Research Article

Staphylococcus aureus Nonribosomal Peptide Secondary Metabolites Regulate Virulence

See allHide authors and affiliations

Science  16 Jul 2010:
Vol. 329, Issue 5989, pp. 294-296
DOI: 10.1126/science.1188888

This article has a correction. Please see:


Staphylococcus aureus is a major human pathogen that is resistant to numerous antibiotics in clinical use. We found two nonribosomal peptide secondary metabolites—the aureusimines, made by S. aureus—that are not antibiotics, but function as regulators of virulence factor expression and are necessary for productive infections. In vivo mouse models of bacteremia showed that strains of S. aureus unable to produce aureusimines were attenuated and/or cleared from major organs, including the spleen, liver, and heart. Targeting aureusimine synthesis may offer novel leads for anti-infective drugs.

Staphylococcus aureus is a human pathogen commonly causing hospital and community-acquired infectious diseases (1). It has an array of virulence factors, including surface proteins responsible for adhesion and invasion of host tissues (e.g., fibrinogen and fibronectin-binding proteins), exoproteins responsible for immune evasion (e.g., chemotaxis-inhibitory protein), and numerous hemolytic and pore-forming toxins (e.g., hemolysins, leukocidins, and enterotoxins) (24). For successful infection, a coordinated release of virulence factors is necessary, and redundancies exist, such that, if one factor is ablated, a productive infection can still ensue. Early research by Novick and colleagues identified an accessory gene regulator (agr) that controls several virulence factors (5). Expression of the agr locus is positively regulated by the agr pheromone, a ribosomally encoded secondary metabolite (6). Subsequent genomic sequencing has revealed that homologs of the agr pheromones exist in several Gram-positive cocci, many of which are not pathogenic (711). Although referred to as the “master” regulator of S. aureus virulence, expression of agr is not always detected in vivo, and agr-deficient clinical isolates are known, which raises the possibility that other small molecules factor prominently in the regulation of virulence factor expression (12).

A major class of bacterial secondary metabolites comprises the nonribosomal peptides, which are produced, in microorganisms, by multifunctional enzyme assembly lines known as nonribosomal peptide synthetases (NRPSs) (13). Antibiotics are the best known nonribosomal peptides produced by soil-dwelling microbes, which use them as weapons and for cell-cell communication (14). Penicillin, for example, is not constructed ribosomally but is dependent on an NRPS that uses valine, cysteine, and α-aminoadipic acid precursors (15). Although penicillin was the first nonribosomal peptide used for S. aureus infections, S. aureus itself has not previously been shown to construct nonribosomal peptides.

Cryptic nonribosomal peptide assembly in Staphylococcus. An NRPS uses adenylation (A) domains and adenosine triphosphate (ATP) to activate adenosine monophosphate esters of selected amino acids and delivers them to posttranslationally modified (phosphopantetheine) NRPS thiolation (T) domains (13). Condensation reactions of the T domain–tethered amino acids produce growing peptide chains, which are released by thioesterase or reductase (Re) domains at the C terminus of the NRPS, the latter as peptide aldehydes or alcohols that frequently undergo intramolecular cyclization reactions (13, 16). Details of NRPS-catalyzed reactions, structural assignments of several NRPS domains, and assembly rules for nonribosomal peptide on NRPSs are reasonably well known. Genes encoding for a given nonribosomal peptide are also clustered. Coopting these genetic and/or biochemical parameters with microbial genomic sequencing has created a link between gene and small-molecule prediction, which has assisted in the discovery of unidentified, or “cryptic,” nonribosomal peptides, a process referred to as secondary metabolite genome mining (17, 18).

We used a genome-mining approach to predict nonribosomal peptides that are exclusive and highly conserved within S. aureus (19). Scanning in excess of 50 S. aureus sequenced genomes led to the identification of a universally conserved (average of 97% identical and 97% similar), yet undescribed, NRPS gene cluster (annotated as a gramicidin synthetase or hypothetical protein) (Fig. 1A and fig. S1) (20, 21). This cluster contains an NRPS gene (7.17 kb) that takes up 0.25% of the S. aureus genome. An ortholog is present in other staphylococci pathogenic to humans, including Staphylococcus epidermidis (53% identical and 71% similar), Staphylococcus capitis (53% identical and 70% similar), and Staphylococcus lugdunensis (53% identical and 70% similar), but is absent in other staphylococci or Gram-positive cocci (figs. S1 and S2) (21, 22). Buoyed by the association of this NRPS with staphylococci pathogenic to humans, we predicted the structure of the encoded nonribosomal peptide.

Fig. 1

Identification of a cryptic NRPS biosynthetic gene cluster within S. aureus. (A) Genetic loci of S. aureus Newman containing the NRPS gene. The NRPS locus is found in all sequenced S. aureus genomes. The NRPS cluster contains two open reading frames: ausA (the NRPS gene) and immediately downstream of it ausB (phosphopantetheinyl transferase). ausB encodes the enzyme (AusB) predicted to posttranslationally modify AusA with a 4′-phosphopantetheine prosthetic group. (B) S. aureus NRPS is a dimodular nonribosomal peptide assembly line encoding a putative cyclic dipeptide. Domains A, C, T, and Re within the S. aureus NRPS (AusA) are shown as round spheres shaded in yellow. Curved blue lines originate from the T domain and indicate the phosphopantetheinyl arm that is predicted to be delivered via action of AusB. Amino acid substrates (valine and tyrosine) were predicted according to established NRPS codes (fig. S2) (17, 21). Release of a linear valine-tyrosine dipeptide aldehyde and the predicted nonribosomal peptide structure are shown. (C) Identification of S. aureus nonribosomal peptides. Structures of aureusimine A and aureusimine B (phevalin) were determined by mass spectrometry and NMR experiments (figs. S4 to S7). (D) Liquid chromatographic separations (HPLC chromatograms) of organic extracts of S. aureus Newman and S. aureus Newman ΔausA (19). Aureusimine A (peak 1) and aureusimine B (phevalin) (peak 2) are present within extracts of S. aureus Newman but absent in extracts of S. aureus Newman ΔausA strain. ERM, erythromycin.

The S. aureus NRPS (2389 amino acids) is a dimodular NRPS with two adenylation (A) domains having strictly conserved NRPS codes for valine and tyrosine for all staphylococci containing the NRPS (Fig. 1B and fig. S3). A Re domain is found at the C terminus of all the S. aureus NRPSs and probably releases the valine-tyrosine dipeptide as an aldehyde (23). The proposed linear dipeptide aldehyde is likely to assume a cyclic imine conformation (predicted mass of 262.17), promoted by nucleophilic attack of the aldehyde by the α-amine of valine (Fig. 1B) (16, 23).

Isolation of tyrosine-valine dipeptides. To isolate the cryptic S. aureus nonribosomal peptide, we collected organic solvent extracts of S. aureus culture broths and subjected them to high-performance liquid chromatography (HPLC) and mass spectroscopy analysis with mass-spectral filtering software to identify products within the range of the predicted dipeptide mass (19) (fig. S4). Two peaks were obtained, one providing a nearly exact match and the second with a slightly later retention time and differing by 16 mass units (fig. S4). Both molecules had a common absorbance spectrum, indicating that the two were congeners (produced in a ~3:1 ratio) or sufficiently similar to suggest that they stem from a common NRPS pathway (fig. S4). One-dimensional and two-dimensional nuclear magnetic resonance (NMR) experiments (figs. S5 to S7) provided the structures of both metabolites (Fig. 1C). The most abundant matched the prediction of the cyclic valine-tyrosine dipeptide bearing a pyrazinone core (Fig. 1C), and we called it aureusimine A. The second molecule (aureusimine B) had a phenylalanine in place of the tyrosine with a structure that matched a previously identified cyclic dipeptide (phevalin) from Streptomyces sp. SC433 (24). Production of two related nonribosomal peptides from a single NRPS is commonplace and is consistent with the second A domain’s incorporation of both tyrosine (aureusimine A) and phenylalanine (aureusimine B or phevalin).

To verify that the aureusimines are synthesized by the S. aureus NRPS (encoded by the gene we have named ausA), an allelic replacement was used to replace ausA with an erythromycin-resistance cassette (19). Culture broths of the resulting ΔausA S. aureus strain were devoid of aureusimine A and B (Fig. 1D). We next compared the growth of the ausA deletion strain with that of the wild type and found that the aureusimines are not necessary for growth and that the ausA deletion strains actually grew better than the wild type (fig. S8).

Microarray analysis of virulence expression. Our discovery of a nonribosomal peptide unique to S. aureus raises the possibility for its role as a regulator of S. aureus virulence factor expression. To evaluate the impact of aureusimine on virulence gene expression, we conducted global microarray analysis. Both S. aureus Newman and Newman ΔausA overnight cultures were diluted 1:100 in tryptic soy broth and grown until early exponential [absorbance at 600 nm (A600) = 0.3] and late exponential phase (A600 = 1.2). mRNA was isolated from each strain and used for microarray experiments (19). In three separate experiments, primary metabolic genes were largely unchanged in the ΔausA strain, a result consistent with growth studies (fig. S6). However, in comparison with its isogenic parent, the ausA mutant displayed significant differences in expression of a large number of virulence genes, including genes encoding immunomodulatory proteins, host cell adhesins, chemotaxis-blocking proteins, and host-targeted lytic proteins and cytotoxins (Fig. 2A) (tables S2 to S5). For example, genes encoding chemotaxis-inhibiting protein and formyl peptide receptor–like 1inhibitory proteins, important for S. aureus immune evasion, are massively up-regulated by aureusimine production, >100 times (145.9) and >50 times (73.5), respectively (3, 25). S. aureus adhesion molecules, such as fibrinogen-binding protein (Efb) and fibronectin-binding protein A (FnbA), are necessary for endothelial cell invasion and endocarditis and are also up-regulated 187.5- and 75.2-fold, respectively (2628). Genes encoding for hemolysins were also significantly up-regulated in strains producing aureusimines. As an illustration showing that the transcriptional profiling corresponds to phenotypic alterations, the blood-lysing capacity of both wild-type and mutant strains was compared on blood agar plates (Fig. 2B). S. aureus Newman colonies lyse red blood cells, whereas little to no clearing or lysis was observed by ΔausA colonies. The hemolytic property of S. aureus could be restored to the ΔausA strains by the addition of aureusimines A and B (100 μg/ml) into the blood agar plates (Fig. 2B).

Fig. 2

Gene regulation by the aureusimines. (A) Differential gene expression caused by the presence of aureusimines A and B in S. aureus in early and late exponential phase growth. Results are presented as mean fold up-regulation (shades of red) and down-regulation (shades of blue) in three separate experiments (see scale bar). The complete microarray results of genes regulated by the aureusimines can be seen in tables S2 to S5. (B) Aureusimines induce hemolysis. (Left) S. aureus Newman wild-type and (center) S. aureus Newman ΔausA were grown on 5% sheep blood agar; (right) S. aureus Newman ΔausA was grown on 5% sheep blood agar supplemented with 100 μg/mL aureusimines A and B. Zones of clearance around colonies indicate hemolysis.

Role of aureusimines in vivo. To gain further insight into the role that aureusimines play in the infectivity and virulence of S. aureus, groups of BALB/c mice were injected intravenously with either S. aureus strain Newman or the isogenic ausA mutant (19). Over the course of 4 days, mice infected with wild-type S. aureus Newman lost, on average, 22% of their original weight, consistent with a productive S. aureus infection (Fig. 3A). In contrast, mice infected with the S. aureus ΔausA deletion strain lost, on average, only 7.5% of their original weight (Fig. 3A). The weight change data for the two groups of mice are significantly different as determined by the Student’s t test (P < 0.001). Organs from both groups of mice were removed and homogenized, and the resulting suspensions were surveyed for viable S. aureus colony-forming units (CFUs) (19). In the group infected with wild-type bacteria, CFUs were high in all organs examined (Fig. 3, B to E). Although CFUs in samples recovered from kidneys of mice infected with the ausA deletion strain were comparable to those from kidneys of mice infected with wild-type S. aureus, CFUs obtained from the hearts, spleens, and livers of mice infected with the ausA mutant were all significantly less (P < 0.01) than those from the respective organs of mice infected with wild-type bacteria, as determined by the Student’s t test (Fig. 3, B to E). In fact, we could not recover detectable CFUs from the hearts of mice infected with the ausA mutant (Fig. 3E).

Fig. 3

(A) Weight change (4 days after infection) for mice infected with S. aureus Newman (filled circles) or S. aureus Newman ΔausA (open circles). Solid bars represent average weight change. (B to E) CFUs obtained from kidneys, livers, spleens, and hearts 4 days after infection. Solid bars represent the average log10 CFUs for the group.

The aureusimines are previously unidentified nonribosomal peptide secondary metabolites that are integral to the ability of S. aureus to act as an infectious agent. Discovery of the aureusimines’ control over a wide range of S. aureus virulence factors presents opportunities for novel anti-infective strategies. Unlike many other nonribosomal peptide secondary metabolites produced by soil microbes, such as penicillin (29), they do not appear to act as antibiotics. However, the original isolation of phevalin (aureusimines B) from a soil-dwelling actinomycete suggests a possible origin of the aureusimine NRPS (24). For S. aureus, acquisition of the aureusimine NRPS biosynthetic machinery was a defining moment.

Supporting Online Material

Materials and Methods

Figs. S1 to S8

Tables S1 to S5


References and Notes

  1. Materials and methods are available as supporting material on Science Online.
  2. This work was supported by generous gifts from McMaster University (N.A.M.), Canadian Institutes of Health Research (MOP-38002 to D.E.H.), and NIH (RA107380 to P.M.D.). For animal infections, all protocols were reviewed and approved by the University of Western Ontario’s Animal Use Subcommittee, a subcommittee of the University Council on Animal Care. Gene microarray data have been deposited in National Center for Biotechnology Information, NIH (accession no. GSE21373). N.A.M. declares no competing financial interests but does declare patent applications related to this work (U.S. patent applications, no. 61/183,152 and no. 61/306,239).
View Abstract

Stay Connected to Science

Navigate This Article