Thymidine starvation promotes c-di-AMP-dependent inflammation during pathogenic bacterial infection

Abstract

Antimicrobials can impact bacterial physiology and host immunity with negative treatment outcomes. Extensive exposure to antifolate antibiotics promotes thymidine-dependent Staphylococcus aureus small colony variants (TD-SCVs), commonly associated with worse clinical outcomes. We show that antibiotic-mediated disruption of thymidine synthesis promotes elevated levels of the bacterial second messenger cyclic di-AMP (c-di-AMP), consequently inducing host STING activation and inflammation. An initial antibiotic screen in Firmicutes revealed that c-di-AMP production was largely driven by antifolate antibiotics targeting dihydrofolate reductase (DHFR), which promotes folate regeneration required for thymidine biosynthesis. Additionally, TD-SCVs exhibited excessive c-di-AMP production and STING activation in a thymidine-dependent manner. Murine lung infection with TD-SCVs revealed STING-dependent elevation of proinflammatory cytokines, causing higher airway neutrophil infiltration and activation compared with normal-colony S. aureus and hemin-dependent SCVs. Collectively, our results suggest that thymidine metabolism disruption in Firmicutes leads to elevated c-di-AMP-mediated STING-dependent inflammation, with potential impacts on antibiotic usage and infection outcomes.

Keywords: STING; Staphylococcus aureus; anti-folate antibiotics; lung; neutrophil; small colony variants; thymidine.

Figure 1.. Antibiotic treatment promotes c-di-AMP production by bacteria.

(A) Heatmap of relative IFN-β production in pBMDMs infected with L. monocytogenes incubated in BIOLOG antibiotic plates as measured by ISRE-luciferase bioassays. The log₂ ratios of IFN-β relative to untreated L. monocytogenes are shown. Black labels indicate different antibiotics, green labels indicate the antibiotic class based on mechanism of action, ND indicates no bacteria were recovered after treatments and IFN-β was not determined. (B) Schematic of thymidine metabolism in bacteria. ThyA: Thymidylate synthase; NupC: Nucleoside permease; DHFR: Dihydrofolate reductase; DHPS: Dihydropteroate synthase. (C) IFN-β production of pBMDMs infected with L. monocytogenes treated with indicated antibiotics relative to the untreated control at 6 hpi as measured by ISRE-luciferase bioassays. Logarithmic phase L. monocytogenes were incubated with either 15 μg/ml penicillin G, 200 μg/ml 2,4-Diamino-6,7-diisopropylpteridine, 1,200 μg/ml sulfamethoxazole, 250 μg/ml trimethoprim, or 50 μg/ml SXT for 6 hrs before macrophage phagocytosis. (D) Schematic of STING activation. Eukaryotic 2′3′-cGAMP, synthesized by cGAS in response to host or pathogen-derived DNA, and bacterial cyclic dinucleotides can both activate STING, resulting in IFN-μ induction. (E) Ifhb1 mRNA expression in WT or Sting ^−/− pBMDMs infected with L. monocytogenes grown in BHI broth with or without 0.25 μg/ml of SXT supplementation at 4 hpi. (F) Quantification of intracellular c-di-AMP production of L. monocytogenes treated with antibiotics as indicated in panel (C) using CDA-Luc assay. For all panels, mean values of biological replicates are plotted, and error bars indicate ±SD. P values were calculated using Two-way ANOVA analysis. Asterisks indicate statistical significance (*, P < 0.05; **, P < 0.01; ***, P < 0.001, ****, P < 0.0001), and ns indicates no significant difference.

Figure 2.. L. monocytogenes ThyA inhibition or mutation leads to higher c-di-AMP production.

(A) Intracellular c-di-AMP concentration of L. monocytogenes grown in BHI broth with indicated SXT and thymidine supplementation quantified by CDA-Luc assay. (B) Ifnb1 mRNA induction in pBMDMs infected with L. monocytogenes grown in BHI broth as indicated in (A). (C) Ifnb1 mRNA induction by WT or ΔthyA L. monocytogenes strains in WT, Sting^−/− or cGas^−/− pBMDMs at 4 hpi. (D-H), Ifnb1 mRNA induction in pBMDMs infected with either E. faecalis (D), S. aureus Newman (E), S. aureus JE2 (F), S. typhimurium (G) or F. novicida (H) grown in the indicated conditions. For all panels, mean values of biological replicates are plotted, and error bars indicate ±SD. P values were calculated using Two-way ANOVA analysis. Asterisks indicate statistical significance (*, P < 0.05, **, P < 0.01; ****, P < 0.0001), and ns indicates no significant difference.

Figure 3.. ΔthyA activates STING-dependent type I IFN production through c-di-AMP in primary macrophages.

(A) LC-MS/MS quantification of c-di-AMP levels of clinical TD-SCV isolates (0115-30 and 0138-11) relative to clonally-related normal-colony (NC) isolates (0115-33 and 0138-5) grown on chocolate agar overnight. (B) Ifnb1 expression in pBMDMs infected with clinical isolates at 4 hpi relative to uninfected control. (C) Auxotrophy test for thymidine and hemin dependence of ΔthyA or ΔhemB plated on Mueller-Hinton agar. Representative image shows enhanced growth, indicated by white arrows, around disks impregnated with thymidine or hemin. (D) LC-MS/MS quantification of c-di-AMP levels of ΔthyA and ΔhemB relative to NC. Cultures of ΔthyA and ΔhemB were supplemented with 5 μg/ml of thymidine and 1 μg/ml hemin, respectively. (E) Ifnb1 mRNA induction by NC, ΔthyA and ΔhemB Newman strains in WT, Sting^−/− and cGas^−/− pBMDMs at 4 hpi relative to uninfected controls. (F) IFN-β concentration in pBMDM supernatants measured by ISRE-luciferase bioassay at 6 hpi. (G) C-di-AMP levels of NC, ΔthyA and ΔthyA::thyA strains quantified by CDA-Luc assay. (H) Ifnb1, Il6, Cxcl10, and Ccl5 mRNA induction by NC, ΔthyA or ΔthyA::thyA Newman strain in WT or Sting^−/− pBMDMs relative to uninfected controls at 4 hpi. (I) Quantification of c-di-AMP levels of NC, ΔthyA, and pde A overexpression Newman (ΔthyA/pdeA) strains by CDA-Luc assay. (J) Ifnb1, Il6, Cxcl10, and Ccl5 mRNA induction in pBMDMs infected with NC, ΔthyA, and ΔthyA/pdeA strains relative to uninfected control. All infections were performed at an MOI of 20. For all panels, mean values of biological replicates are plotted, and error bars indicate ±SD. P values were calculated using Two-way ANOVA analysis. Asterisks indicate statistical significance (*, P < 0.05; **, P < 0.01; ***, P < 0.001, ****, P < 0.0001), and ns indicates no significant difference.

Figure 4.. Thymidine abundance modulates c-di-AMP levels and subsequent Ifnb1 induction of TD-SCV.

(A) Growth curve of NC (left) and ΔthyA (right) S. aureus Newman in BHI broth supplemented with indicated concentrations of thymidine. (B) Quantification of c-di-AMP levels of NC and ΔthyA grown in BHI broth supplemented with thymidine by CDA-Luc assay. (C) LC-MS/MS quantification of intracellular c-di-AMP in iBMDMs infected with NC, ΔthyA and ΔhemB Newman strains at 4 hpi. (D) Schematic of macrophage infections. WT iBMDMs were infected with NC or ΔthyA overnight cultures grown in BHI broth supplemented with the indicated thymidine concentrations for 4 hrs and the intracellular c-di-AMP levels and Ifnb1 expression in iBMDMs were determined by LC-MS/MS and qRT-PCR, respectively. (E-F), The intracellular c-di-AMP concentration (E) and Ifnb1 mRNA relative to uninfected control (F) as described in panel (D). All the infections were performed at the MOI of 20. For all panels, mean values of biological replicates are plotted, and error bars indicate ±SD. P values were calculated using One or Two-way ANOVA analysis. Asterisks indicate statistical significance (**, P < 0.01; ****. P < 0.0001), and ns indicates no significant difference.

Figure 5.. C-di-AMP promotes airway neutrophil infiltration.

(A-C) ELISA quantification of cytokines IL-6 (A), CXCL10 (B), and CCL5 (C) in BALF collected from the WT or Sting^−/− mice intranasally instilled with 15 μg Pam3CSK4 and 15 μg c-di-AMP or Pam3CSK4 alone. (D) Neutrophil numbers in BALF collected from the WT mice with intranasal instillation of 15 μg Pam3CSK4 and 15 μg c-di-AMP or Pam3CSK4 alone. (E) Ifnb1 expression in pBMDMs infected with dacA overexpression (NCdacA) and its control (NC/vector) Newman strain strains relative to uninfected control. Mean values of triplicates are plotted and error bars indicate ±SD. P values were calculated using Two-way ANOVA analysis. (F) CFU recovery from the right lung of WT mice intranasally infected with NC/dacA or NC/vector Newman strain at 8 hpi. (G-H), Neutrophil numbers (G) and active neutrophil elastase concentration (H) in BALF collected from mice infected in panel (F). Neutrophils are identified and enumerated by FACS. For all the mouse infections, biological replicates are plotted, and horizontal black bars are the median of the data. P values were calculated using Mann-Whitney analysis. Asterisks indicate statistical significance (*, P < 0.05; **, P < 0.01; ***, P < 0.001), and ns indicates no significant difference.

Figure 6.. TD-SCVs induce robust airway neutrophil infiltration and inflammation during both acute and recurrent infection.

(A) Ifnb1 expression in pBMDMs infected with 0115-30 or 0115-30::thyA strain relative to uninfected control. Mean values of triplicates are plotted and error bars indicate ±SD. P values were calculated using Two-way ANOVA analysis. (B) WT mice were intranasally infected with 0115-30 or 0115-30::thyA strains and CFU from the right lung was enumerated at 8 hpi. (C) ELISA quantification of IL-6, CXCL10 and CCL5 in BALF collected from mice infected in (B). (D-E), Neutrophil numbers (D) and active neutrophil elastase concentration (E) in BALF collected from mice infected in panel (B). (F) WT or Sting^−/− mice were intranasally infected with NC or ΔthyA strain with 5×10⁸ CFU/mouse and CFU from the right lung was enumerated at 8 hpi. (G) ELISA quantification of IL-6, CXCL10, and CCL5 in BALF collected from mice infected in (F). (H-I), Neutrophil numbers (H) and active neutrophil elastase concentration (I) in BALF collected from mice infected in panel (F). (J) Top panel indicates the repeat infection strategy, in which WT mice were intranasally instilled with NC, ΔthyA or ΔhemB strain with 2.5×10⁷ CFU/mouse or PBS every 4 days for a total of 5 infections. The lungs were harvested 8 hrs after the last infection. Bottom panel indicates the representative images of H&E staining of WT mouse lung sections. The arrows indicate the inflammatory cell infiltrate. Br=bronchiole, V=blood vessel, and A=alveoli. (K) Inflammation scoring for mice infected in (J). Inflammation scores were the summation of perivascular/peribronchiolar, intrabronchiolar, intraalveolar, interstitial inflammation and overall severity. Neutrophils are identified and enumerated by FACS. For all mouse infection panels, biological replicates are plotted. Horizontal black bars indicate the median of the data. P values were calculated using Mann-Whitney analysis. Asterisks indicate statistical significance (*, P < 0.05; **, P < 0.01; ****, P < 0.0001), and ns indicates no significant difference.