Capsule expression is not enhanced in the mucoid compared to the non-mucoid phenotype

S. aureus produces capsule polysaccharides (CP) also termed micro-capsules, which can be differentiated by rabbit serum antibodies [21]. These (CPs) represent an important virulence factor of S. aureus due to protection against phagocytosis. Most clinical S. aureus isolates belong to CP-type 5 or 8 [21]. Therefore, we hypothesized that hyper-expression of CPs could be responsible for the mucoid phenotype. We determined CP expression by a semi-quantitative colony immunoblot assay [22] for selected isolates of patient 8. Mucoid strains varied in their CP expression between weak and strong and did not differ significantly compared to the non-mucoid strains (S1 Table). Quantification of capsule production by ELISA inhibition confirmed the results of the colony immunoblot assays (S1 Fig), indicating that CP expression did not explain the mucoid phenotype of the isolates.

Mucoid isolates are hyper-producers of PIA-associated biofilms

Another possible explanation for the mucoid phenotype could be overproduction of PIA/PNAG leading to enhanced biofilm formation. Assessment of biofilm formation using a conventional microtiter plate assay showed that all mucoid isolates displayed enhanced biofilm production, whereas almost no biofilm was formed by the non-mucoid strains as exemplified with strains 8.1 and 8.2 (Fig 2, Table 2). To investigate the nature of the biofilm, we treated the biofilm with sodium metaperiodate, which breaks down PIA/PNAG-dependent biofilms, and by proteinase K or DNase, which disrupts protein-or DNA-dependent biofilms, respectively. The biofilm of the mucoid isolate was dissolved by sodium metaperiodate, but not by proteinase K or DNase, indicating that PIA/PNAG is functionally involved in the biofilm formation (Fig 2).

Open in a separate window

Fig 2

Biofilm formation of non-mucoid and mucoid phenotypes.

A static biofilm assay was performed after overnight culture of the non-mucoid strain 8.1 and the isogenic mucoid S. aureus strain 8.2 (Table 2) of patient 8 in BHI containing 0.25% glucose, incubation for 24 h at 37°C and staining with 1% crystal violet. Biofilms were solubilized in ethanol-acetone and the absorbance was determined at OD_655nm with a microtiter plate reader. Each experiment was performed at least three times, and in each experiment the samples were prepared in eight replicate wells. To determine the nature of biofilm, detachment of the biofilm was measured after treatment of the wells, which have been cultured as described above, with sodium-metaperiodate, DNase or proteinase K before measuring the absorbance. All experiments were repeated three times. Results are shown as means and standard deviation. **** p-value ≤ 0.0001.

Enhanced biofilm structures in mucoid isolates due to PIA/PNAG

To further characterize biofilm formation of mucoid isolates in more detail, confocal laser scanning microscopy was performed. Staining of adherent bacteria using the LIVE/DEAD BacLight kit detected no stable biofilm for non-mucoid isolates. Here, only occasionally cell clusters with different density and thickness were observed (Fig 3A). Intriguingly, bacteria were only loosely attached to each other, resulting in wobbling of bacterial consortia on mechanical stress, and the average biofilm thickness was low (<10 μM at 24 h). In contrast, the mucoid strain produced a stable biofilm, which revealed the typical mushroom-shaped multicellular structure (Fig 3B). Cells in this biofilm were tightly attached and a biofilm progression over time was present. While the surface was not entirely covered after 12 h, full coverage was reached after 24 h combined with a mushroom-shaped multicellular structure. The average height of the biofilm was 23 μM after 24 h. Furthermore, we observed an accumulation of dead cells inside the mushroom structures (Fig 3B). Staining of the cultures with antibodies against PIA revealed densely packed bacteria embedded in extracellular PIA/PNAG (Fig 4B) for the mucoid but not for the non-mucoid isolate (Fig 4A).

Open in a separate window

Fig 3

Confocal microscopy of biofilms of the non-mucoid (strain 8.1) and mucoid S. aureus (strain 8.2) phenotypes.

Adherent bacteria were stained using the LIVE/DEAD BacLight kit. A). The green color indicates live and the red color dead cells. Only scant amounts of biofilm were detected for non-mucoid isolates. The average biofilm thickness was low (<10 μM at 24h). B) The mucoid strain produced a stable biofilm, which revealed the typical mushroom-shaped multicellular structure. The average height of the biofilm produced by the mucoid strain was 23 μM after 24 h.

Open in a separate window

Fig 4

Confocal microscopy of biofilms stained with PIA/PNAG-specific antibodies.

Non-mucoid strain (A) and mucoid (B) S. aureus strains (strains 8.1 and 8.2) were grown overnight in TSB. After washing, adherent bacteria were stained using SYTO 9 (green). PIA/PNAG was detected using ALEXA-568 labeled wheat germ agglutinin (red), which is a lectin binding to the N-acetylglucosaminyl backbone of PIA/PNAG. Zoom-in shows PIA/PNAG-embedded bacteria in the mucoid strain (B), while hardly any PIA/PNAG is detected with the non-mucoid isolate. White bar = 10 μm.

Mucoid S. aureus isolates are protected against phagocytosis

Since CF lung disease is characterized by enhanced inflammation with an accumulation of polymorphonuclear neutrophils (PMNs) [23], we assessed uptake and killing by fresh human PMNs of representative non-mucoid and mucoid isolates from patient 8 (strains 8.1 and 8.2, Table 2) in an in vitro assay. In contrast to the non-mucoid strain, which was readily killed by PMNs, the mucoid isolate resisted opsonophagocytic killing (Fig 5A). However, if antibodies against PIA/PNAG were added to the assay, robust phagocytic killing was achieved (Fig 5B). In contrast, CP8 antibodies, which interfere with the microcapsule of this isolate, which has been shown to belong to CP8, were poorly opsonic (S2 Fig) indicating that capsule formation had no impact on protection against neutrophilic killing. Therefore, these data revealed that opsonophagocytic killing of the mucoid S. aureus isolate was mediated by opsonic PIA/PNAG antibodies.

Open in a separate window

Fig 5

Opsonophagocytic killing of non-mucoid (strain 8.1) and mucoid S. aureus (strain 8.2) isolates by human neutrophils (PMNs) from healthy volunteers.

The assay contained 2.5 x 10⁶ neutrophils, 5 x 10⁵ CFU S. aureus, 1% guinea pig serum (as a complement source), in a total volume of 500 μl MEM-BSA. Some assays included antibodies against PIA/PNAG raised in rabbits (1% PIA/PNAG). Control samples contained PMNs and S. aureus or antiserum and S. aureus alone with no PMNs. After 2 h of incubation, samples were vortexed and diluted in sterile deionized H₂O to lyse the PMNs to release internalized bacteria. The lysate was plated on TSA plates to enumerate bacterial counts. Percent killing was calculated by the reduction in CFU/ml after 2 h compared with that at time zero. A) Significantly more non-mucoid compared to mucoid bacteria were killed in the presence of neutrophils (97% versus 28%, p<0.0001). In control wells, where neutrophils were absent, only 10% of bacteria died. B) By adding antibodies against PIA/PNAG to wells with mucoid isolates and neutrophils, 93% of bacteria were killed compared to only 26% killing of mucoid isolates in the presence of neutrophils but lacking the specific antibodies (p = 0.0002). All experiments were repeated three times. Results are shown as means and standard deviation. *** p-value ≤ 0.001; **** p-value ≤ 0.0001.

Mucoid isolates carry a 5 bp deletion in the icaR-icaA intergenic region of the ica operon

Because mucoidy of the S. aureus isolates correlated with increased PIA/PNAG-mediated biofilm formation, we further characterized the ica operon of relevant S. aureus strains. As shown earlier by Jefferson et al. [20], a spontaneous laboratory mutant of S. aureus MN8 harbored a 5 bp deletion in the icaR-icaA intergenic region of the ica operon, resulting in constitutive expression of PIA/PNAG. Therefore, at first we sequenced the upstream icaA region of our mucoid isolates. All mucoid isolates carried exactly the same 5 bp deletion at the same position in this intergenic region. Interestingly, in three patients mucoid isolates co-colonized with non-mucoid S. aureus for periods up to 126 months (Table 1), which is exemplified more extensively for patient 8 in S1 Table.

Non-mucoid strains with the 5 bp deletion carried compensatory mutations

Surprisingly, from patients 7 and 8, who were persistently infected by mucoid S. aureus isolates for extended periods, non-mucoid isolates (n = 12), which carried the 5 bp deletion, were isolated. The non-mucoid isolates (Table 2) did not produce rough colonies on CRA or biofilm and therefore did not seem to produce high amounts of PIA/PNAG related biofilm anymore. To determine whether compensatory mutations in these late non-mucoid isolates with 5 bp deletions had occurred, we sequenced the entire ica operon and identified various mutations in icaA (n = 7), icaD (n = 3) and icaC (n = 2, Table 2, Fig 6).

Open in a separate window

Fig 6

Mutations of non-mucoid S. aureus isolates with 5 bp deletions.

During persistence, non-mucoid S. aureus isolates with 5 bp deletions and compensatory mutations in icaA, D and C were identified. The ica-sequences of 12 non-mucoid isolates with 5 bp deletions of patients 7 (A) and 8 (B) are illustrated. The numbering of bases and amino acids (aa) follows the sequences retrieved from S. aureus strain ATCC35556 (SA113). Changes of bases, amino acids and deletions are indicated in red color. Synonymous changes are indicated in green. S. aureus strains are identified by the number of the patient's visit and the date of isolation. For patient 7, the same mutation was identified for isolates from visits 14, 15 and 16 and for patient 8 for isolates from visits 27, 43, 49 and 51, respectively. Only the mutation for the first identified isolate is shown.

Patient 7 carried S. aureus isolates with two different mutations in icaA that were detected in two different isolates recovered from one visit, the same mutation in icaD in three independent isolates during a period of 3 months, and mutations in icaC at one visit (Table 2). Patient 8 carried S. aureus with mutations in icaC identified at one visit and in icaA at five visits. Two of these isolates recovered in April 2007 (strain 8.4) and May 2008 (strain 8.7) carried different mutations in icaA. Six years later, three isolates with the same mutation in icaA as the isolate recovered in 2008 were cultured from March until November 2014 (strains 8.8, 8.9 and 8.10). Since there was a break of 6 years without the culture of isolates with such mutations, it is unlikely that these isolates persisted during this period without detection by culture. In line with this suggestion our whole genome sequencing data revealed that strain 8.7 is more distant from all other sequential isolates making it unlikely that this strain is the founding strain for strain 8.8 (Fig 7). Therefore, the same mutation in icaA emerged again in mucoid isolates. In summary, 10 of 12 isolates with 5 bp deletions but without PIA/PNAG hyper-expression showed mutations in icaA (n = 7) and icaD (n = 3).

Open in a separate window

Fig 7

UPGMA-tree based on up to 1,861 cgMLST allelic profiles of S. aureus isolates of patients 7 and 8 (Table 2).

The tree on the top shows the relationship of the six isolates of patient 7 including the reference strain N315, the tree on the bottom shows the relationship of the six isolates of patient 8 including the reference strain 55/2053. The scale, i. e. distances were given in absolute number of differing alleles. The strains marked with an asterisk exhibit a mucoid phenotype. For more information about isolates see Table 2.

Whole genome sequencing (WGS) of non-mucoid and mucoid isolates of patients 7 and 8

To confirm that the mucoid isolates of patients 7 and 8 evolved from the non-mucoid S. aureus clone, we performed WGS of the first mucoid and nearest non-mucoid S. aureus isolates of these patients and also of the isolates with 5 bp deletion and compensatory mutations (Table 2). WGS and subsequent cluster analysis based on cgMLST (core genome MLST) allelic profiles of the each six strains of patient 7 and patient 8 exhibit a close relationship of all strains within each patient irrespective of the phenotype (Fig 7). Additional ANI (average nucleotide identity) calculation further corroborated this close relationship as all isolates of patient 7 had ANI values of 100% in pairwise comparisons. Similar ANI results were also observed for patient 8, where the ANI values were ≥ 99.99%.

Complementation with vectors expressing wild-type ica genes confers the mucoid phenotype

To assess the role of the icaA, icaD and icaC mutations in non-mucoid S. aureus isolates with the 5 bp deletion, we used the xylose-inducible staphylococcal expression vectors pTXicaBC and pTXicaAD. Complementation of the isolates with the respective vectors restored the mucoid phenotype and allowed strong biofilm formation (Fig 8), corroborating that the compensatory mutations in icaA, icaD or icaC caused abrogation of biofilm and the non-mucoid phenotype.

Open in a separate window

Fig 8

Biofilm formation of isolates with 5 bp deletion, compensatory mutations, which were complemented with the wild-type ica genes.

Biofilm formation of the strains with compensatory mutations in icaA, icaD or icaC was abrogated. A) Isolates of patient 7. B) Isolates of patient 8. Transformation with a vector expressing the appropriate wild-type ica gene conferred biofilm formation.

PIA/PNAG overexpression results in fitness costs in some but not all mucoid strains

Excessive production of PIA/PNAG might confer a fitness loss to mucoid S. aureus compared to non-mucoid isolates. To investigate this, we performed competition experiments between two non-mucoid and mucoid S. aureus strain pairs, which were isolated from patients 7 and 8. Whereas for patient 7, the non-mucoid strain outcompeted the mucoid isolate during co-culture (Fig 9A), for the strain pair of patient 8 no significant differences in fitness could be observed (Fig 9C). These results indicate that the emergence of a mucoid phenotype in vivo can, but is not necessarily associated with a fitness loss.

Open in a separate window

Fig 9

Competition and starvation experiments comparing non-mucoid and mucoid S. aureus isolates from patients 7 and 8.

A and C) Competition growth experiments between the non-mucoid and mucoid strain pairs of patients 7 and 8 with fitness values indicated on top of the graph. During co-culture, the non-mucoid strain of patient 7 out-competed the mucoid strain, while for the strains of patient 8 no significant difference could be observed. B and D) The same strain pairs of patients 7 and 8 were compared concerning survival under nutrient limited conditions for 48h. For both patients, the mucoid strains survived significantly better during starvation compared to the non-mucoid strains. Data were generated by performing three biological with two technical replicates. Graphs show the average of three biological replicates with error bars indicating the standard deviation. Statistical analysis of the generated data was performed using an unpaired two-tailed t-test. * p-value ≤ 0.05; ** p-value ≤ 0.01; *** p-value ≤ 0.001.

Strains and growth conditions

Several mucoid and non-mucoid S. aureus strains isolated from the respiratory tract of CF patients were characterized in this study (Tables ​(Tables1,1, ​,22 and S1). The biofilm-negative S. carnosus TM300 and the biofilm-positive S. epidermidis RP62A (ATCC 35984) were used as controls. Staphylococci were grown in tryptic soy broth (TSB) (Becton, Dickinson and Company, Sparks, MD, USA), brain heart infusion (BHI) (Merck, Darmstadt, Germany), or on tryptic soy agar (TSA) (Becton, Dickinson and Company, Sparks, MD, USA) or Columbia blood agar plates and incubated for 24 h at 37°C. Congo red agar plates prepared with Columbia blood (CRA) were used to confirm the mucoid phenotype of S. aureus isolates [24]. When appropriate, erythromycin (10 μg/ml) was added to the medium. Induction was achieved by the addition of 0.5% xylose.

Genotyping

Cap-typing was performed by multiplex PCR [29]; spa-typing and MLST were done as described before [30,31].

Whole genome sequencing

To determine the clonal relationship of the strains of two patients with long-term persistence, we performed whole genome sequencing (WGS) of 12 S. aureus isolates of these patients as described [32] using the Illumina Nextera XT library preparation for a 250 bp paired-end sequencing run on a Miseq system (Illumina Inc., San Diego, CA, USA). Quality trimming, de novo assembly and subsequent core genome MLST (cgMLST) were performed as described recently [32]. For tree building using the Unweighted Pair Group Method with Arithmetic mean (UPGMA) method within the Ridom SeqSphere⁺ software (Ridom GmbH, Münster, Germany), the allelic profiles of the up to 1,861 cgMLST targets [33] were used applying the parameter “pairwise ignoring missing values”.

Moreover, we determined the average nucleotide identity (ANI) based on the de novo assembled contigs. Here, we used the ANI calculator (http://enve-omics.ce.gatech.edu/ani/ [34], which estimates the ANI using reciprocal best hits (two-way ANI) between two genomic datasets using default parameters. We assume that closely related isolates should exhibit > 99.9% ANI.

For comparison, genome sequences of S. aureus reference strains N315 (GenBank accession no. NC_002745; MLST ST5) and 55/2053 (NC_022113; MLST ST30) were used for patient 7 and patient 8, respectively, which reflect the most related S. aureus lineages to the patients’ isolates as determined by MLST.

All raw reads generated were submitted to the European Nucleotide Archive (http://www.ebi.ac.uk/ena/) under the study accession number PRJEB15647.

Survival in minimal medium

To assess if there is a survival fitness of non-mucoid or mucoid isolates during starvation, bacteria were cultured in minimal medium according to Brooks et al. [24]. Briefly, bacteria were grown in 5 ml tryptic soy broth containing 1% glucose (TSBG) in 50 ml conical cell reactor tubes (Cellstar, No 227 245) at 160 rpm and 37°C. After 24h, cultures of mucoid isolates were supplemented with 5 μg/ml of Dispersin B (Kane Biotech) and cultivated for 30 min at 37°C to break up biofilm clusters. Bacteria were collected by centrifugation (4500 rpm, 7 min, RT) and pellets were resuspended in 5 ml MOPS minimal media (Teknova) lacking glucose to achieve a concentration of 10⁸ cells/ml. Bacteria were cultivated in 50 ml conical cell reactor tubes for up to 48h at 37°C and 160 rpm. After 0h, 4h, 8h, 12h, 24h and 48h cultures were vortexed, serially diluted and plated on Columbia blood agar incubated for 24h at 37°C to determine CFU/ml counts. Survival of isolates in minimal medium was calculated by comparing CFU/ml counts of respective time-points to the 0h CFU/ml counts (presented in %). Transformants harboring the xylose inducible plasmids pTXicaBC or pTXicaAD were cultivated as described above with the exception that the growth medium was supplemented with 10 μg/ml tetracycline and 0.5% (v/v) xylose. Statistical analysis was performed using an unpaired two-tailed t-test. * p-value ≤ 0,05; ** p-value ≤ 0,01; *** p-value ≤ 0,001.

Competition experiments

Competition experiments were performed according to Brooks et al., 2014. Bacteria were grown in 5 ml TSBG in 12 ml tubes at 160 rpm and 37°C. Growth medium of mucoid isolates was supplemented with 5 μg/ml of Dispersin B (Kane Biotech) to prevent strong clustering due to excessive biofilm formation. After 24 h, bacteria were collected by centrifugation (4500 rpm, 7 min, RT) and pellets were resuspended in 5 ml TSBG. Cultures were diluted to a concentration of 10⁸ cells/ml and mucoid and non-mucoid isolates were mixed 1:1 in 50 ml conical cell reactor tubes (Cellstar). The medium was supplemented with 5 μg/ml of Dispersin B (Kane Biotech) to prevent biofilm formation and cultures were incubated for up to 24h at 37°C and 160 rpm. After 0h, 4h, 8h and 24h culture aliquots were vortexed, serially diluted and plated on CRA plates for CFU counting. CRA plates were incubated for 2–3 days at 37°C. Changes in population composition were calculated by comparing CFU/ml counts of mucoid or non-mucoid isolates to the CFU/ml counts of the whole population at the respective time-point (presented in %). Fitness of non-mucoid and mucoid isolates was calculated using the following function developed by Sander et al. M_t = ln [(n_t/m_t)/(n_t-1/m_t-1)^1/gen] where n_t and m_t are the amount of non-mucoid and mucoid cells at a given time-point t, while n_t-1 and m_t-1 represent the amount of non-mucoid and mucoid cells at the preceding time-point [35]. For all isolates t₀ was chosen to be the preceding time-point and fitness was calculated for the given time-points 4h, 8h and 24h according to http://textbookofbacteriology.net/growth_3.html. The function fit_t = 1+M_t was used to calculate the relative bacterial fitness. The fitness value is bigger than 1, if the non-mucoid isolate is fitter than the mucoid isolate. If there is no difference in fitness between the isolates, the value equals 1. If it is lower than 1, the non-mucoid isolate has a reduced fitness compared to the mucoid isolate. Statistical analysis of the generated data was performed using an unpaired two-tailed t-test. * p-value ≤ 0,05; ** p-value ≤ 0,01; *** p-value ≤ 0,001.

Biofilm assay

A static biofilm assay followed by crystal violet staining was modified from a previous report [36]. Briefly, an overnight culture of S. aureus was diluted 200-fold with BHI containing 0.25% glucose, of which 200 μl were added to the wells of a 96-well polystyrene microtiter plate (Greiner Bio-One, Frickenhausen, Germany) and incubated for 24 h at 37°C. To determine the amount of biofilm produced by each clinical S. aureus isolate, all plates were washed three times with PBS and the on the bottom adhering biofilms were stained with 1% crystal violet for 15 min. Following three further washing steps with PBS, biofilms were solubilized in 100 μl ethanol-acetone (80:20). The absorbance was determined at OD_655nm with a microtiter plate reader (Bio-Rad, Hercules, CA, USA). In parallel experiments, the nature of the formed biofilms was analyzed. Therefore, microtiter plates that were incubated for 24 h at 37°C were treated either with 100 μl sodium-metaperiodate in water (40 mM; AppliChem, Darmstadt, Germany), DNase I in 150 mM NaCl/1mM CaCl₂(100 μg/ml; Roche, Mannheim, Germany) or proteinase K in 10 mM Tris-HCl (pH 7.5) (100 μg/ml; MP Biomedicals, Santa Ana, California, USA) for 3 h at 37°C. After three washing steps with PBS, biofilms were stained with crystal violet, washed again with PBS and solubilized in ethanol-acetone as described above. To confirm accuracy and reproducibility, each isolate was investigated in three biological replicates, always in eight wells per microtiter plate. S. epidermidis RP62A, which is known to form a strong biofilm with PIA as the major component [37] served as positive control, and the biofilm-negative and PIA-icaADBC-negative S. carnosus TM300 served as a negative control [38].

Confocal laser scanning microscopy

S. aureus isolates were grown overnight in 300 μl TSB under static conditions in six-well cell culture plates (μ-Dish, Ibidi, Munich, Germany). In some experiments Alexa-568 labelled Wheat germ agglutinin or antibodies against PIA were added to the growth medium. Non-adherent cells were removed by washing with PBS and bacteria were stained using live staining (Live/dead staining, Molecular Probes). Confocal image acquisition was performed on a Zeiss Axiovert 200M inverted microscope equipped with a Yokogawa CSU-22 confocal head and a Hamamatsu C9100-02 EM-CCD camera. Images were taken with a Zeiss Plan Apochromat 63x/1.4 Ph3 Oil objective. Improvision Velocity software was used for image acquisition and quantification.

Quantification of bacterial polysaccharide production

Capsule was semi-quantified by colony immunoblots as described before [22,39], and bacterial polysaccharides (CP and PIA/PNAG) were quantified by enzyme-linked immunosorbent (ELISA) inhibition assays [40,41]. Briefly, 96-well plates were coated overnight at 4°C with purified PNAG (1 μg/ml) or with CP5 or CP8 (4 μg/ml) coupled to poly-L-lysine by the cyanuric acid chloride method. The microtiter plate was washed and blocked with 0.05% skim milk. S. aureus strains were harvested, washed in phosphate buffer, and then trypsinized (1 mg trypsin/ml of 0.1 M phosphate buffer, pH 8) for 60 min at 37°C to remove protein A. After washing, the bacterial suspensions were serially diluted, and the bacterial concentrations were verified by plating on tryptic soy agar plates. Polyclonal polysaccharide-specific antiserum was diluted and incubated overnight at 4°C with serial dilutions of the bacteria or purified polysaccharide (standard curve ranging from 1 μg/ml to 1 ng/ml). Samples were centrifuged, and the absorbed serum samples (supernatants) were added to the coated microtiter plates. Following a 2-h incubation with absorbed or unabsorbed serum samples, the plates were washed with PBS/Tween, and alkaline phosphatase-conjugated protein A/G (Thermo Scientific; 1:3000) was added to each well. After a 2-h incubation at ambient temperature, the plate was washed, and the substrate p-nitrophenyl phosphate was added. When the wells containing unabsorbed serum samples reached an OD_{405 nm} of ~2.0, the plate was read on a Bio-TEK Power Wave HT ELISA reader. The concentration of each sample (CFU/ml) that resulted in 50% inhibition of antibody binding (IC₅₀) was determined, and the polysaccharide content of the sample was calculated from the standard curve.

Opsonophagocytic killing assays

The opsonophagocytic killing (OPK) activity of human polymorphonuclear neutrophils (PMNs) was performed and analyzed as described [42].

Transformation with expression vectors

The staphylococcal expression vectors pTXicaBC and pTXicaAD [43] were used to complement the S. aureus strains with 5 bp deletions and non-mucoid phenotype. The plasmids were transformed into the cells by electroporation using standard procedures as described [44]. Plasmid purification was performed following the PrepEase Quick MiniSpin Plasmid kit protocol (Affymetrix).

We thank G. Pier for providing antibodies against PIA/PNAG for the ELISA inhibition assay.