Functional and comparative genomic characterization of biofilm formation in Staphylococcus aureus

Abstract

Biofilms are structured communities of bacterial cells enclosed in a self-produced extracellular matrix. In the pathogen Staphylococcus aureus, this can enhance resistance to antibiotics and immune responses, contributing significantly to chronic infections associated with medical devices. The underlying mechanisms include the production of polysaccharide intercellular adhesin (PIA), encoded by the icaADBC operon, and surface proteins that mediate adhesion. However, it has been challenging to translate in vitro understanding to explain the molecular mechanisms governing biofilm formation in vivo. Here we combined functional and comparative genomics approaches to investigate genetic factors influencing biofilm formation in isolates belonging to the clinically important ST-8 clonal complex (CC8). Phenotypic and genomic screening of a closely related strain cohort (MRSA USA300 isolates) revealed considerable variability in biofilm formation. Genome-wide association studies (GWAS) identified several genes and polymorphisms linked to biofilm development. These included known biofilm genes and compensatory mutations that restored wild-type biofilm levels in hyper-biofilm forming mucoid isolates. Finally, contextualizing CC8 genomes within diverse S. aureus populations revealed the natural occurrence of biofilm-associated genomic variation as well as evidence for the conservation of the ica loci in CC8. This offers insight into the mechanisms and microevolutionary events that give rise to clinically relevant staphylococcal infections.

Keywords: Biofilm mechanisms; Genome-wide association study; Pathogenicity; Staphylococcus aureus.

Fig. 1

Biofilm formation of 134 clinical MRSA CC8 S. aureus strains. Biofilm development under static conditions was examined using the crystal violet assay. A) Bloodstream isolates (n = 36), B) skin and soft tissue infection (SSTI) isolates (n = 60) and C) isolates derived from asymptomatic carriage (n = 38) were examined and percentage biofilm formation compared to a control MRSA isolate, TW20 (designated 100 %). Three technical and three biological repeats were used with columns and error bars displaying the mean and standard deviation. D) Comparison of biofilm formation from three clinical sources where each dot represents the average biofilm formation of each isolate. A one-way ANOVA with Tukey multiple comparison test showed no significant difference in biofilm formation between groups. (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)

Fig. 2

Maximum likelihood tree displaying biofilm formation and functional validation of biofilm associated genes derived from GWAS. A) A SNP-based maximum likelihood tree mapping biofilm formation and infection source across the collection of isolates. Branch colours describe different disease origins; red = bacteraemia; green = carriage; blue = SSTI. External bars represent normalised biofilm formation; USFL039 biofilm formation is displayed not to scale. B) Transposon insertion mutants were assessed for biofilm activity compared to wildtype JE2. Three biological repeats were included. Statistical differences were calculated by one-way ANOVA. ∗p < 0.05, ∗∗p < 0.01. (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)

Fig. 3

Phenotypic characterisation of strain USFL039. A) Strains LAC (control) and USFL039 were grown on TSA, congo red agar (CRA) or in liquid TSB culture. USFL039 growth on CRA and in liquid culture (red circle) displays biofilm and aggregative phenotypes respectively. B) Growth kinetics of LAC or USFL039 grown in either TSB, TSB-G or TSB-NaCl. Graphs displays average OD_600nm measurements of two biological replicates. C) Biofilm forming capacity of strains LAC, TW20 or USFL039 grown for either 8 or 24 h in TSB-G (grey bar) or TSB-NaCl (white bar). Each dot represents individual biological repeat with error bars displaying the SD. Statistical differences were calculated by two-way ANOVA. ∗p < 0.05. (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)

Fig. 4

Major biofilm component of strain USFL039. Mature biofilms of USFL039, LAC and TW20 grown in TSB-G or TSB-NaCl were incubated with disrupting agents (white bars) A) DNase I (15 mg/mL) B) proteinase K (0.5 mg/mL) C) sodium metaperiodate (20 mM) for 2 h at 37 °C, after which remaining biofilms were stained. Controls (grey bars) were mature biofilms grown under identical conditions that received no challenge. Significance was determined through a 2-way ANOVA, where ∗∗ indicates a p value less than 0.01 and ∗∗∗∗ indicates a p value < 0.0001.

Fig. 5

PIA/PNAG straining and biofilm visualisation by confocal microscopy. USFL039, LAC and TW20 biofilms were grown for 12 h in either TSB-G or TSB-NaCl. Live-dead staining was performed using SYTO9 and propidium iodide (PI); polysaccharide staining was performed using fluorescently labelled wheat germ agglutinin (WGA). The wells were excited at 500 nm for both SYTO9 and WGA and 575 nm for PI on a Zeiss Cell discoverer 7. Green indicates live cells stained by SYTO9, red indicates dead cells stained by PI, and cyan indicates polysaccharides produced by the ica operon stained by WGA. (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)

Fig. 6

Mucoid phenotype is rapidly lost in vitro suggesting a fitness cost. A) Frequency of mucoid to non-mucoid phenotype reversion was calculated every 24 h following screening on CRA agar. B) Biofilm capacity of USFL039 m culture #2 after each passage day grown in TSB-G or TSB-NaCl. C) Co-culture competition experiments between USFL039nm and USFL039 m; mucoid phenotype was examined on CRA agar at time points 2, 4, 8 and 24 h and presented in CFU/ml. Each dot represents individual biological repeat with error bars displaying the SD. Statistical differences were calculated by two-way ANOVA. ∗∗p < 0.01; ∗∗p < 0.0001. D) Co-culture experiments presented in C are displayed as bars representing the value of each phenotype as percent of total population with fitness values indicated on top of the graph.

Fig. 7

The ica locus is present in most S. aureus isolates, with greatest variation in regulatory intergenic region. A) Grapetree of 1991 S. aureus isolates, from carriage (white) and infection (grey). B) Number of unique alleles for icaR, icaADBC and the icaR_A intergenic region for the USA300 isolates used in this study (grey) and the more diverse UK dataset (striped). Frequencies are normalised to the number of isolates and sequence length. C) Number of unique alleles (as above) for each major clonal complex.

Fig. 8

Population analysis of the ica locus suggests loss of mucoid phenotype in natural S. aureus isolates. A) Binary heatmap of mutations at each alignment position of the ica locus for the USA300 dataset. B)ica locus mutations with frequencies over 10 % for the UK dataset, including insertions (red), deletions (blue), synonymous (black) and non-synonymous (gold) mutations and mutations in non-coding sequences (cyan). C) Binary heatmap of mutations <10 % at each alignment position of the ica locus for the UK dataset isolate datasets. D) Diagrammatic representation of the ica operon. (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)