OUP user menu

Quantitative analysis of biofilm formation of methicillin-resistant Staphylococcus aureus (MRSA) strains from patients with orthopaedic device-related infections

Hideki Kawamura, Junichiro Nishi, Naoko Imuta, Koichi Tokuda, Hiroaki Miyanohara, Teruto Hashiguchi, Michihisa Zenmyo, Takuya Yamamoto, Kosei Ijiri, Yoshifumi Kawano, Setsuro Komiya
DOI: http://dx.doi.org/10.1111/j.1574-695X.2011.00821.x 10-15 First published online: 1 October 2011


Biofilms play a pivotal role in medical device-related infections. However, epidemiological analysis of biofilm formation and genotyping among clinical methicillin-resistant Staphylococcus aureus (MRSA) isolates from patients with orthopaedic infections has rarely been reported. A total of 168 MRSA strains were examined: 23 strains from patients with device-related infection (the device group); 55 from patients with device-non-related infection (the nondevice group); and 90 from asymptomatic nasal carriers (the colonization group). Pulsed-field gel electrophoresis analysis and five genotyping methods including agr typing were performed. Biofilm formation was quantified using a microtitre plate assay. The device group had a significantly higher incidence of agr-2 than the colonization group (78.3% vs. 34.4%, P=0.001). The biofilm index of the agr-2 (0.523 ± 0.572) strains was significantly higher than those of agr-1 (0.260 ± 0.418, P<0.0001) and agr-3 (0.379 ± 0.557, P=0.045). The prevalence of strong biofilm formers in the device group (43.5%) was significantly higher than that in the nondevice group (12.7%, P=0.003) and the colonization group (20.0%, P=0.020). agr-2 MRSA strains may be more likely to cause orthopaedic device infection because of their strong biofilm formation ability.

  • MRSA
  • biofilm
  • genotyping
  • orthopaedics
  • medical device


Staphylococcus aureus is one of the common pathogens in orthopaedic infections including bone/joint infections and surgical site infections (SSIs) (Cunningham et al., 1996; Mangram et al., 1999). Methicillin-resistant S. aureus (MRSA) has been increasing as a causative organism in healthcare-associated infections and the incidence of methicillin-resistant strains among S. aureus is now about 50–70% in Japan (Arakawa et al., 2000).

There are many cases in which medical devices (e.g. arthroplasty, internal fixation for bone fracture, etc.) are required in orthopaedic surgery. Biofilm formation has been considered a virulence factor contributing to infections associated with medical devices (Reid, 1999). Biofilm acts as a reservoir for bacteria, making eradication difficult. In many cases of device-related infection, the removal of orthopaedic hardware is required, decreasing activities of daily life (Patel et al., 2008). Implant-associated MRSA infections are associated with a high rate of relapse (Ferry et al., 2010).

The molecular basis of biofilm formation by staphylococci has been studied. The initiation of biofilm formation requires the adherence of cells to a surface, followed by the formation of microcolonies, which develop into a mature biofilm structure (Agarwal et al., 2010). The intercellular adhesion (ica) gene regulates extracellular polysaccharide adhesion. The accessory gene regulator (agr) quorum-sensing (QS) system has been linked to biofilm formation and regulates microbial surface components recognizing adhesive matrix molecules (MSCRAMMs) binding extracellular matrix (e.g. fibronectin, fibrinogen, collagen, and bone sialoprotein) (O'Gara, 2007).

However, epidemiological analysis including biofilm formation and genotyping among clinical MRSA isolates from orthopaedic infections has rarely been performed. The aim of this study was to analyse biofilm formation and perform genotyping among MRSA isolates from patients with orthopaedic infection and asymptomatic carriers, and investigate bacterial factors associated with orthopaedic device-related infections involving MRSA.

Materials and methods

Bacterial isolates

In the Kagoshima University Hospital Orthopaedic Unit, we found 87 patients with orthopaedic infection due to MRSA during the 11-year period from 1999 to 2009 (infection group) and 95 asymptomatic outpatient nasal MRSA carriers scheduled for surgery during the 9-year period from 2001 to 2009 (colonization group). The diagnosis of infection was based on the criteria proposed by the US Centers for Disease Control and Prevention (Horan et al., 2008). Eighty-seven strains of the infection group were divided into two groups: a device group of 23 strains and a nondevice group of 64 strains. The device group included nine cases of prosthetic joint infections, seven cases of SSI after spinal fusion surgery or cervical laminoplasty with a hydroxyapatite spacer, and seven cases of SSI after internal fixation. The nondevice group included 42 device-non-related cases of SSI, seven of pyogenic arthritis, two of pyogenic spondylitis, three of osteomyelitis, and 10 of skin and soft tissue infection. Only one MRSA strain was selected from each patient.

Specimens were incubated on a 5% sheep blood agar at 37 °C for 24 h. Staphylococcus aureus was identified by colony morphology and a coagulase test. Resistance to methicillin was determined by subculturing the isolate on MRSA screening plates (Becton Dickinson, Franklin Lakes, NJ) at 35 °C for 24 h. The strains were individually stored at −80 °C in 10% skim milk.

DNA extraction

As patients were considered to be infected by a single population of MRSA, one colony of bacterial cells was suspended in 100 µL Tris-EDTA with 4 U of lysostaphin added and heated at 95 °C for 10 min. Centrifugation for 2 min at 25 000 g was performed to remove cell debris. Supernatants were frozen at −30 °C.


All of the MRSA isolates collected were examined using six genotyping methods including Staphylococcal cassette chromosome mec (SCCmec) typing (Oliveira & Lencastre, 2002), toxin genotyping (Mehrotra et al., 2000), Staphylococcus protein A (spa) typing (Harmsen et al., 2003), agr typing (Jarraud et al., 2000), mec-hypervariable (HVR) region genotyping (Nishi et al., 2002), and pulsed-field gel electrophoresis (PFGE) analysis. spa typing was performed by sequencing using the spa typing website (http://www.spaserver.ridom.de). agr typing was performed by multiplex PCR with division into four agr types based on the variable regions of agrD or agrC sequences. PFGE was performed using the CHDF-ER III system (Bio-Rad, Hercules, CA). DNA was digested by SmaI (Takara Bio Inc., Shiga, Japan) and electrophoretic patterns were analysed according to criteria described by Tenover (1995).

Detection of adhesion genes

PCR was used to detect icaAD (Ariciola et al., 2001), fibronectin-binding protein (fnb) AB (Montanaro et al., 1999), collagen-binding protein (cna) (Ariciola et al., 2005), clumping factors (clf) A binding to fibrinogen (Ando et al., 2004), and bone sialoprotein-binding protein (bbp) (Tristan et al., 2003). The reactions were performed in a 15 µL volume containing each primer (1 µM each), together with 1 µL of extracted DNA solution, 200 µM each of dATP, dCTP, dGTP, and dTTP, 0.5 U of Taq polymerase (Takara Bio Inc.), and buffer (10 mM Tris-HCl, 50 nM KCl, 1.5 mM MgCl2). Amplification reactions were performed in a Thermal Cycler PC 707 (Astec, Fukuoka, Japan). The PCR products were electrophoresed on agarose gels.

Biofilm quantification

Biofilm formation by MRSA strains on polystyrene was quantified using the microtitre plate assay first described by Christensen and colleagues (Christensen et al., 1985; Manago et al., 2006). Bacterial strains were incubated in 96-well polystyrene microtitre plates (Becton Dickinson) at 37 °C for 18 h. The cells were decanted and the wells were washed with tap water. Adhesive cells were stained with 0.1% crystal violet for 5 min and resolved by the addition of 200 µL of 95% ethanol after the wells were washed with tap water. Absorbance was measured at 595 nm using an ELISA plate reader and the biofilm index was defined at OD595 nm. We performed the biofilm assay of each strain in duplicate in each microtitre plate. In addition, each test was carried out three times.

Statistical analysis

Statistical analysis was performed using the two-tailed χ2-test for independence and the Mann–Whitney U-test using spss 11.0J (SPSS Japan Inc.). Findings of P<0.05 were considered significant.


A total of 182 strains were classified by SCCmec typing (type I, 38 strains; type II, 91; type III, 3; type IV; 50), toxin typing (sec/tst, 91; seb, 26; see, 2; sec, 1; sed, 1; eta, 1; etb, 1; tst, 1; negative, 58), spa typing (t002, 62; t1767, 54; t067, 7; t375, 6; t985, 5; t008, 4; t579, 3; t3148, 3; others, 38), and mec-HVR typing (A, 2; A″, 2; B, 13; C, 10; D, 87; E, 64; F, 4). PFGE analysis of 182 strains demonstrated 101 patterns. As a result, these genotyping methods classified the device group into 20 patterns, the nondevice group into 37 patterns, and the colonization group into 68 patterns. There were seven patients in the colonization group who developed orthopaedic infection (four device infections and three nondevice infections). Among them, five cases showed similar genetic patterns including PFGE between the preoperative nasal isolate and the subsequent surgical wound isolate (four cases in the device group and one patient in the nondevice group). We deleted the five cases in the colonization group. As several strains showed a similar genetic pattern in the nondevice group at <3-month intervals, we excluded nine redundant strains from the nondevice group in order to avoid the influence of an outbreak situation because the study was conducted exclusively in our hospitals. As a result, we analysed 23 strains in the device group, 55 strains in the nondevice group, and 90 strains in the colonization group.

The prevalence of genotypes involving the SCCmec typing, spa typing, toxin typing, mec-HVR typing, and agr typing is presented in Table 1. The most major clone in this study is IV/t1767/1/E/sec+tst (SCCmec/spa/agr/mec-HVR/toxin) and the second major clone is II/t002/2/D/sec+tst.

View this table:

Prevalence of combined genotypes (SCCmec/spa/agr/mec-HVR/toxin) no. (%) in each group

Device group (n=23)Nondevice group (n=55)Colonization group (n=90)
GenotypeNo. (%)GenotypeNo. (%)GenotypeNo. (%)
II/t002/2/D/sec+tst4 (17.4)II/t002/2/D/sec+tst7 (12.7)IV/t1767/1/E/sec+tst15 (16.7)
II/t002/2/D/seb4 (17.4)II/t002/2/D/seb7 (12.7)I/t1767/1/E/-7 (7.8)
II/t002/2/D/-3 (13.0)I/t1767/1/E/-7 (12.7)II/t002/2/B/sec+tst5 (5.6)
II/t002/2/B/sec+tst2 (8.7)IV/t1767/1/E/sec+tst5 (9.1)II/t002/2/D/sec+tst5 (5.6)
II/t002/2/C/sec+tst1 (4.3)II/t002/2/D/-4 (7.3)II/t002/2/D/seb4 (4.4)
IV/t1767/1/D/sec+tst1 (4.3)II/t002/2/A″/sec+tst2 (3.6)IV/t1767/1/E/-4 (4.4)
IV/t1767/1/C/sec+tst1 (4.3)II/t067/2/D/sec+tst2 (3.6)II/t002/2/D/-3 (3.3)
Other types7 (30.4)II/t002/2/A/sec+tst1 (1.8)II/t067/2/D/sec+tst3 (3.3)
II/t002/2/B/sec+tst1 (1.8)I/t1767/1/D/-3 (3.3)
IV/t1767/1/D/sec+tst1 (1.8)IV/t1767/1/D/sec+tst2 (2.2)
IV/t1767/1/C/sec+tst1 (1.8)IV/t1767/1/C/sec+tst1 (1.1)
Other types17 (30.9)II/t002/2/A/sec+tst1 (1.1)
Other types37 (41.1)

The prevalences of agr types are shown in Table 2. The prevalence of agr-2 strains in the device group was significantly higher than that in the colonization group (P<0.0001), while that of agr-1 in the device group was significantly lower than that in the nondevice group (P=0.039) and that in the colonization group (P=0.001). Also, a significant difference was observed between the nondevice group and the colonization group (P=0.047 and P=0.005, respectively). All strains were positive for icaA, icaD, fnbA, fnbB, and clfA genes. Only one strain (1.1%) in the colonization group was positive for the bbp gene. The prevalence of cna-positive strains was one strain (4.3%) in the device group, three strains (5.5%) in the nondevice group, and nine strains (10.0%) in the colonization group. There were no significant differences in the prevalence of the cna gene among the three groups.

View this table:

Prevalence of agr types no. (%) in each group

Device groupNondevice groupColonization group
agr-13 (13.0)20 (36.4)49 (54.4)
agr-218 (78.3)32 (58.2)31 (34.4)
agr-32 (8.7)3 (5.5)9 (10.0)
NT001 (1.1)
Total23 (100)55 (100)90 (100)
  • NT, nontypable.

  • P=0.039 vs. the nondevice group, P=0.001 vs. the colonization group,

  • P<0.0001 vs. the colonization group,

  • P=0.047 vs. the colonization group,

  • P=0.005 vs. the colonization group.

A comparison of the biofilm index among the three groups is shown in Fig. 1. The average biofilm index of the device group (mean ± SD, 0.517 ± 0.517; range 0.066–2.345) was significantly higher than that of the nondevice group (0.328 ± 0.453; range 0.033–2.748, P=0.039). There was no significant difference between the device group and the colonization group (0.411 ± 0.558; range 0.024–3.123, P=0.104).

Figure 1

Comparison of the biofilm index among three groups (device group, n=23; nondevice group, n=55; colonization group, n=90). Box plots show the median and the 10th, 25th, 75th, and 90th percentiles. *P=0.039.

Fifty-four strains of SCCmec type II/spa t002 estimated New York/Japan clone and 31 strains of SCCmec type IV/spa t1767 were isolated in this study. The biofilm index of SCCmec type II/spa t002 strains (mean ± SD, 0.549 ± 0.637; range 0.073–2.889) was significantly higher than that of SCCmec type IV/spa t1767 (0.257 ± 0.304; range 0.033–1.597, P=0.0091) and other sporadic clones (0.353 ± 0.481; range 0.024–3.123, P=0.0088).

We compared biofilm formation ability among agr types (Fig. 2). The biofilm index of the agr-2 strains (0.523 ± 0.572; range 0.073–2.889) was significantly higher than those of the agr-1 strains (0.260 ± 0.418; range 0.033–3.123, P<0.0001) and agr-3 strains (0.379 ± 0.557; range 0.024–2.083, P=0.045).

Figure 2

Comparison of biofilm formation among agr types: agr-1, n=72; agr-2, n=81; agr-3, n=14. Box plots show the median and the 10th, 25th, 75th, and 90th percentiles. *P<0.0001; **P=0.045.

We defined a strong biofilm former as a strain with a biofilm index higher than 0.50, which was the 95th percentile in the agr-1 group. The prevalence of strong biofilm formers in the device group (10/23, 43.5%) was significantly higher than those in the nondevice group (7/55, 12.7%; P=0.003) and the colonization group (18/90, 20.0%; P=0.020) (Table 3). The strong biofilm formers in the device group were all agr-2 strains.

View this table:
Table 3

Prevalence of strong biofilm formers no. (%) in each group

Device groupNondevice groupColonization group
BI>0.5010 (43.5)7 (12.7)18 (20.0)
BI<0.5013 (56.5)48 (87.3)72 (80.0)
Total23 (100)55 (100)90 (100)
  • BI, biofilm index OD595 nm.

  • P=0.003 vs. the nondevice group, P=0.020 vs. the colonization group.

In the device group, there were 13 cases of acute infection that developed in the first 3 months (two haematogenous infections and 11 nonhaematogenous infections), and 10 cases of chronic infection that developed 3 months after primary surgery. Two isolates of acute haematogenous infection were weak biofilm formers. The incidence of the strong biofilm formers in the chronic infection group was higher than that in the acute infection group, although this difference was not significant (6/10, 60.0% vs. 4/13, 30.8%, P=0.222).


In the present study, we found that MRSA strains from patients with device-related orthopaedic infection were more likely to be strong biofilm formers than those from patients with device-non-related infection and asymptomatic nasal carriers. Previous studies showed that biofilm-forming capacity is associated with the pathogenesis of catheter-related urinary tract infection (Hatt & Rather, 2008) and catheter-related blood stream infection (Passerini et al., 1992). Biofilm formation is the most important factor in the development of chronic infection and allows for immune evasion as well as resistance to antimicrobial agents (Francolini & Donelli, 2010), so that the only method of successful treatment is the removal of the injured tissues or devices (Brady et al., 2008). To our knowledge, this is the first study of biofilm formation in an epidemiological analysis of orthopaedic device infection caused by MRSA. Our results suggest that the level of biofilm formation is associated with the pathogenesis of orthopaedic device infection.

The agr system senses the density of bacteria in a ‘QS system’ via cell-to-cell communication with the release of the autoinducer protein, and plays a pivotal role in the inhibition of biofilm formation and expression of many virulence factors (O'Gara, 2007). Previous studies reported that most agr-2 strains exhibited strong biofilm formation ability and that the incidence of MRSA strains carrying agr-2 in the infection group was significantly higher than that in a carrier group in a university hospital (Manago et al., 2006) and that agr-2 strains are defective in the agr system and produce large amounts of biofilm with early transcription of icaA (Cafiso et al., 2007). In the present study, agr-2 strains were significantly stronger biofilm formers than agr-1 and agr-3 strains and the prevalence of agr-2 strains was higher in the device-related SSI group than in the nondevice group and the nasal colonization group. Many agr-2 strains possess the ability to produce strong biofilm and are likely to cause orthopaedic device infection.

Biomaterials are rapidly coated with a conditioning film composed primarily of host-derived extracellular matrix proteins, some of which can act as receptors for the initial attachment of bacteria. Staphylococcal binding to extracellular matrix proteins is mediated by MSCRAMMs. Following the initial attachment to an implanted device surface, biofilm formation involves a cellular accumulation process (O'Gara, 2007). The icaADBC genes encode enzymes that are involved in the production of polysaccharide intercellular adhesion (PIA), which mediates the intercellular adhesion of bacteria and the accumulation of multilayer biofilms (Götz, 2002). The icaA gene product is a transmembrane protein with homology to N-acetyl glucosaminyltransferases, requiring the icaD gene product for optimal activity (O'Gara, 2007). In this study, all strains derived both from the infection group and from the colonization group were positive for the icaAD, fnbAB, and clfA genes. These findings suggest that all MRSA strains isolated from infection and colonization sites have the ability to attach and initiate biofilm formation. Campoccia (2009) reported that the bbp gene and the cna gene were associated with orthopaedic implant infections caused by S. aureus. However, there were no bbp-positive strains and few cna-positive strains in the device-related infection group. In this study, the large genetic diversity in MRSA isolates was observed because we needed to collect MRSA strains over a long period (about 11 years) for this study, and many bacteria involved in biomaterial infections belong to endogenous bacteria or the commensal community of the skin (Rimondini et al., 2005).

A previous study reported that PIA and protein act cooperatively in the biofilm formation of methicillin-susceptible S. aureus (MSSA) and MRSA strains isolated from prosthetic joint infection (Rohde et al., 2007) and that biofilm development in MRSA is ica independent and involves a protein adhesion regulated by SarA and Agr, whereas SarA-regulated PIA/PNAG plays a more important role in MSSA biofilm development (O'Neill et al., 2007). In fact, we observed fibronectin-binding activity using a fibronectin-coated microtitre plate assay in clinical isolates of our study. The biofilm index using a fibronectin-coated plate of an agr-2 group was significantly higher than that of an agr-1 (data not shown).

Currently, to evaluate the ability of differential bacterial strains for the production of biofilm, a number of different in vitro assays are used, such as a microtitre plate, a tube test, radiolabelling, Congo red agar plate test, confocal laser scanning microscopy, etc. (Esteban et al., 2010). Among them, the crystal violet microtitre plate assay is one of the most popular and useful methods. However, the OD value is a reflection of the number of bacteria and is not an indicator of slime production (Harraghy et al., 2006). Furthermore, this method is semi-quantitative, measures the adherence to abiotic not biotic surface, and does not allow analysis of the internal structure of biofilms. Further analysis using other assays will be needed to evaluate biofilm formation ability in our clinical MRSA isolates.

Because of the strong biofilm formation by the agr-2 strain, agr genotyping may aid the determination of the possibility of biofilm-related orthopaedic device infection. Wilcox (2003) reported that the use of perioperative mupirocin is effective in preventing MRSA orthopaedic SSIs. The SHEA guidelines suggest that the use of an active surveillance culture and contact precaution including cohort isolation of all colonized patients is effective in preventing the nosocomial transmission of MRSA (Muto et al., 2003). agr typing and biofilm quantification may be helpful in determining when decolonization and cohort isolation are required to prevent device-related orthopaedic SSIs.

In conclusion, the level of MRSA biofilm formation was related to orthopaedic device infection and agr-2 MRSA strains may be more likely to cause orthopaedic device infection because of their strong biofilm formation ability.


We thank the staff of Kagoshima University Hospital Clinical Laboratory for technical assistance in collecting MRSA strains. All authors report no conflicts of interest relevant to this article.


  • Editor: Jacques Schrenzel


View Abstract