Airway environment drives the selection of quorum sensing mutants and promote Staphylococcus aureus chronic lifestyle

Abstract

Staphylococcus aureus is a predominant cause of chronic lung infections. While the airway environment is rich in highly sialylated mucins, the interaction of S. aureus with sialic acid is poorly characterized. Using S. aureus USA300 as well as clinical isolates, we demonstrate that quorum-sensing dysfunction, a hallmark of S. aureus adaptation, correlates with a greater ability to consume free sialic acid, providing a growth advantage in an air-liquid interface model and in vivo. Furthermore, RNA-seq experiment reveals that free sialic acid triggers transcriptional reprogramming promoting S. aureus chronic lifestyle. To support the clinical relevance of our results, we show the co-occurrence of S. aureus, sialidase-producing microbiota and free sialic acid in the airway of patients with cystic fibrosis. Our findings suggest a dual role for sialic acid in S. aureus airway infection, triggering virulence reprogramming and driving S. aureus adaptive strategies through the selection of quorum-sensing dysfunctional strains.

Fig. 1. Agr system dysfunction correlates with sialic acid hyperconsumption and is selected in an environment containing free sialic acid.

A Schematic depicting the CF lung environment encountered by S. aureus lung-adapted variants with co-occurrence of sialylated mucins and sialidase-producing bacterial species. B Schematic depiction of nan locus and sialic acid metabolism pathway. C, D nan locus expression from bacteria grown in CLM supplemented with 1.3 mM Neu5Ac (CLM+Neu5Ac) was quantified by qRT-PCR relative to that of housekeeping gene gyrB (OD600nm ~ 0.6) (n = 3 samples/group). C USA300-LAC (USA300) strain and ΔagrC derivative; D Early (CF1_early) and late (CF1_late exhibiting an agrC inactivation) isogenic isolates recovered 3 years apart from respiratory samples of cystic fibrosis patient CF1. E, F Kinetics of consumption of sialic acids obtained from chemical desialylation of bovine submaxillary mucins (BSM) (n = 3 samples/group). Bacteria were grown in CLM supplemented with 0.2% chemically desialylated BSM. Free sialic acids were derivatized with DMB (1,2-diamino-4,5-methylene dioxybenzene) before quantification by HPLC. E USA300-LAC (USA300) strain and ΔagrC derivative; F Early (CF1_early) and late (CF1_late) isogenic isolates. G, H Agr dysfunctional variant monitoring in a competition assay. USA300-LAC (USA300) strain was completed for 5 days with ΔagrC derivative starting at a ratio of 1:1 in the presence of G 1.3 mM glucose or H 1.3 mM Neu5Ac. Proportions of nonhemolytic and hemolytic subpopulations were determined based on colony phenotype on sheep blood agar (n = 3 samples/group). The phenotype of USA300 and ΔagrC colonies is hemolytic and non-hemolytic, respectively. Error bars indicate mean with SEM. Statistically significant differences were calculated by one-way ANOVA with Bonferroni’s multiple comparisons test and p value was indicated. Source data are provided as a Source Data file.

Fig. 2. Co-occurrence of free sialic acid, sialidase-producing bacteria, and S. aureus in CF sputa.

A Free sialic acid concentration has been assessed by colorimetric assay in sputa from five children with CF and positive S. aureus culture detection by standard microbiology testing. B Taxonomic characterization at the family level of microbiota in the sputa of 5 patients with CF and positive S. aureus culture detection showing the concomitant presence of Staphylococcaceae, Streptococcaceae, and Prevotellaceae families by 16 S sequencing. C The sialidase activity of three isolates obtained in culture (Streptococcus infantis, Schaalia odontolytica and Prevotella melaninogenica from sputa of Patient A, Patient C, and Patient H, respectively) was evaluated by quantifying the amount of salic acid released from purified CF patient mucins. The USA300-LAC strain was used as a negative control. Source data are provided as a Source Data file.

Fig. 3. Free sialic acid provides a growth advantage to Agr dysfunctional variants in vitro and in vivo.

A Free sialic acid in mucus produced by Calu-3 epithelial cells grown at the air-liquid interface (ALI) was measured 24 h after addition at the apical side of Dulbecco’s phosphate-buffered saline (DPBS), sialidase or S. mitis. A statistically significant difference was calculated by two-sided unpaired t test and p value was indicated (n = 4 samples/group). B Schematic depiction of infection in the ALI culture model. C In Calu-3 ALI model, left panel: 5 μL sialidase (0.5 U/mL DPBS/0.1% BSA) or 5 μL DPBS/0.1% BSA was added 24 hours before infection. Right: 5 mL Neu5Ac (0.4 g/L) or 5 mL H₂O was added 2 h before infection with S. aureus. ALI mucus of wells was apically infected with USA300, ΔagrC, or early and late isolates from patient CF1. Mucus was collected 16 h post infection and plated for CFU counting. Data are represented as fold increases over wells infected with the same strains without sialidase/sialic acid addition (n = 2 wells/groups). Each condition was performed with two biological replicates and two technical replicates. Statistically significant differences were calculated by one-way ANOVA with Bonferroni’s multiple comparisons test and p value was indicated. D Mice were inoculated intranasally with bacterial suspension (2 × 10⁸ CFUs) of S. aureus USA300 or ΔagrC, in PBS (−), or free sialic acid (+), as indicated on the bottom of each panel (n = 5 mice/group). Twenty-four hours post infection, the bacterial CFUs in the lungs were counted. Statistically significant differences were calculated by a two-sided unpaired t test and p value was indicated. Error bars indicate mean with SEM. Source data are provided as a Source Data file.

Fig. 4. Sialic acid triggers a transcriptomic reprogramming favoring chronicity and iron acquisition in S. aureus.

A RNAseq was performed to compare gene expression of the USA300-LAC strain grown in CLM with supplementation with glucose or Neu5Ac (1.3 mM). Numbers of genes with increased (red) or decreased (blue) expression levels between glucose and Neu5Ac growth conditions considering statistically significant genes with a Log2(FoldChange) ≥ |2| and FDR-adjusted p-value ≤ 0.01. Categorization is based on the Clusters of Orthologous Groups (COG) functional classification. Ion, inorganic ion transport and metabolism; Sec. metabol., secondary metabolites biosynthesis, transport and catabolism; Prot. modif., posttranslational modification, protein turnover, chaperones. B Differentially expressed genes were analyzed via the Search Tool for the Retrieval of Interacting Genes/Proteins (STRING) to identify protein–protein interaction networks. C Heatmap showing three biological replicates for selected genes with significant differences between glucose and Neu5Ac growth conditions. D USA300 and ΔagrC were cultivated in CLM supplemented with glucose or Neu5Ac. The sbnA expression (OD600 nm ~ 0.6) were quantified by qRT-PCR with reference to gyrB (n = 3 samples/group). Fold change was calculated relative to the glucose condition. (E) The iron acquisition ability was measured in the presence of glucose or Neu5Ac. Strains were cultured in CLM supplemented with glucose or Neu5Ac, culture supernatants were collected at post-exponential phase and the siderophore units were measured by chrome azurol S (CAS) assay (n = 3 samples/group). F List of differentially expressed small RNAs between glucose and Neu5Ac growth conditions considering statistically significant genes with a Log2(FoldChange) ≥ |2| and FDR-adjusted p-value ≤ 0.01. G, H USA300, ΔnanR deletion mutant, and complemented nanR strain (CP-nanR) were cultivated in CLM supplemented with glucose. isrR (G) and sbnA (H) expression (OD600nm ~ 0.6) were quantified by qRT-PCR with reference to gyrB (n = 3 samples/group). I The iron acquisition ability of USA300, ΔnanR deletion mutant, and complemented nanR strain (CP-nanR) was measured in the presence of glucose by CAS assay (n = 3 samples/group). J Schematic depicting putative pathways for NanR-dependent regulation of iron acquisition-related genes. Error bars indicate mean with SEM. Statistically significant differences were calculated by one-way ANOVA with Bonferroni’s multiple comparisons test and p value was indicated. Source data are provided as a Source Data file.