Comparison of Biofilm Formation between Major Clonal Lineages of Methicillin Resistant Staphylococcus aureus

Objectives Epidemic methicillin-resistant S. aureus (MRSA) clones cause infections in both hospital and community settings. As a biofilm phenotype further facilitates evasion of the host immune system and antibiotics, we compared the biofilm-forming capacities of various MRSA clones. Methods Seventy-six MRSA classified into 13 clones (USA300, EMRSA-15, Hungarian/Brazilian etc.), and isolated from infections or from carriers were studied for biofilm formation under static and dynamic conditions. Static biofilms in microtitre plates were quantified colorimetrically. Dynamic biofilms (Bioflux 200, Fluxion, USA) were studied by confocal laser-scanning and time-lapse microscopy, and the total volume occupied by live/dead bacteria quantified by Volocity 5.4.1 (Improvision, UK). Results MRSA harbouring SCCmec IV produced significantly more biomass under static conditions than SCCmec I–III (P = 0.003), and those harbouring SCCmec II significantly less than those harbouring SCCmec I or III (P<0.001). In the dynamic model, SCCmec I–III harbouring MRSA were significantly better biofilm formers than SCCmec IV (P = 0.036). Only 16 strains successfully formed biofilms under both conditions, of which 13 harboured SCCmec IV and included all tested USA300 strains (n = 3). However, USA300 demonstrated remarkably lower percentages of cell-occupied space (6.6%) compared to the other clones (EMRSA-15 = 19.0%) under dynamic conditions. Time-lapse microscopy of dynamic biofilms demonstrated that USA300 formed long viscoelastic tethers that stretched far from the point of attachment, while EMRSA-15 consisted of micro-colonies attached densely to the surface. Conclusions MRSA harbouring SCCmec types IV and I–III demonstrate distinct biofilm forming capacities, possibly owing to their adaptation to the community and hospital settings, respectively. USA300 demonstrated abundant biofilm formation under both conditions, which probably confers a competitive advantage, contributing to its remarkable success as a pathogen.


Introduction
Staphylococcus aureus is a commensal that colonizes the nasal cavity of approximately 30% of the human population. However, in susceptible people or those predisposed with risk factors, this organism easily leads to infections, ranging from relatively mild skin and soft tissue infections to life-threatening infections such as sepsis, endocarditis and necrotising pneumonia. Although S. aureus infections used to be easily treatable with penicillins, its adaptability resulted in the rise of more successful clones and the discovery of penicillin resistance in the 1940s [1], and of methicillin-resistant S. aureus (MRSA) in the 1960s [2]. Methicillin resistance was due to the uptake of the staphylococcal cassette chromosome mec (SCCmec) encoding an alternative penicillin binding protein PBP2a with reduced affinity for betalactams [3]. The vast use of antibiotics in the hospital setting resulted in the acquisition of additional, often horizontally transmitted, genetic determinants that led to highly resistant hospital-acquired (HA) MRSA, sometimes harbouring more than 20 antibiotic resistance determinants [4]. Although these strains were highly fit for the hospital setting, they often had slower growth rates and a lower overall fitness compared to their community-associated counterparts. The community-acquired (CA-) MRSA arising in the 1990s [5] were therefore associated with the smaller SCCmec type IV-V elements, which resulted in the loss of the antibiotic resistance markers associated with the larger SCCmec type I-III elements while simultaneously leading to strains that were more fit to compete outside the hospital setting [6][7]. Recently however, the borders between CA-and HA-MRSA have started to fade with CA-MRSA invading the hospital environment.
MRSA are not only subdivided into HA-and CA-MRSA, but also into different clones, often originally being described based on geographical site of first isolation [8]. Clones can be genetically characterized based on the sequence type (ST) and clonal complex (CC) by multi locus sequence typing (MLST), by SCCmec typing or by the presence of certain virulence genes such as those coding for Panton Valentine leukocidin (pvl) or the Arginine Catabolic Mobile Element (ACME; arc and opp clusters) [9].
The ability of S. aureus to form biofilms is an important characteristic which complicates infections due to MRSA, especially those associated with foreign materials such as catheters and implants [10]. In biofilms, the surface-associated bacteria are encased in an extracellular matrix and are thus far more resistant to antibiotics, often resulting in the need to remove the infected device in order to be able to treat the infection [11]. Biofilm formation is therefore an important survival strategy employed by bacteria that facilitates a prolonged persistence of infection and increases human morbidity and mortality.
The aim of the present study was to compare the biofilmforming capabilities of MRSA clones that were most commonly associated with infections worldwide. We hypothesized that biofilm formation might be more prevalent in the more virulent and non-multi-drug resistant (CA) MRSA harbouring SCCmec type IV, aiding their rapid spread and the ability to cause severe infections in the community.

Strain characteristics
Seventy-six well-characterized MRSA belonging to the 13 most important epidemic clones, based on MLST and SCCmec typing results, were selected for this study (Table 1). Of the 76 strains, 48 had been isolated from clinical infections (wounds/abscesses, n = 27; blood, n = 5; urinary tract, n = 3; and respiratory tract, n = 13), 1 from a stool sample, 15 from nares of healthy carriers and the rest were of unknown origin. MLST and SCCmec typing were performed as described previously [12][13]. Genetic relatedness was further confirmed by PFGE typing using Sma I [14]. Presence of pvl and the ACME I element (arcA and opp3AB gene clusters) were investigated by real-time PCR and PCR-sequencing, respectively [13,15]. Antibiotic susceptibility testing by disc diffusion was carried out utilizing a panel of antibiotics: chloramphenicol, erythromycin, clindamycin, trimethoprim-sulfamethoxazole, rifampicin, gentamicin, ciprofloxacin, tetracycline and cefoxitin (Table S1).

Biofilm formation in a static model
All strains were tested in triplicate on a 96-well microtitre plate with a polystyrene peg lid as described previously [16], with few modifications. Inoculum was prepared by diluting log phase cultures grown in brain heart infusion (BHI) broth enriched with 1% glucose to OD 600 0.04 after which 200 ml was dispensed in each microtitre plate well (n = 9/strain). Subsequently, microtitre plates were closed with peg lids and incubated at 37uC in air for 72 hours with medium replacement every 24 hours. A known biofilm-forming strain (S. aureus ATCC25923) and a strain consistently negative for biofilm formation (S. aureus 5374) were utilized as controls during each assay. After growth, the pegs were washed 3 times by submersion in PBS at room temperature, fixed with methanol and stained with Hucker's crystal violet (2%) for 15 minutes. Afterwards, pegs were rinsed with water and the stain eluted by submersion for 30 minutes in 200 ml of 33% acetic acid. OD values were measured at 492 nm (Multiskan FC photometer, Thermo Scientific), normalized to the blank and compared to 72 h-old biofilms of the positive control strain. Strains were divided into three groups based on previously published criteria: those showing on average (of three OD values) ,25% biomass of the positive control strain were designated as weak biofilm formers, 25%-75% biomass as moderate biofilm formers, and $ 75% biomass of the positive control as strong biofilm formers [16].

Biofilm formation, viability determination and quantification in a shear flow model
All strains were tested on a medium-throughput continuous flow system (BioFlux 200, Fluxion Biosciences, USA). The BioFlux 200 consists of a 48-well plate with a microchannel connection between 24 paired sets of input and output wells [17].
Electropneumatic regulators allow precise and individual control over the flow (and shear) rate applied in each channel. The same MRSA strain inoculum utilized in the static peg plate assay was also used to inoculate the BioFlux system by reversing the flow and pushing the bacteria into the microchannel from the output well, thus avoiding contamination of the input well. Bacteria were allowed to attach for one hour, followed by 17 hours of incubation at 37uC in BHI with a pressure of 0.5 dyne/cm 2 . Biomass in the microfluidic channels was visualised by addition of 2.5 ml of live/ dead BacLight viability stain (Invitrogen, Life Technologies) to each of the input wells, followed by flowing the stain for 10 minutes at 0.5 dyne/m2 in the dark. Prior to staining, the microchannels were flushed with 100 ml of 0.85% NaCl at 0.5 dyne/cm2 for 10 minutes to wash away unattached or loosely bound bacteria. Biofilms were visualized with an inverted fluorescence microscope (EVOSfl, AMG). Confocal images were obtained with a microlens-enhanced dual spinning disk confocal microscope system (UltraVIEW; PerkinElmer, Seer Green, UK) equipped with an argon-krypton laser source with two excitation lines (488 and 568) for excitation of FITC-and Cy3-like labels, respectively. Images were processed using Volocity 5.4.1 (Improvision, PerkinElmer, Seer Green, UK) and Imaris 7.3.1 software (Bitplane, Switzerland). Time-lapse microscopy of fully grown biofilms (17 hours of growth) under shear flow was performed using a high-definition digital camera (HDR-TG7VE, Sony, Japan). To quantify the total volume occupied by live or dead cells, biofilm stacks obtained by confocal laser microscopy (CLSM) were exported from Volocity followed by a 10-iterations blind deconvolution protocol on Sharp Stack (Image ProPlus AMS 7, Media Cubernetics). After deconvolution, contrasts for green and red cells were equally enhanced in all stacks after which volumes of red and green cells were measured using Volocity, applying automatic threshold settings.

Statistical analysis
Statistical analysis of biomass formation in the static assay was performed using the R Project software (version 2.11.1). The Table 1. Overview of MRSA clone characteristics, site of isolation, and biofilm formation in static and shear flow assays. influence of clonal complex, SCCmec type, site of infection, and presence of pvl and ACME (arcA and opp3AB) on biofilm formation and the comparison of the space occupied by the cells within the dynamic biofilms were analysed using one-way ANOVA, Welch's t-test for two-group comparisons using Bonferroni's post-hoc adjustments where indicated or a Pearson's x 2 test.
Within the different clones of the SCCmec IV group, all SCCmec IV strains belonging to CC22 (EMRSA-15, n = 7) and CC8 (USA300, n = 3; USA500, n = 8) exhibited strong biofilm formation with ODs ranging from 0.036-0.183 and 0.032-0.392, respectively ( and biofilm formation. In our study, the presence of ACME was uniquely associated with USA300 and thus high amounts of biomass (P = 0.011).

Biofilm formation in a shear flow model
Of the 76 strains tested here, 39 (51%) strains successfully formed biofilms under shear flow. Successful biofilm formation was defined as the ability of cells to adhere to the surface in 2 of 2 or 3 independent experiments and was considered negative when no adherent cells could be shown in 2 of 2 or in 3 of 4 such experiments. Of the strains that were found positive for biofilm formation, 22 (66%) harboured SCCmec types I-III (n = 33) and 17 (40%) SCCmec type IV (n = 43) (P = 0.036, x 2 ). All USA600 (n = 3, CC45-SCCmec II) and USA300 (n = 3, CC8-SCCmec IV), but none of the South-West Pacific (n = 5, CC30-SCCmec IV) and Paediatric (n = 5, CC5-SCCmec IV) clones formed biofilms in the shear flow assay (Figure 2). The remaining 33 strains positive for biofilm formation under shear flow belonged to the following clones: Hungarian/Brazilian (8/12), EMRSA-15 (5/7), Berlin (5/ 11), EMRSA-16 (3/4), New York/Japan (3/4), Southern Germany (3/6), USA500 (3/8), Iberian (2/4) and European (1/4). On confocal microscopy, differences in the biofilms between various clones were evident, with the Iberian and USA600 clones forming thicker, more confluent and densely packed biofilms in comparison to those formed by USA500, USA300 and the Hungarian/ Brazilian clones that showed gaps making the biofilms less confluent (USA500) or formed a less densely packed biomass (USA300 and Hungarian/Brazilian) (Figure 3). Since live/dead staining only stains bacterial cells and not the matrix, differences in cell densities within the biofilm could be captured through 3dimensional quantification of the space occupied by microorganisms. Both the USA300 and Hungarian/Brazilian clones demonstrated remarkably lower percentages of cell-occupied space (6.6% and 4.0%, respectively) compared to the other clones (EMRSA-15 = 19.0%, USA500 = 25.2%, USA600 = 15.7% and Iberian = 9.4%) (Figure 3 and 4). Time-lapse microscopy demonstrated the strength of an MRSA biofilm under shear flow conditions (movie S1 and movie S2). Even when flow was 46 higher than normal growth conditions (2.0 dyne/cm 2 ), the USA300 biofilm remained attached to the surface. Long elastic matrix strands kept the cells together. The USA300 (CC8-MRSA-IV-PVL-ACME) strain showed individual cells attached more to each other than to the surface resulting in the formation of long streamers/tethers, while attachment to the surface was more evident for the EMRSA-15 (CC22-MRSA-IV) strain (movie S1 and movie S2).

MRSA showing consistent biofilm formation in static and shear flow assays
Sixteen of the 76 MRSA tested here consistently formed biofilms in both the static and shear flow assays; 19 strains were strong biofilm formers in the static assay and negative in the shear flow assay; 23 strains were weak or moderate biofilm formers in the static assay but positive in the flow assay; and 18 strains were negative in both assays. Of the 16 consistent biofilm formers in both the static and shear flow assays, 13 harboured SCCmec type IV and included all USA300 (n = 3), 5/7 E-MRSA15, 3/8 USA500, and 2/11 Berlin strains. Three MRSA harbouring SCCmec type III were also consistent biofilm formers in both assays and all belonged to the CC8-Hungarian/Brazilian clone.

Influence of SCCmec on biofilm formation
Strains carrying SCCmec type IV were shown to have a significantly higher probability to form biofilms in the static assay compared to those with SCCmec type I-III, with strains carrying SCCmec type II being the least capable of biofilm formation. Conversely, in a shear flow assay strains harbouring SCCmec I-III showed higher potential for biofilm formation compared to strains carrying SCCmec IV.
Genetic differences in the SCCmec element harboured by HAand CA-MRSA strains might underlie differential biofilm formation in static or dynamic conditions and are possibly related to the distinct requirements needed to accommodate community and hospital settings (e.g. skin and soft tissue infections versus catheter related or blood stream infections). For instance, phenol-soluble modulins (PSMs) are toxins that regulate virulence, and biofilm formation and structuring in S. aureus [18]. While most are chromosomally-encoded, psm-mec is present on the SCCmec II and III elements but not on SCCmec IV. Psm-mec has been shown to regulate the core genome-encoded PSMs, resulting in decreased virulence [19], and thicker more compact biofilms [20]. Interestingly, and also validating our in vitro results, MRSA harbouring SCCmec II and an intact psm-mec were found to be more frequently associated with catheter-related blood stream infections than strains with mutated psm-mec that showed PSM levels similar to SCCmec IV harbouring MRSA, and were also more likely to cause pneumonia and abscess formation [21]. Furthermore, differences in biofilm formation in the two models might be due to the different adherent surfaces utilized in the static (polystyrene) and dynamic (glass) systems. Surface-dependent attachment is especially more pronounced in in vitro setups where no human matrix proteins such as fibrinogen or fibronectin, which normally serve as an anchor for attachment on medical implants, are present [22].

SCCmec IV harbouring EMRSA-15 and USA300 are prolific biofilm formers
In comparison to the other globally predominant MRSA clonal lineages, both USA300 and EMRSA-15 formed abundant biofilms. USA300 biofilms were, however, more structured compared to EMRSA-15 that were too dense to allow dye penetration into the deeper biofilm layers. Time-lapse microscopy showed that USA300 formed long viscoelastic streamers or tethers that stretched far from the point of attachment under shear flow, occasionally releasing parts of the biofilm. EMRSA-15 biofilms on the other hand consisted more of micro-colonies densely attached to the surface. USA300 demonstrated to be an excellent biofilm former in both tested biofilm models, showing differences in biofilm characteristics compared to its USA500 progenitor and making its capacity to form biofilms an important characteristic to its transmission and survival, possibly contributing to its highly epidemic behaviour worldwide. The 31-kb genomic island ACME uniquely harboured by USA300 probably underlies its higher propensity to form biofilms. Recent data shows that the ACME-Arc system drives excessive production of host polyamines, which are uniquely toxic to S. aureus. However, the ACME also encodes a polyamineresistance enzyme, SpeG, which combats excess host polyamines and enhances adherence to fibrinogen/fibronectin, resistance to antibiotic and keratinocyte-mediated killing and importantly, an enhanced biofilm forming capacity [23].
Thus, both the SCCmecIV and ACME elements seem to be major contributors to the extraordinary pathogenetic success of the USA300 clone by enhancing its virulence and biofilm forming capacities.

Supporting Information
Table S1 Antibiotic susceptibility data of MRSA clones.

(MTS)
Movie S2 Time-lapse imaging of a USA300 biofilm under increasing shear flow stress. Shear stress was modulated as follows: t = 0 s, pressure = 0 dyne/cm 2 , t = 5 s, pressure = 0.1 dyne/cm 2 , t = 10 s, pressure = 0.2 dyne/cm 2 , t = 15 s, pressure = 0.3 dyne/cm 2 , and so on till pressure = 2.0 dyne/cm 2 in the final 10 s of imaging. When the pressure is switched off at the end of the movie the long stretching tethers are most evident. (MTS)