vitro biofilm-forming potential of this strain, as opposed to its mutant 1457-M11, cultured from explanted coverslips was intact, as assessed by the tube test method [23] at the end of the 4-day culture in MHB.

3.2 Magnitude of tissue-cage infections by S. aureus and S. epidermidis and their ica mutants

Fig. 2 shows the time-course of tissue-cage infections at 1 and 3 days after challenging each tissue cage with 103 CFU of S. aureus strain SA113 compared to its knockout mutant SA113Δica::tet, or S. epidermidis O-47 compared its ica knockout mutant O-47Δica::tet. The average CFU counts reached by strain SA113 and its ica mutant in tissue-cage fluids were nearly identical at day 1, approaching 5.5 log10 CFU ml−1, then slightly increased until day 3 but with no significant difference emerging between the ica-positive and ica-negative strains. Furthermore, viable counts of parental strain SA113 (5.27±0.22 log10 CFU) on tissue-cage-inserted coverslips were only slightly but non-significantly higher than those of its ica mutant (4.71±0.29 log10 CFU).

Figure 2

Experimental tissue-cage infections in guinea pigs by either S. aureus strain SA113 or its mutant strain SA113Δica::tet, S. epidermidis strain O-47 (SEO47) or its ica mutant strain O-47Δica::tet, monitored by the number of viable counts (CFU ml−1) in tissue-cage fluid at 1 and 3 days, after being locally challenged with 3 log10 CFU of each strain. Each value is the mean (±S.E.M.) of eight cage fluids for each strain.

In comparison, the average viable counts reached by strain O-47 and its ica mutant in tissue-cage fluids and coverslips were significantly lower than those of S. aureus SA113 and its ica mutant. Neither strain exceeded 4.5 log10 CFU ml−1 in tissue-cage fluids at either day 1 or 3, and the viable counts of the ica mutant were slightly but nonsignificantly higher than those of its parent. Finally, viable counts of parental strain O-47 (2.49±0.34 log10 CFU) and its ica mutant (2.78±0.29 log10 CFU) were so low on tissue-cage-inserted coverslips that they prevented any further study on in vivo biofilm production by these particular strains. Three of the six implanted coverslips yielded even <2 log10 CFU of strain O-47 per coverslip with one of them scored as culture-negative by the tube test.

In all cages infected with either strain SA113 or O-47, the in vitro biofilm-forming potential of these strains, as opposed to their respective ica mutants, cultured from explanted coverslips was intact, as assessed by the tube test method [23] at the end of the 4-day culture in MHB.

4 Discussion

In the past two decades, several experimental studies performed in a variety of animal models tried to identify major in vivo parameters modulating biofilm production and promoting infections of biomedical implants by S. epidermidis and some other staphylococcal species. A global view of these recently summarised studies [9,10,23,29,30] cannot be obtained because they represent quite heterogeneous experimental conditions that led to quite contrasting results. The reasons for these heterogeneous results were: (i) the comparison of strains of different genetic backgrounds in many early studies [23]; (ii) the high strain-dependent variability in the virulence of wild-type strains of S. epidermidis in animal models [22] as well as in humans. In both animals and humans, S. epidermidis is generally considered as an opportunistic pathogen primarily affecting immunocompromised hosts or recipients whose defences may be locally impaired by the presence of temporary or permanent medical implant [7,23]. The recent development of syngeneic wild-type and ica mutant strains of S. epidermidis or S. aureus was therefore seen as a decisive step to obtain more carefully controlled animal data assessing the in vivo activity of the ica genes and their contribution to bacterial virulence [10].

Both sets of isogenic parental strains 1457 and O-47 and their respective ica mutants used for our study were previously evaluated in two different animal models, namely a mouse model with a subcutaneously implanted catheter [18] and a rat central venous catheter-associated infection model [19,25], which, in contrast to our data, demonstrated the significant role of biofilm production for the virulence of S. epidermidis. The apparent conflicting data derived from both aforementioned murine models with our own findings in a guinea pig tissue-cage model that showed no significant influence of the ica genes on S. aureus and S. epidermidis virulence might be explained by the quite different setting and time-frame of these different infection models: (i) a special characteristic of the tissue-cage model in guinea pigs is the prolonged period separating the time of subcutaneous polymeric implant surgery from the time of experimental infection. This 3–4-week lag time is mandatory for wound healing and required for minimising the risk of spontaneous tissue-cage infection by skin flora microorganisms [3]. During this postsurgical period, both implanted tissue cages and inserted coverslips become coated with an unknown number of host factors that may modulate bacterial adhesion and host defences in a quite complex and still poorly understood manner [7]; (ii) in contrast, most animal studies that demonstrated the contribution of biofilm production in polymer-associated infections used implant surfaces either colonised with in vitro preformed bacterial biofilms [21] or infected by S. epidermidis injection shortly (<60 min) after subcutaneous polymer implantation. These time-dependent conditions might be critical for promoting bacterial biofilm production before the development of a strong host defence response; (iii) in other animal models of medical device infections, injection of bacterial challenges proceeded directly via infusion lines [19,25]. This mode of bacterial administration, which to some extent mimics clinical device infections during blood exchange (e.g. hemodialysis) or fluid administration (intravenous catheters), may represent mixed situations where bacteria are not only located in the intravascular system but also might persist outside the blood flow in the extracorporeal portion of catheters as biofilm-forming microbial reservoirs in direct contact with polymer surfaces [19,25]. A nice description of this mixed situation refers to the development of bacterial biofilms on silastic catheter materials in peritoneal dialysis fluid [31].

The major findings of our animal model can be summarised as follows: (i) both biofilm-producing parental strains 1457 and O-47 of S. epidermidis did not readily attach to and colonise the tissue-cage-inserted coverslips regardless of the challenging inoculum and duration of infection from 1 to several days; (ii) in contrast, S. aureus strain SA113 was recovered in much higher amounts than S. epidermidis on tissue-cage-inserted coverslips, as well as in tissue-cage fluids. Accordingly, the capacity of the different strains to attach to and colonise inserted coverslips was fully independent of their in vitro ability to form a biofilm. A further argument for the independence of inserted coverslips colonisation and biofilm formation was provided by the biofilm-defective mutants that colonised the implanted surfaces to the same extent as their wild-type parents. Thus, the very poor attachment of S. epidermidis compared to S. aureus parental and mutant strains may bring additional support for a significant involvement of host-derived factors specifically interacting with staphylococcal adhesins, collectively designed as MSCRAMMs (microbial surface components recognising adhesive matrix molecules) [32]. Several studies indicate that S. aureus strains can produce a much higher number of MSCRAMMs than S. epidermidis isolates [57,32,33]. The low colonisation rate and lack of significant biofilm formation by S. epidermidis strains on implanted coverslips may therefore result from a defective adhesion mechanism under these in vivo conditions, which may indirectly prevent further significant development of a biofilm. Alternatively or in concert, the local conditions in tissue-cage fluids might downregulate activation of the ica operon required for the production of PIA due to local conditions in tissue-cage fluid [10]. Recent data indicate that expression of the ica operon and PIA synthesis is tightly regulated by various environmental and stress factors in S. epidermidis[17,27,3437,10].

In conclusion, this study indicates that the higher potential of S. aureus to infect implanted tissue-cage fluid and coverslips compared to S. epidermidis likely reflects the global contribution of the much larger number of virulence factors expressed by the former species rather than the isolated contribution of biofilm production [38]. One essential aspect of the well-known virulence of S. aureus is its ability to exhibit improved resistance to host clearance mechanisms compared to S. epidermidis via the elaboration of various cell-damaging toxins. While our observations do not rule out the importance of biofilm production for polymer-associated infections by S. epidermidis and to some extent S. aureus, they rather indicate the necessity for infecting bacteria to find suitable in vivo environments promoting biofilm production before emergence of a strong host antibacterial response.


This work was supported in part by research grants 3200-63710.00 (to P.V.) and 632–57950.99 (to J.S.) from the Swiss National Science Foundation, and by DFG FOR449/GO371/5 (to F.G). The authors thank D. Mack for providing S. epidermidis strains 1457 and 1457-M11, and M. Bento and E. Huggler for technical assistance.


  1. [1].
  2. [2].
  3. [3].
  4. [4].
  5. [5].
  6. [6].
  7. [7].
  8. [8].
  9. [9].
  10. [10].
  11. [11].
  12. [12].
  13. [13].
  14. [14].
  15. [15].
  16. [16].
  17. [17].
  18. [18].
  19. [19].
  20. [20].
  21. [21].
  22. [22].
  23. [23].
  24. [24].
  25. [25].
  26. [26].
  27. [27].
  28. [28].
  29. [29].
  30. [30].
  31. [31].
  32. [32].
  33. [33].
  34. [34].
  35. [35].
  36. [36].
  37. [37].
  38. [38].
View Abstract