Commensal Oral Rothia mucilaginosa Produces Enterobactin, a Metal-Chelating Siderophore

The communication language of the human oral microbiota is vastly underexplored. However, a few studies have shown that specialized small molecules encoded by BGCs have critical roles such as in colonization resistance against pathogens and quorum sensing. Here, by using a genome mining approach in combination with compound screening of growth cultures, we identified that the commensal oral community member R. mucilaginosa harbors a catecholate-siderophore BGC, which is responsible for the biosynthesis of enterobactin. The iron-scavenging role of enterobactin is known to have positive effects on the host’s iron pool and negative effects on host immune function; however, its role in oral health remains unexplored. R. mucilaginosa was previously identified as an abundant community member in cystic fibrosis, where bacterial iron cycling plays a major role in virulence development. With respect to iron’s broad biological importance, iron-chelating enterobactin may explain R. mucilaginosa’s colonization success in both health and disease.

IMPORTANCE The communication language of the human oral microbiota is vastly underexplored. However, a few studies have shown that specialized small molecules encoded by BGCs have critical roles such as in colonization resistance against pathogens and quorum sensing. Here, by using a genome mining approach in combination with compound screening of growth cultures, we identified that the commensal oral community member R. mucilaginosa harbors a catecholate-siderophore BGC, which is responsible for the biosynthesis of enterobactin. The iron-scavenging role of enterobactin is known to have positive effects on the host's iron pool and negative effects on host immune function; however, its role in oral health remains unexplored. R. mucilaginosa was previously identified as an abundant community member in cystic fibrosis, where bacterial iron cycling plays a major role in virulence development. With respect to iron's broad biological importance, ironchelating enterobactin may explain R. mucilaginosa's colonization success in both health and disease. growth of methicillin-resistant strains of Staphylococcus aureus (MRSA). S. aureus is not currently considered a pathogen in oral disease; however, this view is changing as more cases of MRSA and methicillin-susceptible S. aureus (MSSA) have been reported in infections of the oral cavity (23)(24)(25). Our study also establishes that enterobactin chelates both zinc and magnesium ions but with less affinity than Fe(III). Conclusively, we describe a new key ecological mechanism that involves the metal-chelating siderophore enterobactin, which not only supports survival and growth of the underexplored yet frequently occurring oral R. mucilaginosa but also reduces growth of both cariogenic and multidrug-resistant pathogens.
liquid minimal medium M9 cultures supplemented with either 100 mM sucrose or glucose during aerobic incubation at 37°C (Fig. 1A). However, when incubated with other carbon sources (glycerol, lactose, arabinose, or galactose), growth was reduced (Fig. 1A). To explore if a siderophore was produced under any of the above culture conditions, we screened liquid growth extracts using two different assays: a hydroxamate assay that targets carboxylate siderophores (31) and Arnow's assay, which targets catecholate siderophores (32). Arnow's assay showed clear colorimetric changes as the normally colorless M9 medium turned ruby red in cultures incubated with sucrose and glycerol, while the hydroxamate assay showed no color change. The presence of a catecholate siderophore in these cultures was also confirmed using absorbance measurements at 500 nm, which are known to capture catecholate derivatives (33). Not only did these results reveal that R. mucilaginosa ATCC 25296 can produce a catechol siderophore, but they also demonstrated that the antiSMASH program could accurately predict the correct core building blocks (serine and dihydroxybenzoic acid). To facilitate compound isolation and purification, we further explored if glycerol, which is known to elicit secondary metabolite production in other actinobacteria (34), could increase siderophore yields. This was indeed the case as enterobactin production increased significantly in the glycerol-amended cultures (absorbance at 500 nm, ϳ0.25) (Fig. 1B). However, under the same conditions, growth decreased (Fig. 1A). We also tested if liquid cultures of Rothia dentocariosa M567, which also harbors a cat-sid BGC (Fig. S2), can produce a siderophore when subjected to glycerol cultivation (Fig. 1B). Production was observed for this species as well (absorbance at 500 nm, ϳ0.15) but not to the same extent as for R. mucilaginosa (Fig. 1B).
Detailed characterization of R. mucilaginosa's catechol siderophore and cat-sid BGC. Using Phyre2, a protein structure prediction software program (35), only one gene within the enterobactin BGC from R. mucilaginosa ATCC 25296 showed amino acid sequence homology to the BGC that encodes enterobactin in the Escherichia coli JM109 genome (i.e., amino acid location 51 to 463 in the NRPS gene [ Fig. S3]). The overall low structural BGC homology as well as the low amino acid sequence homology between the NRPS genes in R. mucilaginosa and E. coli BGCs illustrates that the local gene environments surrounding the core biosynthetic NRPS genes have diversified greatly between different taxonomic groups of bacteria and have no influence on the final compound structure.
collision-induced dissociation (CID) fragmentation of the parent mass, and comparative analysis with all previously characterized compound spectra available through the Global Natural Products Social Molecular Networking (GNPS) library (36), it became clear that the ion fragments matched the well-characterized catecholate siderophore enterobactin (m/z 670.164, GNPS gold standard spectrum CCMSLIB00005435752, GNPS cosine score of 0.71 to 0.78) ( Fig. 2A). To verify our findings, we performed additional LC-MS/MS on purified extracts from thin-layer chromatography, which again resulted in the identification of enterobactin or a close homologue using the GNPS infrastructure, this time in positive ionization mode (Fig. 2B). Further purification of this compound using high-performance liquid chromatography (HPLC) showed a well-separated peak that tested positive in the Arnow assay (Fig. 3). This peak was eluted and analyzed by 1 H nuclear magnetic resonance (NMR) to confirm that R. mucilaginosa ATCC 25296 produces enterobactin (Table S1). All observed chemical shift values in Table S1 (obtained from R. mucilaginosa's enterobactin) were also reported previously in an infrared (IR) and NMR spectroscopic study of the archetypal enterobactin molecule (37). Proton and two-dimensional (2D) NMR spectra (heteronuclear single quantum correlation [HSQC], heteronuclear multiple-bond coherence [HMBC], and proton correlation spectroscopy [COSY]) for the R. mucilaginosa-derived compound confirmed its identity (Fig. S4).
Enterobactin activity screening using cocultivation assays. Interactions between R. mucilaginosa ATCC 25296 or the purified enterobactin compound and other bacterial species were studied by employing cocultivation agar assays and liquid growth assays. On agar plates, growth of R. mucilaginosa was established prior to spotting the challenging species. Interactions were initially studied on both rich (brain heart infusion [BHI]) and minimal M9 agar media to investigate under which conditions R. mucilaginosa could modulate growth of the competitor test strain, potentially via iron competition (visualized as clearing zones surrounding R. mucilaginosa). In the liquid monocultures, the purified siderophore was added at the same time as the cultures were seeded with the test strain. We found that while growing on minimal M9 agar, supplemented with sucrose (100 mM) during plate interaction studies, R. mucilaginosa inhibited growth of Actinomyces timonensis DSM 23838 (Fig. S5A) and reduced pigment production in four strains of pathogenic Staphylococcus aureus in the presence of catalase ( Fig. 4 and Fig. S6). We added catalase to the agar plates to prevent growth inhibition by eventual reactive oxygen species (ROS) produced by R. mucilaginosa. The results suggest that inhibition of pigment production is not due to ROS but actually due to enterobactin. To further elucidate this interaction, we conducted monococulture experiments on agar plates where we added the purified enterobactin to S. aureus agar plates (Fig. 5). With catalase added, statistically significant differences in pigmentation were observed for all S. aureus strains amended with enterobactin, including MRSA  strains TCH70/MRSA and NR10129 (P Ͻ 0.05, two-tailed t test) (Fig. 5), confirming an important role of enterobactin in the inhibition of S. aureus virulence and growth (38,39). For all the tested oral commensal and pathogenic Streptococcus species, no growth inhibition was observed on agar plates. However, changes in growth were confirmed for some of the species when adding the purified enterobactin compound to liquid cultures (with or without catalase) (Fig. 6). Of the pathogenic Streptococcus strains tested in liquid growth cultures amended with enterobactin and catalase, we observed that growth of both S. mutans UA159 and S. mutans B04Sm5 was significantly reduced (P Ͻ 0.05, t test of averaged mean values) ( Fig. 6A and B), elucidating that R. mucilaginosa has the capacity to reduce growth of cariogenic pathogens, via iron competition. Taken together, these results show that growth and virulence responses to enterobactin are varied for strains belonging to the same species, which warrants further exploration in future studies. The commensal oral bacteria that were tested showed various responses to enterobactin and catalase (for a list of species tested, see Materials and Methods). For example, growth of Streptococcus sanguinis ATCC 49296 was not significantly impacted by enterobactin (Fig. 6C). However, when subjected to catalase treatment (with and without enterobactin), the S. sanguinis growth rate increased as optical density at 600 nm (OD 600 ) reached 0.25 after 10 h of growth compared to the non-catalase-amended cultures (an OD of 0.25 was reached after 20 h) (Fig. 6C). This reflects that catalase protects S. sanguinis from oxidative damage caused by its own ROS production and thereby stimulates its growth (40). Growth of Streptococcus gordonii ATCC 35105 was significantly enriched by enterobactin but only without catalase (P Ͻ 0.0001, t test of averaged mean values), which illustrates that growth was boosted by the presence of enterobactin (Fig. 6D). A similar trend was observed for Streptococcus salivarius SHI-3, which also increased significantly in cultures amended with enterobactin when no catalase was added (P Ͻ 0.05, t test of averaged mean values) (Fig. 6E). Conversely, Streptococcus oralis grew significantly more in the presence of catalase only (P Ͻ 0.0001, t test of averaged mean values), while growth was reduced by enterobactin (P Ͻ 0.05, t test of averaged mean values) (Fig. 6F). These results show that bacteria belonging to the Streptococcus genus have a varied response to enterobactin in the presence or absence of ROS. The liquid culture experiments demonstrate that S. salivarius and S. gordonii can benefit from the presence of enterobactinproducing R. mucilaginosa while S. oralis growth can be reduced. A growth boost of S. salivarius was also observed on coculture agar plates (minimal M9 medium supple-FIG 5 Yellow pigmentation of S. aureus strains exposed to 100 M enterobactin and 8 g/ml catalase on M9 agar plates supplemented with 100 M glucose. Yellow pigmentation was measured with the R package "countcolors" (51). All strains presented statistically significant reductions in pigmentation in the presence of 100 M enterobactin purified from R. mucilaginosa ATCC 25296 and catalase (P Ͻ 0.05, two-tailed t test). Box plots from yellow pixel measurements were generated with the R Studio program (49) and ggplot2 (50).
Oral Rothia mucilaginosa, an Enterobactin Producer mented with sucrose) adjacent to R. mucilaginosa (Fig. S5B). Whether this response was due to the presence of enterobactin needs to be explored further.
Liquid growth culture experiments including enterobactin were also performed for pathogenic Staphylococcus aureus strains. Growth of the enterotoxin H-producing strain S. aureus ATCC 51811 and the MRSA strain S. aureus TCH130 130/ST-72 was significantly reduced by enterobactin when catalase was present, which suggests that enterobactin negatively impacts S. aureus growth (P Ͻ 0.005, t test of averaged mean values) ( Fig. 7A  and B). None of the other tested MRSA strains showed reduced growth in enterobactinamended cultures (with or without catalase added) ( Fig. 7C and D). Interestingly, our findings demonstrate that strains belonging to the same bacterial species show different responses to enterobactin, which warrant further investigations of the role of strain-level sensitivity to this compound in MRSA virulence and growth.
To evaluate if enterobactin was actively chelating free iron throughout the liquid cultivation experiments, we incubated 500 M purified compound in sterile M9 liquid medium as described for bacterial liquid cultures. By using Arnow's assay, we observed that its binding capacity to molybdenum (which in Arnow's assay substitutes for iron) was reduced by 14% after 6 h of incubation. Assuming that this response is linear, we estimated that 56% of enterobactin's activity was lost after 24 h. This suggests that the compound became inactive in our coculture experiments over time for physiochemical and with enterobactin and catalase (right panel). Graphs were generated and statistically validated using R Studio and the "statmod" and "ggplot2" packages (48)(49)(50).
reasons and that a more dramatic effect of enterobactin likely would have been observed if its activity had remained stable.
We also investigated if enterobactin can bind other metal ions by using the calmagite assay, which is normally used for testing the hardness of water by detecting ions such as magnesium (Mg), calcium (Ca), and zinc (Zn) that bind to EDTA (41,42). EDTA is used as a reference compound due to its well-known chelation of divalent cations. Results from this test revealed that enterobactin can chelate both Mg 2ϩ and Zn 2ϩ ions. Enterobactin showed a chelation activity of 40 M EDTA-equivalents at a concentration of 100 M, which is 2.5 times less than EDTA affinity for Mg 2ϩ and Zn 2ϩ (Fig. S7). These results establish that enterobactin can indeed chelate additional ions and not only ferric iron (Fe 3ϩ ), which could have critical impacts on other microbial community members as well as the human host. It is well known that Zn 2ϩ is a critical component of the ubiquitous protein zinc fingers that are able to interact directly with DNA, RNA, and proteins. Magnesium (Mg 2ϩ ) is involved in nervous system signaling, immune system function, and bone formation. Our findings show that R. mucilaginosa has the potential to compete with the host not only for Fe 3ϩ but also for Mg 2ϩ and Zn 2ϩ , which suggest that R. mucilaginosa could play an important role in health and disease outcomes.
Our study reveals that R. mucilaginosa ATCC 25296 produces the archetype siderophore enterobactin when growth conditions are suboptimal in a minimal growth medium supplemented with glycerol, a known inducer of secondary metabolites in Streptomyces (43). We observed different growth responses to purified enterobactin by various members of the oral microbiota: some commensal Streptococcus species increased in growth while others decreased (independently in the presence of ROS). Our findings also demonstrate that growth of pathogenic bacteria (i.e., two cariogenic strains of S. mutans and methicillin-resistant strains of S. aureus) is reduced in growth in cultures amended with enterobactin. Moreover, while methicillin-resistant S. aureus grew adjacent to enterobactin-producing R. mucilaginosa ATCC 25296 (with or without ROS present), the virulence factor known as golden pigmentation (staphyloxanthin) Oral Rothia mucilaginosa, an Enterobactin Producer was highly reduced. While we provide evidence that the loss of this golden pigmentation is caused by enterobactin, the molecular mechanisms of the interaction remain to be explored. Further examination of the role of R. mucilaginosa enterobactin in inhibiting pigment production and bacterial growth may provide a pathway toward the development of new therapeutic leads against not only MRSA strains but also other pathogens, such as cariogenic Streptococcus mutans.
In conclusion, the role of enterobactin in health and dysbiosis of the oral microbiota is yet unexplored. However, based on previous research findings of enterobactin's role in various pathogenicities, it is difficult to overestimate the significance of its ironchelating capacity. It is possible that enterobactin-producing R. mucilaginosa can supply pathogens with additional iron by a siderophore-sharing mechanism and thereby fuel virulence. The fact that bacteria belonging to the Rothia genus not only represent one of the most prevalent oral bacterial groups but also have the capacity to control the availability of iron, both for the host and for other oral microbial community members, illustrates how critical it is to explore the ecological and clinical roles of enterobactin production by Rothia further. Growth media used in this study. A minimal medium was modified and developed from the M9 medium recipe reported by Elbing and Brent (45). All glassware was rinsed thoroughly with 2.5 M HCl and washed with deionized (DI) H 2 O before use. The fundamental M9 medium formulation (pH 7) was supplemented with 0.8% deferrated acid-hydrolyzed Casamino Acids (BD Biosciences, San Jose, CA, USA), 8 mM MgSO 4 (Millipore-Sigma, Carlsbad, CA, USA), and various carbon sources (sucrose, glycerol, galactose, lactose, arabinose, glucose, or lactate), to a final concentration of 100 mM. The acid-hydrolyzed Casamino Acids were deferrated with an equal volume of 3% 8-hydroxyquinoline (Fisher Chemical, Pittsburgh, PA, USA) in chloroform. MgSO 4 , glucose, and sucrose were added postautoclaving to prevent solution clouding and the Maillard reaction ("caramelization" of sugars with amino acids). Plates were made of the same formulations, containing 1% agar. Nutrient-and iron-rich brain heart infusion (BHI) media (Oxoid, Thermo Scientific, Carlsbad, CA, USA) were also used in liquid cultures and agar plates for cultivating experimental strains.

Bacterial isolates used in this study. The isolates used in this study included
Mining for BGCs in Rothia genomes. The biosynthetic gene cluster (BGC) prediction program antiSMASH, bacterial version (30), which is able to predict core secondary metabolite structures from BGC sequences, was used to identify BGCs in 26 genomes (including both completed and draft genomes) representing R. mucilaginosa, available at https://www.ncbi.nlm.nih.gov/genome/genomes/1812?; 12 genomes (including both completed and draft genomes) representing R. dentocariosa, available at https://www.ncbi.nlm.nih.gov/genome/genomes/1968?; and 7 genomes (one full-length and four draft genomes) representing R. aeria, available at https://www.ncbi.nlm.nih.gov/genome/genomes/12163?. The NaPDoS program was used for further classification of the nonribosomal peptide synthase (NRPS) catechol siderophore BGCs by using the C-domain classification tool (46). BGC structural analysis. For structural homology modeling and comparative analysis of the NRPS protein harbored by the Rothia mucilaginosa ATCC 25296 cat-sid BGC, Phyre2 software (35) was used to compare the NRPS with known structures in the Phyre2 database.
Cocultivation agar assays. For bacterial interaction screening studies, all bacterial isolates were seeded from glycerol stocks in BHI (Oxoid, Thermo Scientific) medium and incubated for 24 h before placement on either BHI agar or M9 minimal agar medium supplemented with either 100 mM sucrose (Acros Organics, Pittsburgh, PA, USA), glucose (Millipore-Sigma), or glycerol (Honeywell, Mexico City, Mexico). Plates were incubated either in a 5% CO 2 incubator at 37°C or anaerobically at 37°C in an anaerobic chamber (Coy Laboratory Products, Grass Lake, MI, USA) with a gas mix of 5% H 2 , 5% CO 2 , and 90% N 2 . All tested strains were grown similarly except A. timonensis, which was incubated for 48 h in BHI in an anaerobic chamber at 37°C before plating due to its lower growth rate in M9 minimal medium. R. mucilaginosa or R. dentocariosa was plated by dropping 20 to 30 l, in three to six replicates, onto plates, allowing the drops to air dry before incubation. Control incubations consisted of 20-to 30-l drops of each tested bacterial species. Agar plates with bacteria were incubated 24 to 48 h prior to adding another species. Interactions with other bacteria were tested by adding a 20-to 30-l drop next to the first strain, making sure the drop made contact with the established bacterial strain. Interactions were assessed after another 24 to 76 h of incubation. Interactions were screened with the naked eye for growth inhibitions zones or other growth-related behaviors. All interaction assays were repeated at least three times and documented.

Siderophore detection assays.
To clarify the type of siderophore produced by R. mucilaginosa ATCC 25296 and R. dentocariosa M567, siderophore-positive supernatants were assayed using Arnow's assay (32) for catecholate compounds, and the iron perchlorate assay for hydroxamate siderophores (31). For the hydroxamate siderophore assay, 1.0 ml of a 10 mM solution of Fe(CIO 4 ) 2 (Alfa Aesar, Tewksbury, MA, USA) in 0.1 M HClO 4 (Ricca Chemical, Visalia, CA, USA) was mixed with 1.0 ml of unknown and the absorbance was read at 495 nm. For Arnow's assay, cultures were screened by mixing 200 l culture with 20 l 5 M HCl (Ricca Chemical), followed by 100 l Arnow's reagent (20% sodium molybdate dihydrate [EMD, St. Louis, MO, USA] and 20% sodium nitrite [Fisher Chemical]) in DI H 2 O. To develop the ruby red color indicating the presence of a catecholate siderophore, 20 l 10 N NaOH (Ricca Chemical) was added. These were assessed visually for culture and HPLC fraction screening. For quantitative assays, these were measured spectrophotometrically at 500 nm.
Siderophore purification and structural elucidation. Siderophore enrichment in growth cultures of R. mucilaginosa ATCC 25296 was implemented in several steps. First, after 7 days of incubation, the supernatant of a liquid culture was acidified to pH 2. Twenty grams per 100 ml of Amberlite XAD16 (Alfa Aesar) was added to the supernatant, which was then placed in an orbital shaker at 175 rpm and shaken overnight. The Amberlite was collected, washed three times with DI H 2 O, and then extracted with diethyl ether (DEE) (Millipore-Sigma) and methanol (Millipore-Sigma). The solvent phase (the top diethyl ether layer) was evaporated at room temperature in an open container. The material was separated by silica gel thin-layer chromatography (TLC) (Millipore-Sigma); developed in a mixture of 65% n-butanol (Acros Organics), 25% acetic acid (Fisher Scientific, Pittsburgh, PA, USA), and 15% water; and then developed with Arnow's reagent. The Arnow-positive TLC band was removed and reextracted with DEE, solvent was evaporated, and the remaining material was tested with Arnow's assay. Mass spectrometry (LC-MS/MS) using a triple time of flight (TOF) mass spectrometer (AB Sciex 5600; Framingham, MA, USA) was done in both positive and negative modes on a PS C 18 column (Phenomenex, Torrance, CA; 2.6-m particle size, 4.6-mm diameter, 250-mm length) and eluted with a gradient of 0 to 100% acetonitrile and 0.1% formic acid (Honeywell, Mexico City, Mexico). Upon obtaining data, the msConvert program was used to convert vendor files to the mzXML format (47). The Global Natural Products Social Molecular Networking (GNPS) platform was used to putatively annotate the detected fragment masses (36). For siderophore purification, the crude DEE extract was resuspended in 50% methanol-water and purified with the PS C 18 column using a 20-min gradient of 30% to 65% buffer B (acetonitrile with 0.1% formic acid) using an Agilent 1200 series HPLC with a fraction collector. Buffer A consisted of water with 0.1% formic acid. The pure HPLC fraction was diluted with DI H 2 O, frozen at Ϫ80°C, and lyophilized to remove solvents and water. For NMR, pure lyophilized enterobactin was resuspended in dimethyl sulfoxide (DMSO)-d 6 and analyzed via nuclear magnetic resonance (NMR) with a 600-MHz (14.1-T) Bruker Avance III NMR fitted with a 1.7-mm inverse detection triple resonance cryoprobe with z-gradients.
Liquid cocultivation assay with pure enterobactin. In order to gain further insight into the effects of enterobactin on bacterial growth, liquid growth experiments were prepared with purified enterobactin from R. mucilaginosa. Because hydrogen peroxide can also be responsible for growth inhibition, catalase (also known as hydrogen peroxide oxidoreductase) was used in the culture medium to degrade hydrogen peroxide and isolate the effect of the siderophore itself. Briefly, to enrich for growth of each bacterial species that was included in cocultivation experiments in minimal M9 medium, cultures were first established in BHI from frozen glycerol stocks and incubated for 24 h aerobically at 37°C with 5% CO 2 . For the growth curve assays, glucose was chosen as a universal carbon source to accommodate all strains used, some of which are not able to grow on glycerol. BHI medium was removed from the cells prior to starting experiments in minimal M9 medium as follows: 1 ml of each culture was centrifuged at 7,000 rpm for 10 min, the supernatant was removed from the pellet using a sterile pipette, and the pellet was resuspended in an assay medium consisting of 0.5 ml minimal M9 medium supplemented with 100 mM glucose, 8 mM MgSO 4 , 0.8% deferrated amino acids, and 1 M FeCl 3 , either with or without 8 g/ml catalase (Millipore-Sigma). The bacterial cell suspensions were further diluted 1:20 in the same medium. A stock solution of enterobactin was prepared for the assay by suspending purified enterobactin in M9 plus 100 mM glucose, 8 mM MgSO 4 , and 0.8% deferrated amino acids with or without catalase to 200 M. One hundred microliters of the 200 M siderophore suspension was added to a 96-well plate, in triplicate, for each bacterial strain studied. One hundred microliters of the 1:20 bacterial suspension was added to each well, in triplicate for each condition. The control consisted of the same preparations, but without the siderophore. All plates were incubated at 37°C for 48 h in a Tecan Infinite M Nano spectrophotometer, and growth was monitored every hour by absorbance measurements at 600 nm. Statistical analysis was done with the R package "statmod" using the "compareTwoGrowth-Curves" function with an nsim parameter value of 100,000 (48) based on a mean T calculation. All graphical figures were generated using R Studio, version 1.2.5001 (49), and the package ggplot2 (50).
Agar plate assays of S. aureus strains amended with pure enterobactin from R. mucilaginosa. To assess growth and staphyloxanthin pigmentation (also known as golden pigment) production of S. aureus strains on M9 agar plates supplemented with glucose and pure enterobactin, 24-h cultures of S. aureus strains grown in BHI were premixed 1:2 with a 2ϫ solution containing 2ϫ M9 medium, 200 mM glucose, 16 g/ml catalase, and 200 M enterobactin so that final concentrations were 1ϫ M9, 100 mM glucose, 8 g/ml catalase, and 100 M enterobactin. Controls consisted of the exact same formulation but without enterobactin. Twenty-microliter drops were plated onto M9 agar plates (100 mM glucose) as three replicates and incubated 24 h at 37°C and 5% CO 2 . The R package "countcolors," version 0.9.1 (51), was used for quantifying the level of yellow pigmentation in the bacterial growth areas by setting a pixel color detection range using only the red-green-blue (RGB) values detected in the images obtained. A rectangular range of color (RGB scale) values for yellow pigmentation detection used were upper RGB The command in the "countcolors" package used was "countcolors::countColorsInDirectory," in order to automate yellow pigmentation parameters for all images in the image folder and to standardize detection across the strains and replicates. Separate images were created with the results in order to reveal the detection of yellow pixels by the software. These separate images show the pixels that were detected and counted by the R program "countcolors." The yellow pixels were automatically replaced with blue by the program to better visualize the effect of enterobactin on S. aureus pigmentation. Three images per condition and strain were measured for yellow pigmentation, and a two-tailed t test was performed for each strain using the percent values obtained by the "countcolors" program, comparing each strain with a control strain as described above. Stability testing of enterobactin. To test the stability of enterobactin under conditions used in the liquid growth experiments, pure enterobactin was resuspended at 1 mM in H 2 O and diluted 1:2 to 500 M in a 2ϫ solution of 2ϫ M9, 200 mM glucose, and 2 M FeCl 3 , either with or without 16 g/ml catalase. Twenty microliters of 500 M enterobactin under each experimental condition (with or without catalase) was added to three replicate wells on a 384-well plate, starting at time zero. Every 2 h, another row of 20-l triplicates was added, until 6 h, and incubated at 37°C in a Tecan Infinite M Nano spectrophotometer (Tecan Inc., Männedorf, Switzerland). At 6 h, fresh enterobactin 20-l aliquots were added for the zero time point, and samples representing all time points were assayed using Arnow's assay by adding 2 l 5 N HCl and 10 l Arnow's reagent, followed by 2 l 10 N NaOH. Catecholate absorbance was measured with an Infinite M Nano spectrophotometer (Tecan Inc.) at 500 nm.
Calmagite testing of pure enterobactin. Enterobactin is best known for its ability to bind the insoluble trivalent iron (Fe 3ϩ ). In order to assess the ability of enterobactin purified from R. mucilaginosa to bind magnesium and zinc, the compleximetric dye calmagite was used in an assay format (41,42). Calmagite forms colored complexes with magnesium and zinc as well as other metals and was developed to quantify magnesium in biological samples (41). A chelating compound, such as EDTA or enterobactin, is able to break this complex and elicit a color change from red to blue that can be measured spectrophotometrically. This assay was adapted to the analysis of enterobactin as follows: calmagite dye (Acros Organics) was diluted in DI H 2 O to 0.4 mg/ml. A working solution was prepared consisting of a 25 M metal ion solution (either MgSO 4 or ZnSO 4 diluted in DI H 2 O) and 0.05-mg/ml calmagite compleximetric dye diluted in 62.5 mM NH 4 Cl, pH 10. One hundred seventy-five microliters was mixed with either 25 l of a dilution series of EDTA (for quantification via a standard curve) or 25 l of enterobactin sample at 896 M and 100 M (unknowns). The standard curve and unknown samples were all assayed in triplicate, and the disappearance of the red calmagite-metal complex due to EDTA or siderophore metal binding competition was measured at 650 nm in a Tecan Infinite M Nano spectrophotometer set at room temperature. EDTA equivalents of enterobactin were calculated by extrapolating enterobactin absorbance values from third-order polynomial line-fitting of standard curve data (see Fig. S7 in the supplemental material).

SUPPLEMENTAL MATERIAL
Supplemental material is available online only.