Abstract
Staphylococcus aureus is a pulmonary pathogen associated with substantial human morbidity and mortality. As vaccines targeting virulence determinants have failed to be protective in humans, other factors are likely involved in pathogenesis. Here we analysed transcriptomic responses of human clinical isolates of S. aureus from initial and chronic infections. We observed upregulated collagenase and proline transporter gene expression in chronic infection isolates. Metabolomics of bronchiolar lavage fluid and fibroblast infection, growth assays and analysis of bacterial mutant strains showed that airway fibroblasts produce collagen during S. aureus infection. Host-adapted bacteria upregulate collagenase, which degrades collagen and releases proline. S. aureus then imports proline, which fuels oxidative metabolism via the tricarboxylic acid cycle. Proline metabolism provides host-adapted S. aureus with a metabolic benefit enabling out-competition of non-adapted strains. These data suggest that clinical settings characterized by airway repair processes and fibrosis provide a milieu that promotes S. aureus adaptation and supports infection.
Similar content being viewed by others
Main
Among the numerous opportunistic bacteria that colonize the respiratory tract, Staphylococcus aureus is an especially successful pathogen, infecting both immunocompetent and immunocompromised hosts. While S. aureus can cause overwhelming necrotizing pneumonia1, it is most often associated with persistent pulmonary infections. As S. aureus is a major cause of healthcare-associated pneumonias, it is among the best-characterized human pathogens2. Its patterns of global epidemiology are well documented3, as are its abilities to regulate turnover of surface proteins4,5 and to secrete toxins6,7. Determinants of staphylococcal virulence have been extensively studied, and there is a wealth of data detailing its tissue-specific metabolic activity8. S. aureus has evolved to express an array of gene products that are specific to the human host, thwarting immune effectors, targeting surface receptors, blocking antibody and complement activity and activating excessive T-cell responses via superantigens9,10. Nonetheless, therapeutic and preventative strategies directed at several of these factors have been limited in their success to animal models of infection and have not been protective in human clinical trials11,12.
The pathogenesis of S. aureus pneumonia involves the aspiration of organisms from the upper respiratory tract and the establishment of a nidus of infection in the lower airways7. Airway damage, which may follow endotracheal intubation, viral infection or pulmonary diseases such as chronic obstructive pulmonary disease, idiopathic pulmonary fibrosis or cystic fibrosis (CF), is a major risk factor for S. aureus pulmonary infections13. Unfavourable outcomes are not limited to antibiotic-resistant strains. The bacterial mass, the recruited immune population and airway tissues contribute metabolites and oxidants at the site of infection14. Thus, a successful pathogen must adapt to the combined immune and metabolic pressures of the infected lung.
We postulated that S. aureus strains associated with persistent pulmonary infection acquire a substantial metabolic advantage driving their selection from the population of organisms that initially colonize the airways. We followed the transcriptional and metabolic changes in sequential clinical isolates from patients with pulmonary infection and determined that the successful S. aureus stains adapt to the lung by selectively consuming proline, generated by fibroblasts in the process of homeostatic collagen turnover and airway repair.
Results
Transcriptional changes in metabolic genes in vivo
To determine how S. aureus adapts to the human lung following initial colonization, we surveyed the transcriptional profiles of initial clinical isolates from paediatric patients and chronic isolates from adult patients with CF. We targeted genes expected to be important in S. aureus pathogenesis such as the α-toxin hla, critical in initiating pulmonary infection15; adsA, which generates the potent immunosuppressant adenosine and thwarts immune clearance16,17; putP, one of the proline transporters activated in response to the availability of glucose18 which may be limited in the airways19; and scpA, a collagenase and protease20. Compared with a laboratory strain Newman (WT), hla expression was not increased in any of the isolates. The initial colonizing strains (non-adapted isolates labelled CF paediatric, 0.5–7 years of age) tended to have reduced putP and significantly elevated adsA expression compared with a laboratory control strain, whereas isolates from long-standing infection (adapted isolates labelled CF adult), instead had significantly reduced adsA but increased putP and scpA, with respect to a WT control (Fig. 1a,b). The isolates in a 15 year longitudinal collection of S. aureus from one adult patient (Longitudinal CF Adult) similarly had low adsA and high putP and scpA expression compared with WT control (Fig. 1c). While all isolates could utilize proline as a carbon source (Fig. 1d–f and Extended Data Fig. 1a), it was the most avidly consumed substrate by the adult isolates. Proline boosted the growth (Fig. 1g–i) and metabolic activity (Fig. 1j–k) of the adult isolates (T2015) to a significantly greater extent than the non-adapted paediatric strains. The T2015 strain also showed an increase in extracellular acidification rate (ECAR) compared with the paediatric initial strain or the 2001 isolate (Fig. 1j), but changes in ECAR were much lower than oxygen consumption rates (OCR) (Fig. 1k). These findings were consistent with RNA-sequencing data we mined from a study measuring differential gene expression in a late (2008) S. aureus clinical isolate21 versus early (1995) isolate from one CF patient, which also reflected upregulated tricarboxylic acid (TCA) cycle activity, proline transport and adenosine triphosphate (ATP) synthesis over 13 years of infection (Fig. 1l).
a–c, In vitro RNA transcripts from S. aureus CF clinical isolates grown in LB, all with respect to WT strain Newman control. Representative isolates CF1, CF9, 2001 and 2015 were chosen for functional studies. d–f, Utilization of carbon sources shown as fold change with respect to WT strain Newman control. ‘CF paediatric’ initial isolates (a,d) and ‘CF adult’ adapted isolates (b,e), from distinct patients; longitudinal CF adult adapted isolates (c,f) from single adult patient. g–i, Growth in CDM ± 100 μM proline. P, proline. Please note that (g), (h) and (i) refer to individual strains as labelled on the figure. j,k, Metabolic activity presented as ECAR (j) and OCR (k) upon sequential injections of 100 μM proline comparing initial paediatric (CF1) and later (A2001, T2015) isolates. mpH, milli-pH. l, RNA-seq data19 of early (1995) and late (2008) CF isolates from one patient over 13 years shown as fold change with respect to 1995. Data presented as mean ± s.e.m. In a–k, n = 3. Significance determined by *P < 0.05; **P < 0.01; ***P < 0.001 ****P < 0.0001. In a–c, one-way ANOVA with selective t-Student comparison; in d–f, two-tailed t-Student with Kolmogorov–Smirnov test comparing OCR increase upon proline injection; in g–j, two-way ANOVA with Dunnett’s multiple comparisons.
The patterns of transcription of both scpA and putP in clinical isolates seemed to be a logical response to a collagen and likely proline-rich milieu in the airway. Less obvious was the decreased expression of adsA, given its role in adenosine synthesis expected to promote infection and immune evasion17,22. We hypothesized that adsA expression must also have an impact on staphylococcal metabolism. We performed bulk RNA-sequencing on S. aureus Newman strains grown in Luria–Bertani (LB) media: a WT and an adsA mutant (Δ). Our results confirmed that loss of adsA caused global changes in gene expression (Fig. 2a–c). The adsA mutants significantly upregulated TCA cycle, nucleotide, arginine and proline metabolic pathways and downregulated amino acid biosynthesis and virulence gene expression with respect to the WT strain as established by pathway analyses (Fig. 2b). Specifically, loss of adsA increased expression of arg, acnA and suc loci and decreased expression of genes involved in glycolysis and virulence, including ptsG, pckA, spl and luk (Fig. 2c). We then assessed the effect of added proline, which induced no differential gene expression in the WT strain (Fig. 2d) but was associated with global changes in the Δ transcriptome (Fig. 2e). In response to proline, we observed significant shifts in ΔadsA metabolic pathways with respect to WT control (Fig. 2f), such as upregulated extracellular polysaccharide biosynthesis and proline metabolism itself (Fig. 2g), inducing expression of putP, mur, asp, arg and scpA while downregulating genes involved in glycolysis/gluconeogenesis and virulence, including pckA, nuc, sar, spl, pde and pyk (Fig. 2h). These genotypic effects were coupled to functional changes in which ΔadsA phenocopied the adapted clinical isolates in the utilization of proline as a primary carbon source (Fig. 2i and Extended Data Fig. 1a). ATP production by ΔadsA was increased compared with WT strain in LB media and further increased by the addition of proline (Fig. 2j), enhancing the OCR of ΔadsA but not WT strains (Fig. 2k).
Bulk RNA sequencing of S. aureus WT and ΔadsA Newman grown in LB. a–c, Differential gene expression (a), pathway analysis (b) and fold change (c) in ΔadsA gene expression with respect to WT control. d,e, Differential gene expression of WT (d) and ΔadsA (e) in LB + 100 μM proline. f–h, Differential gene expression (f), pathway analysis (g) and fold changes (h) in ΔadsA genes with respect to WT control grown in LB + 100 μM proline. EPS, extracellular polysaccharide. i, Utilization of carbon sources shown as fold change with respect to WT control. j, ATP generated by WT control and ΔadsA strains grown in LB ± 100 μM proline. k, Carbon flux upon sequential injections of 100 μM proline in WT control and ΔadsA presented as ECAR and OCR. l,m, Growth of S. aureus WT control, ΔadsA and 2015 in ±30 μg ml−1 collagen (Col) (l) and growth of WT control and ΔadsA in ±100 μM proline (P) (m). Data graphics: in a and d–f, all dots represent individual genes; red dots, upregulated; blue dots, downregulated. In b and g, pink bars represent log(P); vertical dotted line is P < 0.05 threshold of significance; blue bars represent number of genes associated to the respective pathway. Data presented as mean ± s.e.m. In a–h, n = 2; in i–m, n = 3. Significance determined by *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. In a–h, Wald test generated P values (<0.05), and absolute log2 fold changes (>1f (greater than onefold)) were called differentially expressed; in i, j, l and m, two-way ANOVA with Dunnett’s multiple comparisons; in k, two-tailed t-Student with Kolmogorov–Smirnov test.
Preferential proline consumption by adapted isolates suggested selection within the lung microenvironment for strains that could use collagen as a carbon source, a protein largely composed of proline, hydroxyproline and glycine23,24. While collagen and proline both increased ΔadsA and clinical strain 2015 growth (Figs. 1e and 2l,m), hydroxyproline and glycine did not (Extended Data Fig. 1b,c), indicating specificity in S. aureus carbon selection and the potential involvement of host collagen as a source of proline for bacterial metabolism. Overall, the metabolic profile of ΔadsA suggested that this mutant reflects the major genotypic and metabolic changes associated with the adaptation of the clinical isolates to the lung: the preferential consumption of proline and associated optimization of energy generation and growth.
The airway metabolome provides proline for S. aureus
The in vivo selection of S. aureus strains that downregulate adsA expression and undergo a metabolic shift toward proline consumption and oxidative phosphorylation implies a survival advantage. We confirmed proline consumption to be higher in ΔadsA compared with both WT and complemented ΔadsA (c:ΔadsA) (Extended Data Fig. 1d,e), deemed to be a stable plasmid with or without antibiotics in both Newman and USA300 backgrounds (Extended Data Fig. 1f,g). Thus, we compared the recovery of WT, ΔadsA and c:ΔadsA in the Newman and USA300 backgrounds in acute (24 h) and sub-acute (72 h) mouse models of pneumonia. While the bacterial burdens were all comparable at 24 h post infection (Fig. 3a), there was significantly increased recovery of the ΔadsA in both the Newman and USA300 backgrounds at 72 h (Fig. 3b) compared with the WT or c:ΔadsA.
Pulmonary infection following intranasal inoculation with S. aureus WT Newman or USA300 LAC controls, ΔadsA or ΔadsA-complemented (c:adsA) strains. a,b, Lung bacterial burden at 24 h (a) and 72 h (b) post infection. c, Pulmonary infection following intranasal inoculation with S. aureus WT Newman or USA300 LAC controls, ΔadsA or ΔadsAΔputPΔproT strains. NS, not significant. d–f, Lung monocytes (d), neutrophils (e) and alveolar macrophages (f). g, Bacterial uptake and killing of murine BMDM assessed at 2, 4 and 6 h. h, Unbiased pathway analysis of BAL metabolites 72 h post-infection. Blue bars, P value; vertical dotted line is <0.05 threshold of significance; orange bars, Z-score or ratio of ΔadsA respective to WT; solid line is threshold of change. DWRG, downregulation; UPRG, upregulation. i, Schematic diagram of citrulline and proline metabolism in host; dotted arrows, translocation; solid arrows, conversion; in bold are critical metabolites/precursors to proline. j, BAL metabolites (selected from Supplementary Fig. 2a) related to citrulline metabolism, and targeted. k,l, Quantification of proline (k) and glucose (l) shown as fold change with respect to PBS. Data presented as mean ± s.e.m. from n = 3 in a; n = 4 in b–f and h–l; n = 2 in g. Significance determined by *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. In a–f and j–l, one-way ANOVA with Tukey’s multiple comparisons; in g, two-way ANOVA with Dunnett’s multiple comparisons; in h, two-tailed t-Student and fold change (Z > 0.5).
We confirmed the importance of proline, as opposed to other substrates that might become available to the organisms in the airway, by constructing an adsA mutant lacking the genes encoding both of the major proline transporters putP and proT in the USA300 background25. In the absence of putP and proT, the increased level of infection achieved by the adsA mutant was lost (Fig. 3c). Despite the published studies suggesting that adsA expression functions as a ‘virulence’ factor, we did not observe major changes in the recruitment of myeloid cells beyond the expected differences associated with bacterial load (Fig. 3d–f). Given the known role of AdsA in the inhibition of phagocytic killing, we evaluated possible differences in the susceptibility of the WT and adsA mutant to phagocytic clearance in vitro. In a gentamicin protection assay, the adsA mutant showed decreased survival within bone-marrow-derived macrophages (BMDM) (Fig. 3g). As might be predicted for the expected effects of adsA as a virulence factor26, we observed differences at 2 h following infection, where the adsA mutant was more readily killed, but by 4 h both WT and ΔadsA strains were killed equally. Given the lack of correlation of the in vitro myeloid cell functionality and the persistence of the adsA mutant in vivo, we postulated that the enhanced ΔadsA infection likely reflects a metabolic fitness advantage in the setting of established infection as preferred substrates become available.
To better understand the metabolic impact of adsA expression on S. aureus persistence in the lung, we performed a metabolomic study on mouse bronchoalveolar lavage (BAL) fluid collected at 72 h of infection with the WT or ΔadsA (Extended Data Fig. 2a). Non-biased pathway analyses indicated a significant upregulation of both purine degradation and citrulline metabolism in ΔadsA infected airways (Fig. 3h).
We evaluated the impact of host purine degradation as an explanation for the enhanced ΔadsA survival in vivo, compared with the WT strain. Host adenosine is predominantly generated through the sequential activity of CD39 (cluster of differentiation 39) (adenosine diphosphate to adenosine monophosphate (AMP)) and CD73 (AMP to adenosine) nucleotidases27 expressed by regulatory T (Treg) cells. We found that Treg cells from ΔadsA-infected mice expressed more ectonucleotidase than those infected with the WT control (Extended Data Fig. 2b,c), possibly compensating for the total pool of adenosine. However, neither Treg cell depletion nor global elimination of CD73 (Extended Data Fig. 2i,j) altered bacterial recovery from the lung (Extended Data Fig. 2d–f), indicating that host adenosine production is not responsible for the increased survival of ΔadsA mutant in the lung.
We proceeded with evaluating the superpathway of citrulline metabolism (Fig. 3i) as the basis for selective advantage of S. aureus strains with decreased adsA expression in vivo. The accumulation of ornithine, citrulline, glutamate and glutamine in the ΔadsA infected airway were consistent with an increase in proline metabolism (Fig. 3j). However, given the involvement of proline in a network of metabolic pathways characterized by the abundance of several other amino acids, we further assessed the central role of proline as the preferred substrate for the ΔadsA and clinical host-adapted strains in vivo. We compared growth in defined media supplemented with other amino acids that generate glutamate (Extended Data Fig. 3a–e). In the absence of proline, growth of WT and ΔadsA were both impaired; the addition of histidine or arginine, but not glutamate, enhanced growth of the ΔadsA but not the WT parental strain. The CF clinical isolates had an absolute requirement for added proline or arginine (Extended Data Fig. 3f–j). Strain T2015 utilized these amino acids in vitro to varying extents, whereas neither A2001 nor T2015 could utilize arginine. Thus, in vitro, both the ΔadsA and the clinical isolates are able to exploit other available amino acids that are provided exogenously, and their enhanced growth is not necessarily limited to proline uptake, as was preferred in vivo.
Targeted metabolomics indicated a trend toward decreased proline in ΔadsA infected airways (Fig. 3k) and a significant decrease in glucose (Fig. 3l), usually a preferred S. aureus substrate but only associated with the WT strain compared with a phosphate-buffered saline (PBS) control. These findings overall are consistent with a metabolic switch by the ΔadsA mutant and the clinical strains to proline catabolism in the setting of established infection, as S. aureus prefers to transport and not synthesize proline when glucose is perceived as limited18,25,28.
While BAL represents a combination of eukaryotic and prokaryotic products, we expected most of these metabolites to be of eukaryotic origin. We verified the eukaryotic origin of the staphylococcal substrates by demonstrating the upregulation of the corresponding host enzymes involved in proline and collagen (I and IV) synthesis (Fig. 4a) and greater abundance in ΔadsA infected hosts (Fig. 4b,c). Increased transcription of collagen triple-helix repeat complex (cthrc) and alpha smooth muscle actin (α-sma) suggested the participation of activated fibroblasts in ΔadsA infection (Fig. 4c) as did the combination of increased IP-10, IL-6, MCP-1 and TGFB1 (Fig. 4d,e)29,30. Despite the increased expression of fibroblast-associated genes in ΔadsA infection (Fig. 4a–c), fibroblast numbers in either BAL or lung were equivalent to WT (Fig. 4f). We also explored other mechanisms of collagen turnover which can foster a similar response31. We found roughly equivalent production of matrix metalloproteases (MMPs) in BAL (Extended Data Fig. 4a) and their transcripts (mmp) in the lung (Fig. 4c) which are typically activated in frank tissue damage (Extended Data Fig. 4a). Cytokines and chemokines commonly associated with stress and inflammation during infection were also comparable in ΔadsA and WT infected hosts (Extended Data Fig. 4b), as was the gross histology of the infected lung (Extended Data Fig. 4c). These findings suggest that this host response was specific to ΔadsA infection and not simply a reflection of the increased ΔadsA bacterial burden or frank tissue damage. Thus, ΔadsA mutants appear to enjoy an advantage in vivo by virtue of their ability to exploit available proline, which seems likely to be synthesized by activated fibroblasts.
a, Schematic diagram of proline and collagen biosynthesis pathways in the host. Dotted arrows, translocation; solid arrows, conversion; in bold are critical enzymes. b,c, At 72 h following intranasal inoculation with S. aureus Newman WT control or ΔadsA, we monitored host lung RNA transcripts associated with proline biosynthesis (b) and collagen biosynthesis turnover via mmps and markers of fibroblast activity (c). d,e, Selected cytokines were quantified in BAL. f, CD140a+SCA-1+ fibroblast numbers in lung and BAL fibroblasts quantified. Data presented as mean ± s.e.m. In b–f, n = 3. Significance determined by *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. In b–f, one-way ANOVA with Tukey’s multiple comparisons.
Airway fibroblasts provide proline for S. aureus consumption
We sought to confirm the role of fibroblasts as the source of collagen and hence proline fuelling ΔadsA growth. We isolated primary murine fibroblasts (Extended Data Fig. 4d) to use as a substrate for S. aureus infection. We observed that while the WT bacterial burden remained unchanged over time, the ΔadsA bacterial burden was significantly increased at the 72 h time point (Fig. 5a and Extended Data Fig. 4e), coincident with the depletion of collagen in the fibroblast cultures infected with either ΔadsA or the 2001 clinical isolate, but not the WT or PBS control at 72 h (Fig. 5b). We inhibited collagen synthesis by the fibroblasts with halofuginone32,33,34, which significantly decreased the ΔadsA bacterial burden but not that of WT controls and corresponded with collagen depletion in the culture supernatant (Fig. 5c,d). As in the untreated condition, we observed increased ΔadsA bacterial burden compared with WT at 24 and 48 h post infection. We confirmed the in vivo significance of ongoing collagen synthesis in selectively supporting infection by the ΔadsA mutants as ΔadsA mice treated with halofuginone (by the intraperitoneal route) lost their in vivo advantage. Infection by WT S. aureus at 72 h was not significantly altered by halofuginone compared with a vehicle control (Fig. 5e). Neither bacterial (Extended Data Fig. 4f–i) nor fibroblast (Extended Data Fig. 4j,k) viability was significantly decreased by halofuginone. This set of experiments strongly supports our hypothesis that host-adapted clinical isolates of S. aureus, as well as our surrogate ΔadsA strain with similar metabolic properties, exploit host fibroblast deposition of collagen and hence proline synthesis to persist in the lung.
Primary murine fibroblasts were infected with S. aureus WT control or ΔadsA strains. a,b, Bacterial burden (a) and quantification of collagen by ELISA (b) in primary murine fibroblasts culture supernatants following infection with WT control, ΔadsA or clinical strains. c,d, Bacterial burden (c) and collagen quantification by ELISA (d) in primary murine fibroblast cultures treated with halofuginone, an inhibitor of collagen synthesis or vehicle (DMSO). e, Bacterial burden 72 h post-WT control or ΔadsA infection of mice treated with intranasal halofuginone or DMSO. Data presented as mean ± s.e.m. In a–d, n = 6; in e, n = 3. Significance determined by *P < 0.05, **P < 0.01. In a–d, two-way ANOVA with Brown–Forsythe test; in e, one-way ANOVA with Tukey’s multiple comparisons. PF, primary murine fibroblasts; HF, halofuginone.
Carbon catabolite repression directs proline utilization
Our data thus far suggested that S. aureus clinical isolates, along with ΔadsA mutants that also consume host-generated proline, enjoy a growth advantage in vivo compared with WT organisms. In vitro, carbon catabolite repression (CCR) directs the S. aureus metabolic switch from glycolysis, fuelled by glucose, to proline metabolism and oxidative phosphorylation via the TCA cycle31,35. The perception of limited glucose in the airway is mediated by carbon catabolite protein A (CcpA) and phosphorylated Hrp to direct the expression of >300 genes36,37, a complex system conserved in many bacterial species38,39,40. CCR also provides a mechanism to link bacterial metabolism and the expression of virulence determinants41,42. In S. aureus, CcpA negatively regulates putP and other amino acid transporters and positively regulates pckA, a gene central to glycolysis35, whereas carbon catabolite protein E (CcpE) positively regulates citB and feeds the TCA cycle activity43 (Fig. 6a). It seemed likely that CCR is the dominant factor regulating S. aureus metabolism in the lung and drives selection of the host-adapted clinical isolates as well as the parallel metabolic behaviour of the adsA mutant.
a, Schematic diagram of proposed CCR regulation of S. aureus consumption of collagen and proline. b–e, In silico identification of CRE binding sites using composite and verified consensus sequence targets for CcpA (Seq1, Seq2), known to regulate pckA, and CcpE (Seq3), known to regulate citB. CcpA homologous sequences in pink; CcpE homologous sequences in green. Putative consensus sequences in S. aureus Newman WT (accession number NC_009641.1) indicating CcpA regulation of putP, scpA and pckA (control) (b) and CcpE regulation of adsA and citB (control) (c); yellow highlighting, mismatch. Putative consensus sequences in S. aureus respiratory clinical isolates 2001 and 2015 indicating CcpA regulation of putP, scpA and pckA (control) (d) and CcpE regulation of adsA and citB (control) (e); yellow highlighting, mismatch. f–k, Effects of CcpA and CcpE in mice infected with S. aureus WT Newman control and ΔadsA (f,g) or JE2 WT control and isogenic ccpA::Tn, ccpE::Tn and putP::Tn transposon mutants (h–k) for collection of prokaryotic RNA and bacterial burden at 72 h. f,g, Transcripts from in vivo lungs (f) and respective in vitro inoculum (g) grown in LB. h,i, Lung bacterial burden (h) and bacterial in vivo competition (i). j,k, In vivo transcripts from ccpA::Tn (j) or ccpE::Tn (k) compared with a WT JE2 control in the lung. l, Activation of the adsA promoter detected using a LUX reporter expressed in ccpA::Tn, ccpE::Tn or JE2, shown as a function of bacterial growth (OD600). Data presented as mean ± s.e.m. In f, n = 4; in g and l, n = 3; in h, n = 4; in i, n = 5; in j and k, n = 4. Significance determined by P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. In f, g, j and k, two-tailed multiple t-Student with Kolmogorov–Smirnov tests; in h, one-way ANOVA with Kruskal–Wallis comparisons; in i, Mann–Whitney test; in l, two-way ANOVA with Dunnett’s multiple comparisons.
We utilized a bioinformatic approach to verify the presence of catabolite responsive elements (CRE) associated with both CcpA and CcpE in the promoter regions of selected genes in the clinical isolates 2001 and 2015 (Supplementary Tables 2 and 3) and in the Newman genome. Published consensus sequences associated with CcpA regulation (TGTAAAANNNTTTACA, WTGNAANCGNWNNCW)44,45 were found in putP and pckA as expected35,46 and also in the promoter region of the collagenase scpA (Fig. 6b,d) in Newman strains and 2001 and 2015 isolates. A composite sequence associated with CcpE regulation of citB (ATAANNNNNNTTAT)47 was identified in both of these clinical isolates and Newman strains in the promoter region of adsA (Fig. 6c,e).
We predicted that adsA must also be regulated by CCR which would explain the preferential consumption and growth on proline in vivo as well as reliance on oxidative phosphorylation and the TCA cycle to generate ATP. We compared rates of transcription of our selected genes in organisms collected directly from the murine lung at 72 h of infection with WT or ΔadsA compared with bacteria grown in LB (Fig. 6g). The ΔadsA strains had increased putP, scpA and citB but decreased pckA expression (Fig. 6f), consistent with positive regulation by CcpE activity and with the inactivity of CcpA.
The impact of CCR regulation on S. aureus in vivo was addressed by testing the ability of mutants lacking specific CCR elements to cause pneumonia. Bacterial burden in mice infected with ccpA::Tn in the JE2 background stain was greater than that of mice infected with WT, ccpE::Tn or putP::Tn strains (Fig. 6h). A competition experiment infecting mice with equal inocula of the WT and ccpA::Tn strains also illustrated the advantage of the ccpA mutant (Fig. 6i) also indicating that the loss of adsA expression increases persistence in the lung. We noted increased expression of putP, scpA and citB but decreased expression of pckA in ccpA::Tn infection, compared with WT control (Fig. 6j). ccpE::Tn S. aureus collected directly from infected mice showed increased expression of adsA and achieved similar levels of infection as the WT strain (Fig. 6k,h). The combination of inactive CcpA with active CcpE generates the pattern of gene expression observed in vivo in the ΔadsA and adapted clinical isolates (Figs. 1b,c and 6f).
We confirmed CCR regulation of adsA expression using a luciferase (LUX)-promoter fusion construct to monitor the relative activation of the adsA promoter in the parental JE2 background, ccpA::Tn or ccpE::Tn strains, having established equivalent rates of growth over the assay conditions (Fig. 6l). In the ccpA::Tn strain, we observed less adsA reporter activity, compared with the ccpE-deficient strain in which adsA expression was significantly increased (Fig. 6l). The control for our promoter fusion assay documented the expected positive regulation of citB by CcpE and the relative lack of influence of CcpA on its expression43. The CCR regulatory system, which optimizes staphylococcal bioenergetics, directs S. aureus metabolism in this setting in which collagen deposition, as part of the ongoing process of host airway repair, provides proline.
CCR is engaged in S. aureus adaptation to available nutrients, and hence, we predicted that isolates that had undergone adaptation to another collagen-rich site, such as the skin, would also show CCR-directed metabolic activity. We screened ten S. aureus strains from children with atopic dermatitis, although the chronicity of infection was not known, for patterns of gene expression and preferred substrate utilization (Extended Data Fig. 5a). Roughly half of the isolates reflected CCR control with suppressed adsA and increased putP and scpA expression. A representative strain AD2 showed significant preference for proline consumption compared with a strain Newman control and showed increased growth rates in proline (Extended Data Fig. 5b,c). Using the ΔadsA mutant as a surrogate for the clinical and less virulent host-adapted strains, we observed increased ability to infect murine skin (Extended Data Fig. 5d) generating smaller lesion sizes compared with WT (Extended Data Fig. 5e). We also found increased induction of host collagen synthesis, with increased expression of col1a1, col1a2 and col4a1 (Extended Data Fig. 5f), findings similar to those of the pulmonary isolates. We utilized the same bioinformatic approach to verify the presence of CRE associated with both representative clinical isolates AD2 and AD8 in the promoter regions of selected genes and found similar matches (Extended Data Fig. 5g,h). We also found that the ccpA::Tn mutant was associated with significantly greater bacterial loads in a murine skin infection model, consistent with the pattern of increased infection in the lung (Extended Data Fig. 5i). Thus, CCR direct gene expression in S. aureus isolates to optimize utilization of available carbon sources is not limited to the lung.
Our findings overall indicate that S. aureus adapts its metabolic activity in the lung and perhaps at other sites as directed by CCR. In a setting of accessible and abundant collagen, as in the airway and possibly the skin, the suppression of CcpA activity enables utilization of amino acid transporters. The organisms that preferentially utilize proline derived from collagen show a substantial growth advantage and establish persistent infection.
Discussion
S. aureus is a complex pathogen that optimizes its metabolic activity according to the constraints of the local environment. We found that organisms that persist in the lung exploit the availability of proline released from collagen, which is actively synthesized by fibroblasts during the process of tissue repair. Proline consumption along with expression of adsA is responsible for the increased growth of clinical isolates in the damaged lung. These metabolic properties apparently predominate over the expected anti-inflammatory consequence of AdsA as have been described in models of sepsis or renal abscess formation48,49,50 and indicate tissue-specific adaptation. Strategies to prevent or suppress S. aureus infection have focused upon their expression of multiple toxins and gene products that interfere with host immune clearance6,7,8. While these genes may be important in the initial stages of infection, an increasing literature indicates that the organisms causing more persistent infection have an entirely different pattern of gene expression. Such host-adapted strains have substantially altered metabolic activity, giving them a growth advantage specifically in the local microenvironment. Our data were generated from S. aureus strains isolated over years of chronic infection compared with initial isolates or laboratory strains. Exactly how long it takes for S. aureus metabolic rewiring to take place in vivo remains to be established and whether analogous metabolic changes occur in S. aureus at other sites of infection remains to be established. Appreciating the metabolic directives that drive S. aureus gene expression in vivo may provide novel targets for therapy or infection prevention in patients, especially those with damaged airways, fibrosis and collagen deposition who are at increased risk for S. aureus pneumonia.
Methods
Ethics statements
Human participants
S. aureus strains were obtained from adults 22–44 years of age in the CF program at Yale University, from children 0.5–2 years of age in the CF program at Seattle Children’s hospital and from a longitudinal collection from a single adult in the CF program at the University Hospital of Münster, all as part of routine care. None of the CF patients studied were under CF transmembrane conductance regulator modulator therapy (for example, Kaleydeco, Orkambi).
Bacterial strains
The bacterial strains used in this study are shown in Supplementary Table 4. All strains were grown at 37 °C on LB plates supplemented with 1% agar; LB broth shaking, with the exception of Newman c:adsA in/on LB + 10 mg l−1 chloramphenicol; USA300 c:adsA in/on LB + 5 μg ml−1 tetracycline; or ccpA::Tn, ccpE::Tn, putP::Tn in/on LB + 5 μg ml−1 erythromycin (Nebraska Transposon Mutant Library). For the experiments requiring supplementation with ATP, AMP, adenosine, proline, collagen, glycine, hydroxyproline or ornithine, the medium was pH corrected to 7.0 before filter sterilization using 0.20 μm filters. Inoculum was estimated on the basis of an optical density of 600 nm (OD600) and verified by serial dilution plating.
Mouse experiments
CD57/BL6 and 129S6/SvEvTac mice 8–10 weeks old were purchased from Jackson Laboratories and Taconic, respectively, and housed in a humidity-controlled facility at 18–23 °C and 12 h light/dark cycles. All animal studies were approved (Columbia Institutional Animal Care and Use Committee Protocol AABE8600) and performed in accordance with the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health (NIH), the Animal Welfare Act and US federal law. Two transgenic strains were utilized: GFP FoxP3 DTR (stock number 016958) with background (stock number 005304) and CD73−/− (stock number 018986) with background (stock number 000664). Each in vivo and ex vivo experiment was performed using an equal ratio of male to female animals. Sex was not expected to influence results. Animals were randomly assigned to cages, and their health was routinely monitored by an Institutional Animal Care and Use Committee veterinarian.
Methods details
Materials and resources
A comprehensive list of materials and resources has been provided in Supplementary Table 4.
Cloning of USA300 adsA mutant
The adsA mutant in strain background USA300 was generated using the pIMAY protocol21,51. A fully synthetic 1,200 bp double-stranded DNA fragment was designed and ordered from Invitrogen (GeneArt). This DNA fragment contained the following features (from 5′ to 3′): SphI and BamHI restriction sites, a Pxyl/tet promoter, a ribosome-binding site, a codon-optimized pheS* gene, and KpnI and BglII restriction sites. The synthetic DNA was cloned in the topoisomerase vector pCR8 and transformed into XL1-Blue cells, resulting in strain ANG3944. The resulting plasmid pCR8-pheS* was found to have two base exchanges in the 10 promoter region, altering the optimal sequence TATAAT to CTTAAT (Pxyl/tet*). The Pxyl/tet* pheS* construct was sub-cloned from pCR8-pheS* via the SphI and BglII restriction sites into pIMAY [7], and the resulting plasmid pIMAY* was introduced into XL1-Blue, creating strain ANG4005. pIMAY* was then shuttled through IM08B [18], yielding strain ANG5002, and further used to transform S. aureus RN4220 and LAC*, creating strains RN4220 pIMAY* (ANG5079) and LAC* pIMAY* (ANG5080), respectively.
Plasmid pIMAY*-DmgtE was constructed by amplifying 1,000 base pairs of genomic DNA upstream and downstream of mgtE from S. aureus LAC* (ANG1575) chromosomal DNA using primer pairs P2374 (TATA CCCGGG AATGT- TAATT CAATACAATA CCGTGTAGC)/P2375 (AAAT- GATGTA GCACGCTCTT TTTCATCTGT GTTCATTGAC) and P2376 (GAAAAAGAGC GTGCTA- CATC ATTTATGGCT TACTTAATTT AAG)/P2377 (ACGT GAATTC TTGAAATGAT AAATGCAACG ATTAAAATCG), which were then spliced together using the primer pair P2374/P2377. This fusion construct contained the first and last 30 bases of mgtE and thus created a non-functional truncated open reading frame with flanking regions. The spliced DNA was cut with XmaI and EcoRI, ligated into pIMAY* (ANG5002) cut with the same enzymes and introduced into XL1-Blue, yielding strain ANG5081. The plasmid was subsequently introduced into IM08B to create strain ANG5082, and the plasmid isolated from this strain was introduced into S. aureus strains RN4220 [36] and LAC*. Two colonies each were selected, yielding strains ANG5083–86, and used in the allelic exchange procedure.
For the construction of the complementation plasmid piTET-mgtE, the mgtE gene, including its ribosome-binding site, were amplified using primers P2384 (CGATCCTAGG ACAGGGGGTGTAAGTATGTCAATGAACAC) and P2385 (AAGCAGATCT TTAAATTAAGTAAGCCA TAAATGATGTAGC). The PCR product was cut with AvrII and BglII and ligated with plasmid piTET (ANG284), which had been cut with the same enzyme. The resulting plasmid was recovered in Escherichia coli XL1-Blue cells, yielding strain ANG4144. The plasmid was then shuttled through IM08B, creating strain ANG4155, and used to transform one of the LAC* DmgtE deletion strains (ANG5091), resulting in the construction of the complementation strain LAC*DmgtE piTET-mgtE (ANG5288). In addition, the empty vector piTET isolated from strain ANG3928 was introduced into the mgtE mutant strain LAC*DmgtE 581 (ANG5091), (ANG5287) creating strain LAC*DmgtE piTET.
RNA-sequencing of S. aureus Newman WT and ΔadsA
S. aureus WT and ΔadsA were grown in LB with or without 100 μM proline to late exponential phase. Bacterial pellets were incubated in a cell wall lysis mixture TE buffer at pH 8 (30 mM Tris, 1 mM EDTA, 15 mg ml−1 and 200 μg ml−1 proteinase K) containing mutanolysin, lysostaphin and lysozyme at 37 °C for 30 min. TRK lysis buffer Total RNA lysis Kit (Omega #PRQ21) and 70% ethanol were added to each sample, before transferring to E.Z.N.A RNA isolation columns. RNA was isolated following the manufacturer’s instructions, and DNA was selectively degraded using the DNA-free DNA removal kit. The RNA was precipitated with 0.1 volume 3 M sodium acetate (Thermo Fisher #S209) and 3 volumes of 100% ethanol, recovered by centrifugation and washed with ice-cold 70% ethanol. A ribosomal RNA-depleted complementary DNA library was prepared according to the manufacturer’s instructions using the Universal Prokaryotic RNA-Seq Prokaryotic AnyDeplete kit (NuGEN #0363-32) and sequenced with Illumina HiSeq. Raw base calls were converted to fatsq files using Bcl2fastqs. Filtered reads were aligned to the Newman (Refseq NC_009641.1) reference genome using STAR-Aligner v2.7.3a. The mapped reads were annotated for read groups and marked for duplicates using the Picard tools v2.22.3. The raw counts were quantified using Subreads:FeatureCounts v1.6.3 and processed for differential gene expression using DEseq2 in R v3.5.3. Analysis of sequencing for clinical isolates was performed by Windmüller and colleagues. Briefly, data were stratified into ≥2f, ≥4 f, ≥8 f and genes that were either not expressed in both isolates or not present on the genome level in both isolates. To exclude low abundant transcripts, a threshold of less than 3 reads per gene was used (accession number GSE268637).
Mouse lung infection
Mice were anaesthetized with a 20:1 ketamine xylazine formulation infected intranasally with 1 × 108 colony-forming units (c.f.u.) of the specified S. aureus strain. Bacteria were delivered in 50 μl PBS, and sterile PBS was used as control (vehicle). All mice were killed 72 h post infection for BAL fluid collection and lung tissue processing. Lung tissue was homogenized through 40–70 μm cell strainers. Colony-forming units were determined by serially diluting aliquots of the BAL and of lung homogenates, which were plated on LB agar plates (with the exception of Newman c::adsA on LB + 10 mg l−1 chloramphenicol; USA300 c::adsA on LB + 5 μg ml−1 tetracycline; ccpA::Tn, ccpE::Tn and putP::Tn on LB + 5 μg ml−1 erythromycin). The remaining BAL was spun and stored for cytokine and metabolomic analysis. BAL and lung cells were treated with a hypotonic lysis solution for elimination of red blood cells, counted and prepared for fluorescence-activated cell sorting (FACS) as described below.
In vivo Treg cell depletion
GFP FoxP3 DTR (strain number B6.129(Cg)-Foxp3tm3(DTR/GFP)Ayr/J; stock number 016958) allows for depletion of FoxP3+ cells upon daily administration of diphtheria toxin. Mice were injected intraperitoneally with 50 μg kg−1 diphtheria toxin resuspended in sterile water once a day for 5 days. Mice were inoculated on the third day and processed on the sixth day as indicated above.
Untargeted metabolomic analysis
Aliquots of BAL supernatants were mixed in equal volume (v/v) with methanol and stored at −80 °C. Metabolites were identified and quantified by high-resolution mass spectrometry. Sample runs were performed on a Q Exactive HF Hybrid Quadrupole-Orbitrap Mass Spectrometer (Thermo Fisher) coupled to a Vanquish UHPLC System (Thermo Fisher). Chromatographic separation was achieved on a Syncronis HILIC UHPLC column (2.1 mm × 100 mm × 1.7 µm, Thermo Fisher) using a binary solvent system at a flow rate of 600 µl min−1. Solvent A, 20 mM ammonium formate at pH 3.0 in mass spectrometry grade H2O; solvent B, mass spectrometry grade acetonitrile with 0.1% formic acid (%v/v). A sample injection volume of 2 μl was used. The mass spectrometer was run in negative full-scan mode at a resolution of 240,000 scanning from 50 to 750 m/z. Metabolites were identified using the known chromatographic retention times of standards, and metabolite signals were quantified using E-Maven v0.10.0.
Pathway analyses
Following bulk RNA sequencing, bacterial pathways were identified using the National Institutes of Allergy and Infectious Diseases (NIAID)/NIH free Database for Annotation, Visualization and Integrated Discovery. The use of the tool relies on the identification of the bacterial species and genus and specific gene coding. In brief, a comprehensive rubric of NCTC8325 codes in the SAOUHSC ortholog format matching the sequencing locus discovery was assembled manually using the AureoWiki free online repository unique to S. aureus. The input of the codes in Database for Annotation, Visualization and Integrated Discovery generated unbiased functional annotations and pathway suggestions as well as a list of corresponding gene counts per pathway.
Following untargeted metabolomics of BAL fluid, host pathways were identified using the Ingenuity Pathway Analysis bioinformatic software (QIAGEN). In brief, a rubric of the Kyoto Encyclopedia of Genes and Genomes metabolite codes matching the macromolecular discovery was assembled. Data were paired for t-Student comparison and averaged to a single value for the generation of a fold change (log2). The input of the above in Ingenuity Pathway Analysis generated unbiased pathway suggestions as well as significance with respect to cumulative pathway P values and relative fold change.
Cytokine analysis
Aliquots of BAL supernatants were stored at −80 °C. Cytokine concentrations in BAL supernatants were quantified with the Mouse Cytokine Proinflammatory Focused M31-plex Discovery Assay (Eve Technologies), which quantifies eotaxin, G-CSF, GM-CSF, IFNγ, IL-1α, IL-1β, IL-2, IL-3, IL-4, IL-5, IL-6, IL-7, IL-9, IL-10, IL-12p40, IL-12p70, IL-13, IL-15, IL-17A, IP-10, KC, LIF, LIX, MCP-1, M-CSF, MIG, MIP-1α, MIP-1β, MIP-2, RANTES, TNFα and VEGF-A; TGFβ 3-plex, which quantifies TGFβ1, TGFβ2 and TGFβ3; and MMP 5-plex, which quantifies MMP-2, MMP-3, MMP-8, proMMP-9 and MMP-12, using a bead-based multiplexing technology also known as addressable laser bead immunoassay.
FACS
BAL and lung cells were treated with a hypotonic lysis solution for elimination of red blood cells and stained with fluorescent antibodies as follows. All antibodies were purchased on BioLegend unless otherwise indicated—lymphocytes: live/dead, CD45, CD4, CD8, CD25, FoxP3, CD39 and CD73; granulocytes and leukocytes: SiglecF, CD11b, CD11c, MHCII, LY6G, LY6C, CD39 and CD73; fibroblasts: live/dead, CD45, Dump gate (CD31, EPCAM, TER119, THY 1.2, CD146), SCA-1 and CD140a. Gating strategies are shown in Supplementary Figure 4.
All antibodies were used at a dilution of 1:200 with the exception of CD45 and FoxP3 (1:100) and CD25 (1:50). All dilutions were performed in PBS, and cells were stained for 1 h at 4 °C. For lymphocytes, after washing, the cells were fixed, permeabilized and intracellularly stained with a monoclonal anti-FoxP3 antibody in permeabilization buffer for 30 min at room temperature. After a final wash, the cells were stored in 2% paraformaldehyde until analysis on the BD LSRII (BD Biosciences) using FACSDiva v9. Flow cytometry was analysed with FlowJo v10.
Primary fibroblast isolation
Immediately after euthanasia of the experimental animal, the chest cavity was opened and the lungs flushed with PBS through the heart from the right ventricle. The lungs were both inflated and immersed in dispase (~0.9 ml, 50 U ml−1), incubated at room temperature for 20 min and isolated from surrounding structures. Lobes were chopped mechanically and incubated at 37 °C for 10 min in MEM + DNase to complete digestion. The mixed cells were filtered through 100 µm and 40 µm cup filters, and fibroblasts were removed from the suspension by three successive adherence steps on plastic. With respect to other cells, fibroblast adhesion is faster and facilitated by the continuous and sequential generation of filopodial structures containing integrins52. Moreover, adhesion is facilitated by the treated plasma of cell culture plates, which permits obtaining a consistent cell attachment and growth. Primary fibroblasts are washed three times with 1× PBS and incubated with Dulbecco's modified Eagle's medium (DMEM) 10% fetal bovine serum (FBS) and antibiotics.
Primary murine fibroblast immunostaining and imaging
Primary fibroblasts were fixed with 4% paraformaldehyde for 10–20 min at room temperature and washed three times with PBS for 5 min. The cells were permeabilized with 0.25% Triton X-100/PBS for 30 min at room temperature followed by blocking in 10% donkey serum diluted in PBS for 1 h. Vimentin primary antibody was diluted (1:400) in 5% donkey serum in PBS and incubated at 4 °C overnight. The following day, cells were washed three times with PBS before applying the secondary antibody (1:400) for 1 h incubation at room temperature. Following secondary incubation, sections were washed three times for 10 min with PBS and finally stained with DAPI (1:1,000) for 10 min at room temperature. PenStrep (1%) was added in 200 μl PBS before imaging with a motorized Leica Stellaris DMi8 (Leica Microsystems) inverted confocal microscope. Positive cells were counted over total number of cells. LasX v1.4.5 was used for fibroblast imaging.
Primary murine fibroblast infection time course
Primary fibroblasts were isolated from mouse lungs as described above and incubated in antibiotics for 4–5 days until 70–80% confluency was achieved. Cells were counted and plated (including two extra wells per plate) the day before infection. On the morning of the infection, the two extra wells were counted to determine the multiplicity of infection (MOI). After media change, the cells were infected with bacterial subcultures of WT and ΔadsA Newman strains grown to late exponential phase at a MOI of 10, or treated with PBS control. The infection was allowed to proceed until pre-determined time points. Processing of plates at the selected time points of 2, 24 and 72 h involved retrieving and plating one aliquot of supernatant for c.f.u. counts to determine extracellular multiplication due to collagen consumption. After 1 h incubation at 37 °C in 500 ng ml−1 gentamicin, aliquots of supernatants were collected and stored at −80 °C for protein quantification and RNA isolation before addition of 0.25% trypsin for 9 min at 37 °C. Aliquots of cells were either stored at −80 °C or used to measure cell viability.
Inhibition of collagen synthesis
Halofuginone, a semisynthetic quinazolinone (Collgard Biopharmaceuticals), was reconstituted in DMSO at a stock concentration of 5 mg ml−1.
For in vitro study, inhibition of collagen synthesis in cell culture was achieved via the administration of 0.25 mg ml−1 of halofuginone per well32 once before infection.
For in vivo study, inhibition of collagen synthesis was achieved through the intraperitoneal administration of halofuginone at a dose of 0.1 mg kg−1 in 100 μl of di-water for 3 sequential days, starting from the day of infection33,34. The health of control/uninfected mice was monitored to survey potential effects of the drug or vehicle. Animal weight was recorded every 24 h until euthanasia.
Collagen detection enzyme-linked immunosorbent assay
Collagen was quantified using the Mouse Pro-Collagen alpha 1 enzyme-linked immunosorbent assay (ELISA) kit (abcam) according to manufacturer’s instructions. Briefly, supernatant and cell aliquots from in vitro primary fibroblast infections and from in vivo airway BAL fluid were treated solubilized with chilled 1× Cell Extraction Buffer PTR, incubated on ice for 20 min and centrifuged at 18,000 g for 20 min at 4 °C. Supernatants (50 μl) were transferred to the assay plate along with 50 μl of antibody cocktail (600 μl 10× capture antibody, 600 μl detector antibody and 2.4 ml antibody diluent 5BR) per well and incubated at 400 r.p.m. for 1 h at room temperature. After washing, 100 μl of tetramethylbenzidine (TMB) substrate was added, and the plate was incubated shaking for 10 min in the dark. The addition of 100 μl of stop solution occurred before the endpoint reading at OD 450 nm.
Collagen staining and pathology assessments
Following euthanasia, the mouse was tracheostomized and cannulated. The thorax was opened to facilitate lung expansion upon gentle infusion of formalin-free tissue fixative through the cannula, attached to a 5 ml syringe positioned above the mouse. Lungs were collected en bloc, placed in formalin-free tissue for 24 h, in 70% ethanol for 24 h and prepared in paraffin blocks. Haematoxylin and eosin and Masson’s trichrome staining were performed on 5 μm sections (two sections cut 25 μm apart for each lung) for gross pathology assessment.
In vitro RNA isolation
Bacterial strains Newman WT, Newman ΔadsA and all clinical isolates were grown to the exponential phase and normalized to an OD600 of 1. Bacteria were incubated in RNAprotect cell reagent and TE buffer at pH 8 (30 mM Tris, 1 mM EDTA and 200 μg ml−1 proteinase K) containing mutanolysin, lysostaphin and lysozyme at 37 °C for 30 min. TRK lysis buffer and 70% ethanol were added to each sample, before transferring to E.Z.N.A RNA isolation columns. RNA was isolated following the manufacturer’s instructions, and DNA was selectively degraded using the DNA-free DNA removal kit.
In vivo RNA isolation from mouse lungs
Total RNA (eukaryotic and prokaryotic) was isolated from infected whole mouse lungs added to TRIzol Reagent following the manufacturer’s instructions and homogenized. Aliquots of these samples were stored for immediate reverse transcription to obtain largely eukaryotic cDNA analysis. The remaining sample was depleted of eukaryotic RNA with a MICROBEnrich Kit. The samples were added to binding buffer and capture oligonucleotide mix, incubated at 70 °C for 10 min and 37 °C for 1 h to hybridize prokaryotic RNA with the oligonucleotides. Host RNA was then depleted using OligoMag beads, and enriched prokaryotic RNA was recovered following the manufacturer’s instructions. DNA was selectively degraded using the DNA-free DNA removal kit. Prokaryotic RNA was precipitated using MICROBEnrich Kit as per manufacturer’s instructions.
cDNA synthesis and quantitative real-time PCR
RNA was converted to cDNA using a High-Capacity cDNA Reverse transcription kit and a SimpliAmp thermocycler (Applied Biosystems). Quantitative real-time PCR was performed using the relevant primers and PowerUp SYBR Green PCR Mastermix. All primers used in this study are listed in Supplementary Table 1. The experiment was performed on a StepOnePlus Real-time PCR System (Applied Biosystems) using StepOne v2.3. Data were analysed using the ∆∆CT method. 16S was used as a control housekeeping gene for bacteria, and the Newman strain of S. aureus was included as a wild-type control. GAPDH was used as a control housekeeping gene, and the uninfected hosts were used as controls.
Carbon utilization assays
S. aureus WT, adsA mutants and clinical isolates were grown to the exponential phase and normalized to an OD600 of 2. A stock solution of 2 × 107 bacteria per ml was made in 1× IF-Oa buffer supplemented with 1× Redox Dye Mix H. About 100 μl of this stock solution was added to each well of a BIOLOG PM1 Microplate (BIOLOG) delivering 2 × 106 bacteria. This plate was incubated at 37 °C for 24 h, and absorbance was read at 590 nm on a SpectraMax M2 plate reader.
Growth curves
A U-bottomed, clear 96-well plate was prepared with LB or chemically defined media (CDM), LB or CDM supplemented with 100 µM proline and LB or CDM supplemented with 30–10 µg ml−1 collagen. Each well was inoculated with 2 × 106 bacteria. Absorbance at 600 nm was read every 30 min for 20–50 h on a SpectraMax M2 plate reader. The plate was incubated at 37 °C with shaking before and after every read. Compound are listed in Extended Data Table 5.
Biofilm assays
A flat-bottomed, clear, 96-well plate was prepared with LB or LB with 0.5% glucose (w/v), supplemented with serially diluted, pH-corrected proline. Each well was inoculated with 1.5 × 106 bacteria, and the plate was left to incubate statically overnight at 37 °C. The next morning, absorbance at 600 nm was determined on an Infinite M200 plate reader (Tecan). To stain the biofilm, the supernatant was discarded, the plate was washed and dried, and the biofilm was fixed with 100% methanol, then stained with 1% crystal violet. After discarding the staining solution and washing and drying the plate, the stained biofilm was resuspended in 33% acetic acid. Absorbance at 540 nm was determined on the Infinite M200 plate reader using iControl v1.10.4.
ATP luciferase assays
Bacterial strains were cultured overnight in LB + 100 μM proline or LB alone, standardized to an OD of 1 and subcultured 1:100 with the respective fresh broths for 2 h. In a 96-well plate, an ATP 10-fold serial dilution was used as internal standard control. ATP concentrations were measured as per BacTiter Glo Microbial Cell Viability Assay kit instructions. For extracellular ATP, supernatant was obtained from 100 μl of bacterial culture and mixed with assay reagents for 5 min before reading luminescence on a SpectraMax M2 plate reader. For intracellular ATP, 200 μl of bacterial culture was mixed with 100 μl of ice-cold 1.2 M perchloric acid, incubated on ice for 15 min, and neutralized with 0.72 M KOH and 0.16 M KHCO3.
Extracellular flux analysis
Bacterial cultures were grown overnight and subcultured. The XFe24 sensor cartridge was calibrated as per the manufacturer’s instructions overnight at 37 °C without CO2. About 500 μl of XF base medium supplemented with 2 mM glutamine was added to each well of a Seahorse XF24 well plate and inoculated with 2 × 106 bacteria for a 1 h incubation at 37 °C. The OCR and ECAR were measured on a Seahorse XFe24 Analyzer (Agilent Technologies) using Seahorse Wave Desktop v2.6.0. Proline was added at a final concentration of 100 μM at the three time points indicated.
In silico bioinformatic identification of ccpA and ccpE binding motifs
De novo assembly of the sequenced airway isolates, A2001 and T2015, was performed using Shovill (v. 1.1.0) (https://github.com/tseemann/shovill). Following the completion of assembly, the genomes were identified as S. aureus using Mash (v. 2.1)53. Assembled genomes were annotated using Prokka (v.1.14.6)54. The atopic dermatitis genomes used in comparison—AD2 (GCF_011319675.1) and AD8 (GCF_011319565.1)—were downloaded from RefSeq database in the National Center for Biotechnology Information (NCBI) using ncbi-datasets (v. 14.6.0)55. Basic Local Alignment Search Tool (Blastn) (v. 2.13.0) was then used to confirm the presence of the genes of interest in each of these five genomes56,57,58. After confirming the presence of each of these genes, samtools (v. 1.16.1) was used to extract each gene59. In extracting each gene, 1,000 base pairs upstream from the start of the gene and 1,000 base pairs downstream from the end of the gene were included. Using a custom script, the extracted regions for each genome were evaluated for matches to the cis-acting sequence called catabolite responsive element (cre) sites. Taking into consideration that diverse cre sites have been published39,60,61,62, we used two candidate motifs and allowed up to two mismatches. One of the cre motifs that was searched for was WTGNAANCGNWNNCW, where W represents A or T, and N represents any nucleotide44. The other candidate sequence that was used in the custom script was TGTAAA-Yx-TTTACA, where N represents any base and ranges between 0 and 40 nucleotides45. The custom script reported the sequence that was identified, the motif that the sequence matched, the location of the sequence in relation to the start of the gene of interest and the number of substitutions (that is, 1 substitution, 2 substitutions or exact match). The best match was to be the sequence closest to the start of the gene of interest, with the least number of mismatches. We then checked for nonsynonymous mutations in ccpA and ccpE, in the four clinical isolates. A protein alignment was performed using Snippy (v. 4.6.0) (https://github.com/tseemann/snippy). The gene sequences of ccpA (region 1813146.1814135) and ccpE (region 724466.725332) from strain Newman (Refseq NC_009641.1) were used as the reference for each alignment. SnpEff (v. 4.3.0) was then used to analyse the effect of each single-nucleotide polymorphism on the protein sequence63. The alignments were visualized using snipit (v. 1.1.2), with the respective gene from strain Newman as the reference (https://github.com/aineniamh/snipit).
Bacterial luciferase reporter plasmid
To construct pIMK1-LUX, the Listeria phage integrase vector pIMK64 was digested with SphI/BglII to excise the PSA integrase and replace it with the PCR-amplified (IM1241/IM1242) low copy number pSK41 replicon from pLOW65, yielding pIMK1. Vectors pIMK1 and pPL2lux were then digested with SalI/PstI, and the gel-extracted pIMK1 backbone ligated to the bacterial luciferase operon from pPL2lux. The vector pIMK1-LUX produces exact promoter fusions which can be cloned into the SalI/SwaI digested vector, as described previously64. Promoters for adsA (IM1793/IM1794), citB (IM1785/IM1786) and pckA (IM1787/IM1788) were amplified from S. aureus strain JE2 genomic DNA. The amplimers were digested with SalI, gel extracted and cloned into the pIMK1-LUX double digested vector. Plasmids isolated from IM08B66 were transformed into S. aureus JE2, JE2 ccpA::Tn or JE2 ccpE::Tn and selected on brain heart infusion agar containing 50 µg ml−1 kanamycin. An overnight 5 ml LB culture containing kanamycin was diluted 1:100 in fresh LB containing kanamycin. The culture was then dispensed in triplicate (200 µl) black/clear bottom 96-well plates (PhenoPlate 96-well microplate, Perkinelmer). The plates were sealed with MicroAmp Optical Adhesive Film (Thermo Fisher) and incubated at 37 °C with dual orbital shaking at 300 r.p.m. (Clariostar Plus, BMG). Every 10 min the plate was read at OD600, and light emission (1 s exposure) collected over an 8 h period. Primers are listed in Supplementary Table 1.
Colony PCR
Transposon mutants from the Nebraska Transposon Mutant Library were confirmed via colony PCR. Primers were generated manually and consisted of the first and last 20 bp of the genes of interest. Agarose gels (1%) in 1× Tris–acetate–EDTA buffer were cast after addition of ethidium bromide and allowed to solidify. LB agar + erythromycin plates were streaked with mutants of interest (ccpA::Tn, ccpE::Tn, putP::Tn) and grown overnight. A minimum of 10 single colonies were picked per transposon mutant, and a minimum of 3 for the isogenic WT JE2, both to be resuspended in 25 μl NON-DEPC nuclease-free water. A Go-Taq master mix is prepared (12.5 μl Go Taq, 1 μl forward primer, 1 μl reverse primer, 5 μl emulsified colony, 5.5 μl nuclease-free water) and set at an enzyme/oligonucleotide-dependent thermal cycle (Applied Biosystems) as indicated in the Extended Data. Upon the completion of the cycle, the samples were loaded in the solidified agar gel soaked in an electrophoresis chamber with 1× Tris–acetate–EDTA at room temperature. Gel was run at 140 V for 20 min and imaged with a chemiluminescent detection imager (ProteinSimple). Emulsified colonies of confirmed mutants were streaked and incubated overnight.
In vivo skin infection
Mice were infected intradermally on the shaved back with 1 × 107 c.f.u. of S. aureus Newman WT or adsA mutant. Bacteria were delivered in 100 μl PBS, and sterile PBS was used for an uninfected control (vehicle). Lesion size was measured daily, and mice were killed after 6 days for tissue collection and processing. Skin tissue was collected via 5 mm punch biopsy in the centre of the infected lesion and homogenized through a 40 μm cell strainer. Aliquots of the homogenates were serially diluted and plated on LB agar (±antibiotics if needed) plates to determine c.f.u. counts. Separate aliquots were added to TRK lysis buffer and frozen for later host RNA isolation.
Cloning of the ΔadsAΔputPΔproT USA300 mutant
The putP and proT genes were deleted (between the start and stop codon) from S. aureus LAC or LACΔadsA by allelic exchange, as described in ref. 67. For the deletion substrate, gBlocks for putP (screening primers IM1789/IM1790) and proT (screening primers IM1791/IM1792) were synthesized by Integrated DNA Technologies, cloned into pIMAY-Z by SLiCE and transformed into E. coli IM08B. The genes were deleted sequentially, yielding LACΔproTΔputP and LAC ΔadsAΔproTΔputP and genome validated by Oxford nanopore sequencing68.
BMDM killing assay
Bone marrow was isolated from mouse femurs and tibias spun at 500 g for 6 min and resuspended in ammonium chloride potassium (ACK) lysis buffer for 5 min at room temperature. After lysis buffer inactivation, the cells were plated and differentiated in DMEM with penicillin–streptomycin in the presence of 20 ng ml−1 of murine macrophage colony stimulating factor (M-CSF) for 5 to 6 days (Peprotech). Newly differentiated BMDMs were seeded at 0.3 × 106 cells per ml in DMEM supplemented with 1% FBS. BMDMs were infected at an MOI of 10 for 60 min and washed. Fresh DMEM containing 1% FBS and 50 μg ml−1 gentamicin (Sigma) was then added until the desired time point. The cells were washed, detached using TrypLE Express (Life Technologies), serially diluted and plated on LB agar at 2, 4 and 6 h post infection. Cell viability was determined by trypan blue (Life Technologies) staining assay.
Quantification and statistical analysis
Experiments in this study were not performed in a blinded fashion except for the histopathology qualitative score assessments. Animal assignment to control or experimental groups was randomized. No statistical methods were used to pre-determine sample sizes, but our sample sizes are similar to those reported in previous publications from our laboratory. All analyses and graphs were performed using the GraphPad Prism 9 software. Values defined as n represent individual experiments, and graphed data points represent total biological replicates (please note that some replicates with value 0 will not be visible on logarithmic scale). Values in graphs are shown as average ± s.e.m., and data were assumed to follow a normal distribution. For comparison between average values for more than two groups, we performed one-way analysis of variance (ANOVA) with a multiple posteriori comparison (Tukey’s multiple comparisons or Dunnett’s multiple comparisons). When studying two or more groups along time, data were analysed using two-way ANOVA with a multiple posteriori comparison. Differences between two groups were analysed using a Student t-test with Kolmogorov–Smirnov test. Differences were considered significant when at least a P value under 0.05 (*P < 0.05) was obtained (**P < 0.01; ***P < 0.001; ****P < 0.0001). Statistical details of experiments are indicated in each figure legend. Data distribution was assumed to be normal. No data points were excluded.
Reporting summary
Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.
Data availability
The authors confirm that the data supporting the sequencing findings in this study are available at GEO (accession number GSE268637), Zenodo (https://doi.org/10.5281/zenodo.11450126) (ref. 69) and NCBI (Bio Project PRJNA1115226, accession numbers JBECZQ000000000 and JBECZR000000000). Bacterial strains used in this study are available from the lead contact without restriction. Data, material and resources requests should be directed to and will be fulfilled by the lead contact, A.S.P.
Code availability
No custom code was used or developed during the analysis of the data in this study.
References
Diep, B. A. & Otto, M. Methicillin-resistant Staphylococcus aureus infection is associated with increased mortality. Future Microbiol. 7, 189–191 (2012).
Manara, S. et al. Whole-genome epidemiology, characterisation, and phylogenetic reconstruction of Staphylococcus aureus strains in a paediatric hospital. Genome Med. 10, 82 (2018).
Giulieri, S. G. et al. Niche-specific genome degradation and convergent evolution shaping. Elife 11, 10.7554/eLife.77195 (2022).
Schneewind, O. & Missiakas, D. Sortases, surface proteins, and their roles in Staphylococcus aureus disease and vaccine development. Microbiol. Spectr. https://doi.org/10.1128/microbiolspec.PSIB-0004-2018 (2019).
Raghuram, V. et al. Species-wide phylogenomics of the Staphylococcus aureus. Microbiol. Spectr. 10, e0133421 (2022).
Crosby, H. A. et al. The Staphylococcus aureus global regulator MgrA modulates clumping and virulence by controlling surface protein expression. PLoS Pathog. 12, e1005604 (2016).
Cheung, G. Y. C., Bae, J. S. & Otto, M. Pathogenicity and virulence of Staphylococcus aureus. Virulence 12, 547–569 (2021).
Tam, K. & Torres, V. J. Staphylococcus aureus secreted toxins and extracellular enzymes. Microbiol. Spectr. https://doi.org/10.1128/microbiolspec.GPP3-0039-2018 (2019).
Gjertsson, I., Foster, S. & Tarkowski, A. Polarization of cytokine responses in B- and T-lymphocytes during Staphylococcus aureus infection. Microb. Pathog. 35, 119–124 (2003).
Foster, T. J. Immune evasion by staphylococci. Nat. Rev. Microbiol. 3, 948–958 (2005).
Fowler, V. G. & Proctor, R. A. Where does a Staphylococcus aureus vaccine stand? Clin. Microbiol. Infect. 20, 66–75 (2014).
Wong Fok Lung, T. et al. Adaptive evolution: recent insights on how immune evasion, immunometabolic subversion and host genetics impact vaccine development. Front. Cell. Infect. Microbiol. 12, 1060810 (2022).
Warheit-Niemi, H. I. et al. Fibrotic lung disease inhibits immune responses to staphylococcal pneumonia via impaired neutrophil and macrophage function. JCI Insight 7, 10.1172/jci.insight.152690 (2022).
Howden, B. P. et al. Staphylococcus aureus host interactions and adaptation. Nat. Rev. Microbiol. 21, 380–395 (2023).
Wilke, G. A. & Bubeck Wardenburg, J. Role of a disintegrin and metalloprotease 10 in Staphylococcus aureus alpha-hemolysin-mediated cellular injury. Proc. Natl Acad. Sci. USA 107, 13473–13478 (2010).
Kim, H. K., Thammavongsa, V., Schneewind, O. & Missiakas, D. Recurrent infections and immune evasion strategies of Staphylococcus aureus. Curr. Opin. Microbiol. 15, 92–99 (2012).
Thammavongsa, V., Kim, H. K., Missiakas, D. & Schneewind, O. Staphylococcal manipulation of host immune responses. Nat. Rev. Microbiol. 13, 529–543 (2015).
Wengender, P. A. & Miller, K. J. Identification of a PutP proline permease gene homolog from Staphylococcus aureus by expression cloning of the high-affinity proline transport system in Escherichia coli. Appl. Environ. Microbiol. 61, 252–259 (1995).
Baker, E. H. & Baines, D. L. Airway glucose homeostasis: a new target in the prevention and treatment of pulmonary infection. Chest 153, 507–514 (2018).
Potempa, J., Dubin, A., Korzus, G. & Travis, J. Degradation of elastin by a cysteine proteinase from Staphylococcus aureus. J. Biol. Chem. 263, 2664–2667 (1988).
Windmüller, N. et al. Transcriptional adaptations during long-term persistence of Staphylococcus aureus in the airways of a cystic fibrosis patient. Int. J. Med. Microbiol. 305, 38–46 (2015).
Thammavongsa, V., Schneewind, O. & Missiakas, D. M. Staphylococcus aureus synthesizes adenosine to escape host immune responses. J. Exp. Med. 206, 2417–2427 (2009).
Onursal, C., Dick, E., Angelidis, I., Schiller, H. B. & Staab-Weijnitz, C. A. Collagen biosynthesis, processing, and maturation in lung ageing. Front. Med. 8, 593874 (2021).
Karna, E., Szoka, L., Huynh, T. Y. L. & Palka, J. A. Proline-dependent regulation of collagen metabolism. Cell. Mol. Life Sci. 77, 1911–1918 (2020).
Lehman, M. K. et al. Proline transporters ProT and PutP are required for Staphylococcus aureus infection. PLoS Pathog. 19, e1011098 (2023).
Bruchmann, S., Feltwell, T., Parkhill, J. & Short, F. L. Identifying virulence determinants of multidrug-resistant Klebsiella pneumoniae in Galleria mellonella. Pathog Dis. 79, ftab009 (2021).
Ferretti, E., Horenstein, A. L., Canzonetta, C., Costa, F. & Morandi, F. Canonical and non-canonical adenosinergic pathways. Immunol. Lett. 205, 25–30 (2019).
Li, C. et al. CcpA mediates proline auxotrophy and is required for Staphylococcus aureus pathogenesis. J. Bacteriol. 192, 3883–3892 (2010).
Dufour, A. M., Alvarez, M., Russo, B. & Chizzolini, C. Interleukin-6 and type-I collagen production by systemic sclerosis fibroblasts are differentially regulated by interleukin-17A in the presence of transforming growth factor-beta 1. Front. Immunol. 9, 1865 (2018).
Van Linthout, S., Miteva, K. & Tschöpe, C. Crosstalk between fibroblasts and inflammatory cells. Cardiovasc. Res. 102, 258–269 (2014).
Kendall, R. T. & Feghali-Bostwick, C. A. Fibroblasts in fibrosis: novel roles and mediators. Front. Pharmacol. 5, 123 (2014).
Jain, P. P. et al. Halofuginone, a promising drug for treatment of pulmonary hypertension. Br. J. Pharmacol. 178, 3373–3394 (2021).
Nagler, A. et al. Reduction in pulmonary fibrosis in vivo by halofuginone. Am. J. Respir. Crit. Care Med. 154, 1082–1086 (1996).
Popov, Y. et al. Halofuginone induces matrix metalloproteinases in rat hepatic stellate cells via activation of p38 and NFkappaB. J. Biol. Chem. 281, 15090–15098 (2006).
Halsey, C. R. et al. Amino acid catabolism in Staphylococcus aureus and the function of carbon catabolite repression. mBio. 8, e01434–16 (2017).
Deutscher, J., Küster, E., Bergstedt, U., Charrier, V. & Hillen, W. Protein kinase-dependent HPr/CcpA interaction links glycolytic activity to carbon catabolite repression in gram-positive bacteria. Mol. Microbiol. 15, 1049–1053 (1995).
Pätzold, L. et al. Impact of the histidine-containing phosphocarrier protein HPr on carbon metabolism and virulence in. Microorganisms 9, 10.3390/microorganisms9030466 (2021).
Sadykov, M. R. et al. CidR and CcpA synergistically regulate Staphylococcus aureus. J. Bacteriol. 201, 10.1128/JB.00371-19 (2019).
Martin-Verstraete, I., Stülke, J., Klier, A. & Rapoport, G. Two different mechanisms mediate catabolite repression of the Bacillus subtilis levanase operon. J. Bacteriol. 177, 6919–6927 (1995).
Ramström, H. et al. Properties and regulation of the bifunctional enzyme HPr kinase/phosphatase in Bacillus subtilis. J. Biol. Chem. 278, 1174–1185 (2003).
Seidl, K. et al. Staphylococcus aureus CcpA affects virulence determinant production and antibiotic resistance. Antimicrob. Agents Chemother. 50, 1183–1194 (2006).
Bischoff, M. et al. CcpA affects infectivity of Staphylococcus aureus in an hyperglycemic environment. Front. Cell. Infect. Microbiol. 7, 172 (2017).
Hartmann, T. et al. The catabolite control protein E (CcpE) affects virulence determinant production and pathogenesis of Staphylococcus aureus. J. Biol. Chem. 289, 29701–29711 (2014).
Miwa, Y., Nakata, A., Ogiwara, A., Yamamoto, M. & Fujita, Y. Evaluation and characterization of catabolite-responsive elements (cre) of Bacillus subtilis. Nucleic Acids Res. 28, 1206–1210 (2000).
Yang, Y. et al. A flexible binding site architecture provides new insights into CcpA global regulation in Gram-positive bacteria. mBio 8, 10.1128/mBio.02004-16 (2017).
Liao, X. et al. Regulation of DNA-binding activity of the Staphylococcus aureus catabolite control protein A by copper (II)-mediated oxidation. J. Biol. Chem. 298, 101587 (2022).
Ding, Y. et al. Metabolic sensor governing bacterial virulence in Staphylococcus aureus. Proc. Natl Acad. Sci. USA 111, E4981–4990 (2014).
Thammavongsa, V., Schneewind, O. & Missiakas, D. M. Enzymatic properties of Staphylococcus aureus adenosine synthase (AdsA). BMC Biochem. 12, 56 (2011).
Thammavongsa, V., Kern, J. W., Missiakas, D. M. & Schneewind, O. Staphylococcus aureus synthesizes adenosine to escape host immune responses. J. Exp. Med. 206, 2417–2427 (2009).
Urso, A. & Prince, A. Anti-inflammatory metabolites in the pathogenesis of bacterial infection. Front. Cell. Infect. Microbiol. 12, 925746 (2022).
Schuster, C. F., Howard, S. A. & Gründling, A. Use of the counter selectable marker PheS* for genome engineering in Staphylococcus aureus. Microbiology 165, 572–584 (2019).
Partridge, M. A. & Marcantonio, E. E. Initiation of attachment and generation of mature focal adhesions by integrin-containing filopodia in cell spreading. Mol. Biol. Cell 17, 4237–4248 (2006).
Ondov, B. D. et al. Mash: fast genome and metagenome distance estimation using MinHash. Genome Biol. 17, 132 (2016).
Seemann, T. Prokka: rapid prokaryotic genome annotation. Bioinformatics 30, 2068–2069 (2014).
Sayers, E. W. et al. Database resources of the National Center for Biotechnology Information. Nucleic Acids Res. 52, D33–D43 (2024).
Camacho, C. et al. BLAST+: architecture and applications. BMC Bioinformatics 10, 421 (2009).
Altschul, S. F. & Lipman, D. J. Protein database searches for multiple alignments. Proc. Natl Acad. Sci. USA 87, 5509–5513 (1990).
Liu, J., Faeder, J. R. & Camacho, C. J. Toward a quantitative theory of intrinsically disordered proteins and their function. Proc. Natl Acad. Sci. USA 106, 19819–19823 (2009).
Li, H. et al. The Sequence Alignment/Map format and SAMtools. Bioinformatics 25, 2078–2079 (2009).
Seidl, K. et al. Effect of a glucose impulse on the CcpA regulon in Staphylococcus aureus. BMC Microbiol. 9, 95 (2009).
Stülke, J. & Hillen, W. Regulation of carbon catabolism in Bacillus species. Annu. Rev. Microbiol. 54, 849–880 (2000).
Zalieckas, J. M., Wray, L. V. & Fisher, S. H. Expression of the Bacillus subtilis acsA gene: position and sequence context affect cre-mediated carbon catabolite repression. J. Bacteriol. 180, 6649–6654 (1998).
Cingolani, P. et al. A program for annotating and predicting the effects of single nucleotide polymorphisms, SnpEff: SNPs in the genome of Drosophila melanogaster strain w1118; iso-2; iso-3. Fly 6, 80–92 (2012).
Bron, P. A., Monk, I. R., Corr, S. C., Hill, C. & Gahan, C. G. Novel luciferase reporter system for in vitro and organ-specific monitoring of differential gene expression in Listeria monocytogenes. Appl. Environ. Microbiol. 72, 2876–2884 (2006).
Liew, A. T. F. et al. A simple plasmid-based system that allows rapid generation of tightly controlled gene expression in Staphylococcus aureus. Microbiology 157, 666–676 (2011).
Monk, I. R., Tree, J. J., Howden, B. P., Stinear, T. P. & Foster, T. J. Complete bypass of restriction systems for major Staphylococcus aureus lineages. mBio 6, e00308–00315 (2015).
Monk, I. R. & Stinear, T. P. From cloning to mutant in 5 days: rapid allelic exchange in Staphylococcus aureus. Access Microbiol. https://doi.org/10.1099/acmi.0.000193 (2021).
Wick, R. R., Judd, L. M., Monk, I. R., Seemann, T. & Stinear, T. P. Improved genome sequence of Australian methicillin-resistant Staphylococcus aureus strain JKD6159. Microbiol. Resour. Announc. https://doi.org/10.1128/mra.01129-22 (2023).
Theiller, E. & Moustafa, A. Staphylococcus aureus adapts to exploit collagen-derived proline during chronic infection. zenodo https://doi.org/10.5281/zenodo.11450126 (2024).
Acknowledgements
We thank L. Hoffman (University of Seattle), B. Kahl (Münster University), J. Koff (Yale University) and D. Missiakis (University of Chicago) for providing strains critical to our studies. This work was supported by NIH grant 1R35HL135800 to A.S.P. and National Science Foundation grant DGE-2036197 to A.U. This study was also supported by NIH grant S10RR027050 to the Columbia Center for Translational Immunology Flow Cytometry Core and NIH (National Cancer Institute) grant P30CA013696 to the Molecular Pathology Shared Resource of the Herbert Irving Comprehensive Cancer Center at Columbia University. The Calgary Metabolomics Research Facility is supported by the International Microbiome Centre and the Canada Foundation for Innovation (CFI-JELF 34986). I.A.L. is supported by an Alberta Innovates Translational Health Chair. BioRender platform was used for the creation of some schematics in the manuscript.
Author information
Authors and Affiliations
Contributions
A.U. conceived the project, designed and performed the experiments and wrote the paper. I.R.M., B.P.H. and T.P.S. constructed bacterial strains and reporter constructs; S.P. and V.T. made USA300 mutants; Y.-T.C., G.L. and J.B. performed bacterial metabolic experiments; C.P. and N.V.D. contributed to the fibroblast experiments; T.W.F.L. contributed to data analyses; E.M.T. and A.M.M. performed the bioinformatic studies of clinical strains; I.A.L. provided the metabolomic data; S.U.B. analysed histopathologic data. A.S.P. proposed the central hypothesis, contributed to the experimental design, supervised the project and wrote the paper.
Corresponding author
Ethics declarations
Competing interests
The authors declare no competing interests.
Peer review
Peer review information
Nature Microbiology thanks the anonymous reviewers for their contribution to the peer review of this work.
Additional information
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Extended data
Extended Data Fig. 1 Relative single carbon source consumption by S. aureus strains.
Utilization of carbon sources by CF clinical isolates (2001, 2015, CF1, CF9) and Newman adsA mutants shown as fold-change respective to Newman WT. b, Growth of S. aureus WT Newman and ΔadsA in CDM b, ± 100 μM hydroxyproline (HyP), c, ± 100 μM glycine (G), and c:adsA in d, CDM and e, CDM + proline. f, g, Newman and LAC plasmid stability in LB ± antibiotics. Mean ± SEM, n = 3. Significance determined by *P < 0.05; **P < 0.01; ***P < 0.001 ****P < 0.0001, (a) two-tailed t- Student with Kolmogorov-Smirnov test; (b-e) One-Way ANOVA with Tukey’s Multiple Comparisons; (f, g) Two- Way ANOVA with Dunnett’s Multiple Comparisons.
Extended Data Fig. 2 Impact of host generated adenosine on WT and ΔadsA mutant infection.
Pulmonary infection at 72 hours following intranasal inoculation with S. aureus WT Newman, ΔadsA or PBS. a, Untargeted metabolomics shown respective to PBS control, b, T regulatory cell (Tregs) count and c, relative expression of surface CD39 and CD73 ectonucleotidases on various immune cells. Pulmonary infection at 72 hours following intranasal inoculation of GFP FoxP3DTR mice with S. aureus WT Newman or ΔadsA and in vivo diphtheria toxin (DT)-mediated Treg depletion. d, Total lung Tregs e, bacterial burden f, immune cell populations and g, CD39 and CD73 expression upon treatment with DT or vehicle. Pulmonary infection at 72 hours following intranasal inoculation of CD73-KO mice with S. aureus WT Newman or ΔadsA. h, Bacterial burden in WT and CD73-KO mouse lungs. i, Weight loss, j-o, lung immune cells, and p, untargeted metabolomics on BAL fluid respective to PBS. Mean ± SEM, (a, b) n = 3; (c-e, h-p) n = 3; (f) n = 4. Significance determined by *P < 0.05; **P < 0.01; ***P < 0.001 ****P < 0.0001; (a-h. j-p) One- Way ANOVA with Tukey’s Multiple Comparisons; (f). Two-Way ANOVA with Dunnett’s Multiple Comparisons.
Extended Data Fig. 3 Consumption of components of the citrulline superpathway by ∆adsA and CF clinical isolates.
Growth of a-e, WT, ΔadsA, f-j, A2001 and T2015 in a, f, complete defined medium (CDM) lacking b, g, proline, c, h, histidine, d, i, glutamate or e, j, arginine. Mean ± SEM from (a-j) n = 4.
Extended Data Fig. 4 Induction of collagen synthesis by WT and ∆adsA infection.
Induction of host responses by pulmonary infection with S. aureus WT or ΔadsA strains at 72 h. a, BAL fluid matrix metalloprotease and b, pro-inflammatory cytokines quantified by ELISA and plotted with respect to PBS. Primary murine fibroblasts were freshly isolated and cultured for infection with S. aureus WT or ΔadsA strains and treatment with collagen inhibitor halofuginone (HF) ex vivo. c, Representative lung sections stained with Masson’s Trichrome (purple) and hematoxylin and eosin staining (pink) from a PBS control, WT Newman and ΔadsA infection at 72 h are shown. Scale bar= 50 μM; m = 40x. d, Representative confocal images of primary fibroblasts (PF) stained with vimentin and DAPI to confirm population purity (scale bar 40μm). e, Cell count of live PF with trypan blue at each infection timepoint in vitro. S. aureus growth in f, g, LB only and h, i, CDM + P supplemented with HF or vehicle. In vivo intranasal 72- hour lung infection with S. aureus WT or ΔadsA strains in mice treated with HF or vehicle. j, BAL and k, lung total cells from infected mice treated with HF or vehicle. Mean ± SEM from (a-b) n = 3; (e) n = 4; (f-i) n = 6; (j-k) n = 6. (a-b, d-j) Two-Way ANOVA with Dunnet’s Multiple Comparison.
Extended Data Fig. 5 Impact of adsA and CCR on skin infection.
a, RNA transcripts from S. aureus Atopic Dermatitis (AD) clinical isolates. b, Carbon utilization assays shown as fold-change increase w.r.t. WT. c, Growth curves in CDM ± 100 μM proline. d, Bacterial burden and e, skin lesion size at 6 days post-infection from a 5 mm punch biopsy of skin infected with S. aureus Newman. f, Host RNA transcripts from 5 mm punch biopsy of skin infected with S. aureus Newman. g-h, Putative cre binding sites in S. aureus atopic dermatitis (AD) isolates AD2 and AD8 for ccpA in silico (black dashes, mismatches; red dots, matches) using binding sequences Sequence 1 and Sequence 2 (red). i, Bacterial burden at 6 days post- infection from a 5 mm punch biopsy of skin infected with WT JE2, ccpA::Tn, ccpE::Tn, putP::Tn. Data presented as mean ± SEM, (a-c) n = 3, (d,e) n = 4, (i) n = 3. Significance determined by *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001, (a,c,e) Two-Way ANOVA with Dunnett’s Multiple Comparisons, (b,d) two-tailed t- Student with Kolmogrov-Smirnov test (f,i) One-Way. ANOVA with Tukey’s Multiple Comparisons.
Extended Data Fig. 6 FACS Gating strategies.
a, Neutrophils, alveolar macrophages and monocytes; b, lymphocytes and T regulatory cells; and c, lung fibroblasts.
Supplementary information
Supplementary Information
Legends for Extended Data Figs. 1–6 and Supplementary Tables 1–3.
Supplementary Tables
Supplementary Tables 1–6.
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.
About this article
Cite this article
Urso, A., Monk, I.R., Cheng, YT. et al. Staphylococcus aureus adapts to exploit collagen-derived proline during chronic infection. Nat Microbiol 9, 2506–2521 (2024). https://doi.org/10.1038/s41564-024-01769-9
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1038/s41564-024-01769-9
