Fermentative growth decreases the iron demand of Staphylococcus aureus

Abstract

Iron (Fe) is an essential nutrient for S. aureus survivability and pathogenesis, but excess Fe can catalyze the formation of toxic radicals, emphasizing the importance of maintaining proper Fe homeostasis. The essentiality of Fe for bacteria is exploited by host immunity strategies, which employ metal-binding proteins to decrease the availability of metal ions such as Fe. S. aureus responds to Fe limitation using the ferric uptake regulator (Fur) and the Fur protein antagonist (Fpa). During Fe-replete conditions, Fur functions as a transcriptional repressor of target genes. Upon Fe deprivation, Fur repression is relieved with the aid of Fpa, allowing for the increased expression of Fur-regulated genes such as iron uptake systems. We demonstrate that fur inactivation is required during Fe-limited growth and is independent of the described high-affinity Fe uptake systems. Using transcriptomic and metabolomic analyses, we demonstrate that fur inactivation or Fe limitation triggers a decrease in respiration and an increase in fermentation. Triggering fermentative growth allows S. aureus to cope with Fe limitation by having a metabolism less reliant on Fe but allowing for redox balance. Our work provides insight into how S. aureus adapts to iron limitation; a common stress encountered during infection.

Graphical abstract

During Fe-deplete growth conditions (right panel), Fur repression is alleviated with the aid of Fpa, resulting in increased transcription of the Fur regulon. Altered expression of the Fur regulon results in decreased dioxygen respiration and increased use of fermentative pathways to balance redox.

Introduction

Staphylococcus aureus is a human commensal bacterium and a leading cause of infectious disease-related illness and death worldwide (1). One reason that S. aureus infections continue to burden society is the increase in infections caused by S. aureus that are resistant to one or more antibiotics (2). In addition to increasing lethality, infections caused by antimicrobial-resistant S. aureus prolong hospitalization and have an increased financial burden (3–5).

S. aureus must acquire iron (Fe) from the environment for survival. The human body attempts to prevent S. aureus infections by decreasing access to Fe, while making it available for human cell use. Fe in the circulatory system is often ligated by proteins, including lactoferrin and transferrin (6). This process is called nutritional immunity. S. aureus attempts to circumvent nutritional immunity by synthesizing two siderophores and using monovalent Fe uptake systems. It can also take up and utilize the Fe in heme, and it synthesizes a metallophore, which may have a role in Fe ion uptake in addition to Zn(II) (7–10). Fe acquisition is essential because it is used to synthesize iron-sulfur (Fe-S) clusters and make Fe-S proteins, which are used for numerous metabolic processes (11).

Because the cytosolic accumulation of Fe ions is toxic, the cellular concentration of Fe ions is tightly regulated (12, 13). During iron-replete conditions, the staphylococcal ferric uptake transcriptional regulator (Fur) is thought to be metalated with Fe and act as a repressor of genes coding for Fe uptake pathways (14). In the canonical model for Fur function, when the cytosolic Fe concentration decreases, Fur loses its regulatory Fe ions, resulting in an altered affinity for DNA and alleviated repression of Fur-regulated genes (15). We recently discovered a protein (Fpa) that influences the alleviation of Fur-dependent repression in S. aureus (16). The presence of Fpa is necessary for the transcription of Fur-repressed genes. A Δfpa mutant has defective growth in Fe-starved media, and this phenotype is suppressed by a null mutation in fur. Similar results were reported in B. subtilis (17). S. aureus Fpa associates with Fur in vivo and in vitro (16). Biochemical analyses determined that Fpa decreased the ability of Fur to bind to DNA. Fpa binds an Fe atom with a K_d of ≈2 μM, which suggests that Fpa would be holo in Fe-replete conditions and apo in Fe-deplete conditions. It is currently unknown if Fe binding by Fpa impacts its ability to alleviate Fur repression or whether the S. aureus Fur directly senses the concentration of cytosolic Fe.

Besides directly controlling Fe acquisition gene targets, Fur impacts other cellular processes such as central metabolism. A proteomics analysis of S. aureus strain Newman demonstrated that in a Δfur strain or upon addition of a divalent metal chelator to alleviate Fur repression, expression of glycolysis and fermentation were upregulated, and expression of the tricarboxylic acid cycle (TCA) cycle was downregulated (18). We and others have discovered that Fur represses the transcription of isrR, which codes for a non-coding RNA (sRNA) (19–21). IsrR binds select messenger RNAs that code proteins that utilize Fe-containing cofactors or are utilized in processes that require Fe ions. We found that IsrR directly bound to and downregulated transcripts coding for TCA cycle enzymes AcnA, SdhC, and Mqo, explaining the decrease in TCA cycle in the Δfur mutant (19).

As stated above, the growth defect of an S. aureus Δfpa mutant is suppressed by the introduction of a null fur allele (22). In the present study, we took a multi-omics approach to understand how inactivation of fur led to suppression of the fpa phenotype and why Fur derepression is required under Fe-limited conditions. Our findings led to a model wherein Fe starvation causes Fur derepression, resulting in 1) increased transcription of genes coding Fe uptake systems, and 2) altered gene expression to favor a metabolism that requires fewer Fe ions. Alleviating Fur repression impacts metabolism by forcing an increased reliance on fermentation for redox balance and a concomitant decreased reliance on dioxygen respiration. We also demonstrate that fermenting S. aureus has a lower concentration of cell-associated Fe ions.

Experimental procedures

Chemicals and growth conditions.

All bacterial strains used in this study (Table 4) were derived from the community-associated methicillin-resistant Staphylococcus aureus isolate USA300_LAC (23). Bacteria were grown at 37 °C in tryptic soy broth (TSB) (MP Biomedicals) and shaking at 200 rotations per minute. Solid media was generated by adding 1.5% (wt/vol) agar (VWR). Chelex-treated TSB was prepared as previously described (24). Unless stated otherwise, cells were cultured in 10 mL capacity culture tubes containing 1.5 mL of liquid medium. Quantitative growth was conducted using a 96-well plate reader, BioTek 808E visible absorption spectrophotometer, at 37 °C and shaking. For quantitative growth, bacterial strains were grown for 18 hours and washed twice with PBS before diluting them to an optical density at 600 nm (OD600) of 0.02 in 200 μL. Anaerobic growth was conducted using a 37 °C incubator inside of a COY anaerobic chamber containing an oxygen scavenging catalyst to maintain oxygen levels <1ppm. Anaerobic growth was quantified by measuring culture OD₆₀₀ after 18 hours.

Table 4.

Bacterial strains used in this study.

Staphylococcus aureus USA300_LAC strains
Strain name	genotype	Reference
JMB1100	USA300_LAC wild type (WT)	(72)
JMB8689	Δfpa::tetM	(16)
JMB1432	Δfur::tetM	(73)
JMB4536	hemB::Tn	(59)
JMB10638	Δfpa::tetM fur*	(16)
JMB7525	fhuC::Tn	(11)
JMB10722	Δfpa::tetM fur* fhuC::Tn	(16)
JMB10664	Δfur::tetM fhuC::Tn	This study
JMB10675	proC::Tn	(16, 74)
JMB10677	proC::Tn Δfpa::tetM	(16, 74)
JMB10678	proC::Tn Δfpa::tetM fur*	(16, 74)
JMB15540	proC::Tn Δfpa::tetM fur*ΔisrR	This study

When necessary, antibiotics were added at the final following concentrations: 100 μg mL⁻¹ ampicillin (Amp); 10 μg mL⁻¹ erythromycin (Erm); 3 μg mL⁻¹ tetracycline (Tet). Protein concentrations were determined using Bradford reagent (Bio-Rad Laboratories Inc., Hercules, CA). DNA primers were purchased from IDT and are listed in Table S5. Molecular reagents were purchased from New England Biolabs. Unless stated otherwise, all chemicals were purchased from Sigma-Aldrich (St. Louis, MO). Sanger DNA sequencing was performed at Azenta (South Plainfield, NJ). The restriction minus strain S. aureus RN4220 was used for transformation (25), and transductions were carried out using bacteriophage 80α (26).

Aconitase enzyme assay

The aconitase assays were performed as previously described (27) with slight modifications. S. aureus strains were grown in 1 mL of TSB with 0 or 120 μM DIP for four hours with shaking at 200 RPM. Cells were washed twice with PBS. The cell pellet was collected by centrifugation and then stored at −80 °C before thawing and assaying. Protein concentrations were determined using the Bradford protein colorimetric assay modified for 96-well plates (Bio-Rad Protein Assay Dye Reagent Concentrate).

RNA-seq analysis

S. aureus strains were cultured overnight in biological triplicates. Cultures were diluted into 2.5 mL of fresh TSB to an OD₆₀₀ of 0.05 in 10 mL culture tubes with or without 120 μM DIP and subsequently cultured for eight hours aerobically with shaking. After eight hours, cultures were placed on ice, one mL of cells was harvested by centrifugation, washed with PBS, resuspended in 500 μL with RNA protect (QIAGEN), pelleted by centrifugation, and stored at −80°C. RNA extraction was performed as previously described (24). Cell pellets were thawed and washed twice with 0.5 mL of lysis buffer (20 mM RNase-free Sodium acetate, 1 mM EDTA, 0.5% SDS). The cells were lysed by the addition of 4 μg lysostaphin and incubated for 30 min at 37°C until confluent lysis was observed. RNA was isolated using TRIzol reagent according to the manufacturer's instructions. RNA-seq libraries and DNA sequencing was conducted by SeqCenter (Pittsburg, PA).

RNA-seq data were submitted to GEO and the accession identifier GSE228225 was provided. RNA-seq data analysis was performed using CLC Genomics Workbench (Qiagen) as described previously (28). Raw data was imported and any remaining rRNA reads were filtered out using the “map reads to reference” function. RNAseq analysis was performed using the S. aureus USA300 FPR3757 genome updated to include annotations for sRNAs (29). Individual differential expression analyses were performed between biological triplicates and data normalized using quantile normalization. To eliminate lowly expressed genes, we excluded from further analysis genes in which the average expression values in both samples was less than 10 RPKM. To eliminate genes with ambiguously assigned reads we excluded from further analysis genes in which the percentage of assigned reads in any of the samples were less than 80% unique. Genes were considered significantly altered between two strains if they demonstrated a fold change >2 or <2 and a p-value <0.05 as determined by Students t-test. Transcripts exhibiting significant differences in expression were placed into functional categories based on their annotations in the COG database (30) as described previously (31). PCA plots were generated in CLC Genomics Workbench and exported to GraphPad Prism for visualization. Gene set enrichment analyses were performed using ShinyGo v. 0.82 (32). Genes with fold change >2 or <2 and p-value <0.05 between WT and Δfur or between DIP challenged and control WT were selected and analyzed for all available enriched pathways (all available gene sets) using S. aureus Reference ID 1280. Pathways with FDR-corrected p-values <0.05 were selected and sorted by fold enrichment.

RNA extraction, cDNA synthesis, and qPCR

S. aureus strains were cultured overnight in TSB and subsequently diluted in triplicate to an OD₆₀₀ of 0.1 in 2.5 mL TSB with or without 500 μM DIP in 10 mL glass culture tubes. Cells were incubated at 37 °C with agitation for six hours after which time 1 mL cell pellets were treated with RNAprotect (Qiagen). Cell pellets were washed in 0.5 mL PBS pH 7.4, resuspended in 100 μL 50 mM tris pH 8 containing 6.7 μg lysostaphin, and then incubated at 37 °C with agitation for 30 minutes. The cell suspension was incubated at 65 °C for 5 minutes following the addition of 200 μL of 20 mM sodium acetate, 1 mM EDTA, 0.5% SDS. RNA isolation was performed as previously described (33). cDNA libraries were constructed using the High-Capacity cDNA Reverse Transcription kit (Biosystems). Quantitative real-time PCR was performed using an Applied Biosystems StepOnePlus thermal cycler and SYBR Green Master Mix. DNA primers used are listed in Table S7. Data were analyzed using the comparative C_T method (24, 34).

Metabolic modeling analyses

Model simulations were performed using the iYS854 genome-scale metabolic model of S. aureus (35) as previously described (36). RNA sequencing data were applied as modeling constraints using the iMAT algorithm (37, 38) using Gurobi and oxygen consumption rates were applied as additional modeling constraints for sampling in the COBRApy toolbox (39). The Fur deletion strain was modeled by setting the upper bounds for Fe(II) and Fe(III) to 0 to represent model uptake derepression. DIP treatment was modeled by creating sink reactions for Fe(II) and Fe(III) to represent Fe chelation. Simulations were performed by sampling each model 10,000 times using optGpSampler (40). Fluxes were visualized using Escher (41) using an iYS854 core metabolism map. The core metabolism map for iYS854 was derived from the core metabolism map of the e_coli_core model.

Metabolomic analyses

Strains were cultured overnight as biological triplicates before diluting into 2.5 mL of fresh TSB with or without 120 μM DIP to an OD₆₀₀ of 0.05 in 10 mL culture tubes. Cells were subsequently cultured for eight hours aerobically with shaking or statically in an anaerobic chamber. After eight hours, OD₆₀₀ was measured, and cultures were placed on ice. Samples for the metabolite profiling were prepared as described previously (42). Briefly, 1.5 mL of samples were collected, individual strains were normalized to the lowest OD₆₀₀ of the triplicates, cells were pelleted, and washed twice with PBS. Cell pellets were resuspended in 1 mL of Methanol:Acetonitrile:Water (2:2:1) solution. Cells were lysed by bead beating (2 cycles, 40 s each, 6.0 m s⁻¹) using a FastPrep homogenizer (MP Biomedicals) and 0.1-mm silica glass beads (MP Biomedicals). Samples were centrifuged twice at 14,500 rpm at 4 °C for 2 minutes and supernatant was retained. The supernatant was filtered with nylon membrane syringe filters (13 mm, 0.22 μm, Fisherbrand) and then samples were stored at −80 °C until metabolite analysis was performed.

Samples were analyzed at the metabolomics core of the Cancer Institute of New Jersey. HILIC separation was performed on a Vanquish Horizon UHPLC system (Thermo Fisher Scientific, Waltham, MA) with an XBridge BEH Amide column (150 mm × 2.1 mm, 2.5 μm particle size, Waters, Milford, MA) using a gradient of solvent A (95%:5% H₂O:acetonitrile with 20 mM acetic acid, 40 mM ammonium hydroxide, pH 9.4) and solvent B (20%:80% H₂O:acetonitrile with 20 mM acetic acid, 40 mM ammonium hydroxide, pH 9.4). The gradient was 0 min, 100% B; 3 min, 100% B; 3.2 min, 90% B; 6.2 min, 90% B; 6.5 min, 80% B; 10.5 min, 80% B; 10.7 min, 70% B; 13.5 min, 70% B; 13.7 min, 45% B; 16 min, 45% B; 16.5 min, 100% B; and 22 min, 100% B. The flow rate was 300 μL min⁻¹. The column temperature was set to 25 °C. The autosampler temperature was set to 4 °C, and the injection volume was 5 μL. MS scans were obtained in negative mode with a resolution of 70,000 at m/z 200, in addition to an automatic gain control target of 3 × 106 and m/z scan range of 72 to 1000. Metabolite data was obtained using the MAVEN software package (43) (mass accuracy window: 5 ppm).

Whole-cell metal quantification

S. aureus strains were grown for 18 hours overnight in TSB before diluting them to an OD of 0.05 (A₆₀₀) in 7.5 mL of TSB or Chelex (Bio-Rad)-treated TSB in a 30 mL capacity culture tubes as described previously (24). Cells were allowed to grow with shaking for eight hours. Pre-weighted metal-free 15 mL propylene tubes were used to pellet the cells using a prechilled tabletop centrifuge (Eppendorf, Hauppauge, NY). Pellets were washed three times with 10 mL of ice-cold PBS. All samples were kept at −80 °C or on dry ice until processing.

Cell pellets were acid digested with 2 mL of Optima grade nitric acid (ThermoFisher, Waltham, MA) and 500 μL hydrogen peroxide (Sigma, St. Louis, MO) for 24 hr at 60°C. After digestion, 10 mL of UltraPure water (Invitrogen, Carlsbad, CA) was added to each sample. Elemental quantification on acid-digested liquid samples was performed using an Agilent 7700 inductively coupled plasma mass spectrometer (Agilent, Santa Clara, CA). The following settings were fixed for the analysis Cell Entrance = −40 V, Cell Exit = −60 V, Plate Bias = −60 V, OctP Bias = −18 V, and collision cell Helium Flow = 4.5 mL min⁻¹. Optimal voltages for Extract 2, Omega Bias, Omega Lens, OctP RF, and Deflect were determined empirically before each sample set was analyzed. Element calibration curves were generated using ARISTAR ICP Standard Mix (VWR). Samples were introduced by a peristaltic pump with 0.5 mm internal diameter tubing through a MicroMist borosilicate glass nebulizer (Agilent). Samples were initially up taken at 0.5 rps for 30 s followed by 30 s at 0.1 rps to stabilize the signal. Samples were analyzed in Spectrum mode at 0.1 rps, collecting three points across each peak and performing three replicates of 100 sweeps for each element analyzed. Sampling probe and tubing were rinsed for 20 s at 0.5 rps with 2% nitric acid between each sample. Data were acquired and analyzed using the Agilent Mass Hunter Workstation Software version A.01.02.

Dioxygen consumption assay.

S. aureus strains were grown overnight in 5 mL TSB in 30 mL culture tubes at 37 °C in a shaker (200 RPM). Strains were diluted 1:100 into 10 mL TSB containing 0, 250, 500 or 750 μM DIP in 125 mL flasks and incubated at 37 °C with agitation for four hours. The cultures were diluted to an OD₆₀₀ of 0.025 in TSB containing 0, 250, 500 or 750 μM DIP prior to transfer of 200 μL to the wells of a Seahorse XF96 V3 PS cell culture microplate (Agilent). The seahorse XF sensor cartridge (Agilent) was hydrated in a non-CO₂ incubator with sterile water overnight and equilibrated in XF calibrant (Agilent) for two hours prior to measurement. Measurements were taken for 15 cycles with a 3-minute mix and 3-minute measure cycle.

Proton motive force measurement

Overnight cultures were grown in 2 mL TSB in 10 mL culture tubes at 37 °C with shaking. Strains were diluted 1:100 into 2 mL TSB +/− 500 μM DIP and incubated at 37 °C with shaking for six hours. Cell pellets were collected from 2 mL of culture and washed with 0.5 mL PBS, pH 7.4, and then resuspended in 0.5 mL PBS. The OD₆₀₀ was adjusted to 0.085 in PBS. Thirty micromolar 3,3′-diethyloxacarbocyanine iodide (DiOC₂(3)) +/− 5 μM carbonyl cyanide m-chlorophenyl hydrazone (CCCP) were added to the cell suspension followed by incubation at room temperature for 30 minutes. Red fluorescence (excitation of 450 nm, emission of 670 nm) was measured using a Varioskan Lux plate reader (Thermo Scientific).

Results

Iron deprivation or the loss of Fur alters the transcription of genes that function in fermentation.

We previously demonstrated that a S. aureus Δfpa mutant has a growth defect on solid tryptic soy broth (TSB) medium containing 2,2-dipyridyl (DIP), which is a cell-permeable divalent metal chelator with a high affinity for Fe(II) (44, 45). As demonstrated in Figure 1A, introducing a null fur mutation (fur*) relieves this phenotype (16). The fur* mutation codes for a null allele containing a nonsense mutation after the eleventh codon (E11stop) (16). FhuC is required to utilize Fe acquired using Fe-binding hydroxamates, staphyloferrin A, and staphyloferrin B, which are organic molecules that have a high affinity for Fe ions (16, 46). The suppression of the Δfpa mutant growth defect with DIP does not require these three high-affinity Fe uptake systems, as the introduction of a fhuC::Tn mutation does not alter the growth of the Δfpa fur* strain. Further analysis found that the fhuC::Tn mutant also had a growth defect on TSB medium containing DIP, and this growth defect was partially relieved by introducing a Δfur mutation (Figure 1B). These findings are consistent with the hypothesis that alleviating Fur repression aids growth during Fe depletion by altering gene expression to favor a metabolism that requires fewer Fe ions.

Figure 1. Introducing a null fur allele improves growth on low Fe medium independent of the high affinity iron uptake systems.

Panel A. Overnight cultures of the WT (JMB1100), Δfpa::tetM (JMB8689), Δfpa::tetM fur* (JMB10638), fhuC::Tn (JMB7525), and Δfpa::tetM fur* fhuC::Tn (JMB10722) strains were cultured in TSB before being standardized for optical density (OD₆₀₀), serially diluted, and spot plated on TSA with or without 700 μM 2,2-dipyridyl (DIP). Panel B. Overnight cultures of the WT, Δfur::tetM (JMB1432), fhuC::Tn, and Δfur::tetM fhuC::Tn (JMB10664) strains were cultured in TSB before being standardized for optical density (OD₆₀₀), serially diluted, and spot plated on TSA with or without 700 μM 2,2-dipyridyl (DIP). The experiments were performed three times and pictures of a representative experiment are displayed.

To provide insight into this phenotype, we examined how Fur shapes the transcriptome to overcome Fe limitation. We quantified RNAs using RNA-sequencing after the WT and Δfur strains were cultured in TSB with and without 120 μM DIP. This concentration of DIP was chosen because it does not cause a substantial defect in growth during logarithmic growth in either strain under the growth conditions utilized (Figure S1). Principal component analysis (PCA) of the regulons indicated that each strain and growth condition had a unique regulon and, for the most part, clustered together (Figure S2). We verified the RNA abundances corresponding to some of the genes that had significantly increased or decreased transcript abundances in each strain and growth condition using quantitative PCR (qPCR) (Figure S3 and Table S1).

There were 851 protein-coding transcripts that had significant log2 fold-changes greater than 1 or less than −1 between the WT and Δfur strain (Figure 2A). There were 498 protein-coding transcripts that had significant log2 fold-changes greater than 1 or less than −1 between the WT strain cultured with and without DIP (Figure 2B). As expected, there was an overlap between the Δfur versus WT and the WT + DIP versus WT regulons (Tables S2 and S3 and heat maps in Figures S4 and S5). The Δfur and WT + DIP regulons shared ≈ 25% of the upregulated genes with a log2 fold-change in abundance > + 1 and ≈ 30% of the downregulated genes with a log2 fold-change in abundance < −1 (Figures S6). The causes for the differences between these regulons are likely threefold. First, adding DIP competes for divalent metals, altering the DNA-binding affinity of divalent metal-sensing transcriptional regulators such as MntR (47). In support of this idea, there was increased transcription of MntR-regulated Mn(II) uptake genes (e.g. mntABC) in the DIP regulon but not in the Δfur strain (Figure S7A). There was no increase in mntABC transcription in the Δfur + DIP regulon, which could be caused by the interactive nature of Fur, PerR, and MntR in regulating transcription (48, 49). In support of this, the sitABCD operon, which is responsible for Mn(II) uptake in Salmonella enterica, and functionally similar to MntABC, is repressed by both Fur and MntR (50). Second, the Fe uptake regulon is derepressed in the Δfur strain, resulting in increased Fe uptake. This increased Fe uptake could result in physiological changes that are sensed by transcriptional regulators, resulting in altered transcription of regulatory networks. The increased Fe uptake in the Δfur + DIP strain could have directly impacted MntR regulation, explaining why mntABC expression was not derepressed. Third, we did not add enough DIP to fully derepress the Fur regulon. In support of this, the transcription of Fe uptake genes (e.g., iruO and sstB) was not fully derepressed in DIP-treated cells (Figures S7B and C).

Figure 2. Transcriptome analyses suggest Fur-dependent transcriptional changes redirect central metabolism upon Fe limitation.

Panel A: Volcano plot of genes that have a Log2 fold-change of ≥ 1 or ≤ −1 in the Δfur::tetM (JMB1432) strain compared to the WT (JMB1100). Panel B: Volcano plot of genes that have a Log2 fold-change of ≥ 1 or ≤ −1 in the 120 μM 2,2-dipyridyl (DIP) challenged WT strain compared to unchallenged.

Detailed analysis of the regulons showed that genes encoding nutrient uptake, energy generation enzymes, and iron-requiring proteins had altered regulation in the Δfur strain and DIP-challenged WT (Tables 1 and S2). The Δfur strain and DIP-challenged WT had increased expression of RNAs encoding Fe-uptake (sbnB) and fermentation enzymes (formate dehydrogenase: fdh) (Figures 2A and B, Table S2). Additionally, the Δfur strain and DIP-challenged WT had decreased expression of iron-requiring proteins such as terminal oxidases (qox, cyd), anaerobic ribonucleotide reductase (nrdG), and succinate dehydrogenase (sdh) (Figures 2A and B, Tables 1 and S2). These observations are supported by gene set enrichment analyses (GSEA), which revealed significant enrichment of fermentation-related pathways (lactose catabolic processes and glycolysis) in the Δfur strain (Table S4). GSEA also enriched for respiratory pathways (electron transport and cytochrome complex), which were significantly downregulated in expression in the Δfur and WT + DIP regulons (Table S4).

Table 1.

Fold-change in significantly altered RNAs that code predicted iron-requiring proteins.^*

Gene name	Function	Δfur / WT	WT + DIP / WT
pflA	Pyruvate formate lyase activating enzyme	−36.1	−7.5
sdhB	Succinate dehydrogenase	−2.5	-
acnA	Aconitase	-	5.2
nrdG	Ribonucleotide reduction	−45.5	−3.0
qoxC	Quinol oxidase	−24.4	−4.3
bioB	Biotin synthase	−17.9	-
ilvD	Dihydroxy-acid dehydratase	−10.1	-
cydA	Quinol oxidase	−4.9	−4.5
gltB	Glutamate synthase	-	3.7

^*Data were acquired from Table S2 and genes listed are significantly up or down regulated at least two-fold. The dash indicates that the gene was not found to be significantly altered in the regulon. Negative fold-change was reported as the negative reciprocal.

Iron deprivation or the absence of Fur decreases dioxygen respiration.

We tested the hypothesis that inactivation of Fur or Fe-limiting conditions downregulates cellular respiration by quantifying dioxygen (O₂) consumption. First, we validated the technique using the WT strain and a hemB::Tn mutant, which cannot maturate cytochrome oxidases (51). Whereas O₂ consumption could be monitored in the WT strain, the hemB::Tn strain had no detectable O₂ consumption, verifying that respiration is the primary consumer of O₂ under the growth conditions examined (Figure S8A).

We next quantified the rates of O₂ consumption using the WT and Δfur strains cultured with varying concentrations of DIP. We did not see a statistically significant decrease in the rate of O₂ consumption when the WT was co-cultured with 120 μM DIP (not shown), but there was a significant decrease in the rate of O₂ consumption at DIP concentrations ≥250 μM, and the rate decreased as a function of increasing DIP (Figure 3A). The basal rate of O₂ consumption was decreased in the Δfur mutant compared to the WT. Dioxygen consumption was unaffected in the Δfur strain challenged with 250 μM DIP, but there was a slight but significant decrease in O₂ consumption upon treatment with >500 μM DIP. These data are consistent with the hypothesis that Fur derepression is responsible for the decreased O₂ consumption in the DIP-challenged WT.

Figure 3. Iron limitation decreased dioxygen respiration.

Panel A: Dioxygen consumption rates (OCR) of the WT (JMB1100) and Δfur::tetM (JMB1432) strains cultured in TSB media with and without 2,2-dipyridyl (DIP). OCR was monitored using a Seahorse Analyzer. Panel B: The membrane potentials of the WT and Δfur::tetM strains were measured using the fluorescent dye 3′3′-diethyloxacarbocyanine iodide (DiOC₂) after culture in TSB media with and without 2,2-dipyridyl (DIP) and membrane decoupler carbonyl cyanide m-chlorophenyl hydrazone (CCCP). The data presented represent the average of biological triplicates, and standard deviations are displayed. The data were analyzed using Two-Way ANOVA, and ad hoc Tukey tests were conducted for pairwise comparisons. The * denotes a p-value ≤ 0.05.

We further tested the ability of the WT and Δfur strains to generate a membrane potential using the fluorescent dye 3′3′-diethyloxacarbocyanine iodide (DiOC₂). When bacteria generate a membrane potential, DiOC₂ accumulates intracellularly, resulting in a shift from emitting green fluorescence to emitting red fluorescence (52, 53). We validated the technique by analyzing the ability of DiOC₂ to detect altered membrane potential between the WT and hemB::Tn strains. Whereas there was a red shift in fluorescence in the WT strain, very little red fluorescence was noted in the hemB::Tn mutant, indicative of a reduced membrane potential (Figure S8B). The addition of carbonyl cyanide m-chlorophenyl hydrazone (CCCP), which dissipates the transmembrane electric potential (ΔΨ) and ΔpH, resulted in decreased fluorescence in the WT (Figure S8B). We next examined the membrane potential in the WT and Δfur strains after culture with and without 500 μM DIP. We chose this concentration of DIP because it was the minimal concentration found to fully reduce the rate of O₂ consumption displayed in Figure 3A. The membrane potential of the WT strain decreased upon co-culture with DIP (Figure 3B). The membrane potential was lower in the Δfur strain compared to the WT. There was not a significant descrease in the membrane potential in the Δfur strain when co-cultured with DIP, consistent with the hypothesis that Fur derepression was responsible for the decreased membrane potential in the WT. Our findings are consistent with a model wherein Fe starvation leads to Fur derepression, resulting in decreased cellular respiration.

Iron deprivation or the loss of Fur alters metabolite abundances.

We performed genome-scale metabolic modeling using our RNA-sequencing data and O₂ consumption measurements as modeling constraints (Table S5) (35). Consistent with these data, modeling analyses predicted decreased NADH dehydrogenase, succinate dehydrogenase, and cytochrome oxidase activity (Figure 4A), as well as decreased TCA cycle activity (Figure 4B) in iron-limiting conditions. To validate these predictions, we experimentally measured the activity of aconitase (AcnA), a TCA cycle FeS cluster-requiring enzyme, in the WT and Δfur strains co-cultured with DIP under the conditions used for RNA sequencing analyses (33). AcnA activity decreased in Fe-limiting conditions and was nearly undetectable in the Δfur strain (Figure 4C).

Figure 4. Genome-scale metabolic modeling analyses suggest Fe limitation induces fermentation.

Panels A and B: Modeling analyses predict decreased respiratory activity (Panel A) and TCA cycle activity (Panel B). Flux balance analysis was performed on the iYS854 genome-scale metabolic model of S. aureus using transcriptome and dioxygen consumption measurements for WT, 2,2-dipyridyl (DIP)-treated cells, and Δfur::tetM strains as modeling constraints. Panel C: AcnA activity in cell-free lysates generated from the WT and Δfur::tetM strains after four hours of culture with (green bars) and without (blue bars) 120 μM DIP. Panel D: Modeling analyses predict increased activity in lactate fermentation. For modeling analyses, data represent distributions for metabolic reaction activity simulations using 10,000 sampling points. Black lines represent median values for each distribution. Kruskal-Wallis tests were performed on data in panels A, B, D, and **** indicates p < 0.0001. For aconitase experiments in panel C, data represent the average of three biological replicates with standard deviations. The data were further analyzed using Two-Way ANOVA, and ad hoc Tukey tests were conducted for pairwise comparisons. The * denotes a p-value ≤ 0.05.

Modeling analyses further predicted several other global metabolic changes upon Fur derepression, including increased activity through lactate fermentation (Figure 4D), branched-chain amino acid biosynthesis, nucleotide biosynthesis, and the urea cycle. (Table S5). Modeling analyses also predicted decreases in lipid and cell wall biosynthesis and tRNA synthetase pathways (Table S5). Collectively, these data and modeling analyses are consistent with the hypothesis that upon Fe deprivation, Fur-dependent transcriptional repression is relieved, resulting in carbon flux being redirected to decrease reliance upon Fe-dependent enzymes, and increasing carbon flux through fermentation pathways to maintain redox homeostasis. Fe-dependent processes that are decreased upon Fur derepression include tRNA synthesis (MiaB), respiratory complexes, and TCA cycle enzymes. Visualizations of the flux predictions for the WT and Δfur strain cultured with and without DIP are depicted in supplemental Figures S9–12. The flux predictions highlight the increase in L- or D-Lactate formation under fur inactivation or Fe limitation.

To validate some of these modeling predictions, we cultured the WT and Δfur strains as described for the transcriptomic experiments before isolating metabolites and conducting untargeted metabolomic analyses. We also cultured the strains without a terminal electron acceptor (anaerobic incubation) to force fermentation. The DIP-treated WT and the Δfur strain shared several metabolites that had altered abundances. In the aerobic samples, when compared to the WT, the Δfur strain had more significantly altered metabolites than the DIP-challenged WT (Tables 2 and S6). In contrast, the number of significantly altered metabolites was similar in the fermenting Δfur and DIP-treated WT strains (Tables 3 and S6).

Table 2.

Log2 fold-changes of select metabolites that were significantly altered after aerobic growth^*

WT + 120 μM DIP / WT		Δfur / WT
Common Metabolites
Pyruvate	0.5	Pyruvate	0.8
AICAR (ZMP)	1.1	AICAR (ZMP)	2.4
Glycine	1.5	Glycine	2.5
Adenine	1.1	Adenine	1.0
Guanine	1.1	Guanine	2.3
Ribulose-5-phosphate	−0.4	Ribulose-5-phosphate	2.0
Proline	−0.8	Proline	−0.6
Aspartate	−0.8	Aspartate	−2.8
Serine	1.9	Serine	2.4
N-Formyl-L-methionine	−0.4	N-Formyl-L-methionine	0.9
Succinic acid	1.0	Succinic acid	2.6
Lactate	0.4	Lactate	0.8
Unique Metabolites
Acetyl-serine	2.4	Cystathionine	1.0
Uracil	1.0	Histidine	1.9
Thymidine	−0.5	Argininosuccinate	−2.9
		N-acetyl-L-ornithine	−2.4
		N-acetyl-glutamine	−2.1
		O-Phospho-serine	2.4
		Glutamine	−1.2
		Citrulline	−1.3
		Arginine	2.5
		UMP	0.6
		Guanosine	2.4
		Adenosine	2.2
		dTMP	0.9
		Uridine	2.0
		GMP	1.3
		ADP	2.0
		Thymidine	2.2
		2-Deoxyguanosine	2.9
		IMP	6.0
		Inosine	7.1
		Cytidine	1.4
		O-Succinyl-L-homoserine	−3.1
		Methionine	1.0
		Methionine sulfoxide	2.0
		Urocanate	1.7
		Alanine	0.7
		Tyrosine	1.8
		Threonine	3.1
		Lysine	1.8
		Tryptophan	2.0
		Phenylalanine	1.5
		Aconitate	2.8
		Alpha-Ketoglutarate	1.2
		Malate	1.8
		Thymine	1.1
		NAD	1.3
		Mevalonate	0.8
		Pyridoxal	1.2
		Pterine	−1.4

^*Bold compounds are decreased in the Δfur strain or to WT + DIP when compared to the WT strain. Student’s t-tests were used to compare the metabolite concentrations (Table S6) between strains (Δfur and WT) or growth conditions (WT + DIP and WT) and metabolites that had significantly altered abundances are included (p-value of ≤ 0.05).

Table 3.

Log2 fold-changes of select metabolites that were significantly altered after anaerobic growth

WT + 120 μM DIP / WT		Δfur / WT
Common Metabolites
Proline	−0.7	Proline	−1.8
Urocanate	−1.5	Urocanate	−1.5
Ornithine	−1.0	Ornithine	−1.1
Histidine	−2.3	Histidine	−1.6
Dihydroorotic acid	−2.4	Dihydroorotic acid	2.5
5-Aminolevulinate	−1.1	5-Aminolevulinate	−1.4
Unique Metabolites
Orotate	−0.7	N-acetyl-L-ornithine	−0.4
N-carbamoyl-aspartate	−1.0	O-Phospho-serine	2.8
Asparagine	−0.7	Citrulline	−2.6
Arginine	−0.4	Glutamine	−0.7
Pyruvate	0.4	Argininosuccinate	−3.3
Phosphoenolpyruvate	−1.2	N-acetyl-glutamine	−1.0
Glyceraldehyde-3-phosphate	2.2	2-Keto-3-deoxy-D-gluconate	−1.3
Dihydroxyacetone phosphate	2.4	Diaminopimelic acid	2.5
Deoxycytidine	0.7	Thiamine	0.5
2-Deoxyguanosine	2.6	dUMP	2.4
dUMP	2.0	CMP	−0.8
AMP	−0.5	AICAR (ZMP)	1.1
Pterine	−1.2
NADH	−0.9
Lysine	−0.6
Acetyl-serine	2.2
Cystine	−1.0

^*Bold compounds are decreased in the Δfur strain or to WT + DIP when compared to the WT strain. Student’s t-tests were used to compare the metabolite concentrations (Table S6) between strains (Δfur and WT) or growth conditions (WT + DIP and WT) and metabolites that had significantly altered abundances are included (p-value of ≤ 0.05).

Independent of respiration, the absence of Fur or the presence of DIP altered pools of amino acids with nitrogen-containing side chains (His, Gln, Pro) and decreased abundances of metabolites associated with nitrogen homeostasis (citrulline, ornithine, argininosuccinate, glutamine, N-acetyl-glutamine). Consistent with model predictions, Fe limitation induced nucleotide metabolism (Tables 1, 2, S5, and S6). Aerobic co-culture of the WT with DIP or the absence of Fur increased the abundance of adenine, guanine, thymine, and uracil. We hypothesize that this metabolic shift occurs to increase the production of the nitrogen-containing metallophore, staphylopine, and the siderophores, staphyloferrin A and B. Consistent with this hypothesis, there were altered abundances of diaminopimelic acid, pyruvate, O-phospho-serine, serine, ornithine, and histidine, which are metabolic precursors used for the synthesis of these metallophores (52, 54, 55).

During aerobic growth, the Δfur strain had increased pyruvate, malate, α-ketoglutarate, and succinate accumulation. The WT treated with DIP also accumulated pyruvate. The abundances of these metabolites were not altered in the fermenting cells, suggesting that flux into and/or through the TCA cycle was altered during respiratory growth upon Fur derepression (Tables 2 and 3). The fur mutant also accumulates Fe during aerobic growth, so at this time, we cannot rule out that the increased Fe in the Δfur strain was contributing to the altered abundances of the TCA cycle intermediates (16).

Fermentative growth bypasses the essentiality of Fpa during Fe deprivation.

Fur has difficulty alleviating repression during Fe starvation in the absence of Fpa, which acts as a Fur anti-repressor (16). We tested the hypothesis that a null fur mutation alters metabolism, bypassing the essentiality of Fpa in Fe-depleted media. The fur* mutation is genetically linked to the proC::Tn allele, allowing us to make a fur* Δfpa::tet strain. Both DIP and ethylenediamine-N,N’-bis(2-hydroxyphenylacetic acid) (EDDHA) are divalent metal ion chelators that have a high affinity for Fe (44, 45); however, DIP can cross the cell membrane, whereas the larger EDDHA likely does not (56, 57). We cultured the parent (proC::Tn) and isogenic fur*, Δfpa, and Δfpa fur* strains on solid or in liquid media with or without a metal chelator in the presence and absence of O₂. Since no other respiratory substrate was added to the media, anaerobic growth forces fermentation. Aerobic culture of the Δfpa strain on solid medium with DIP resulted in a growth defect (Figure 5A), and a growth defect was noted with either DIP or EDDHA in liquid media (Figure 5C) (16). The growth defect of the Δfpa mutant on solid medium containing DIP was relieved by introducing the fur* allele, consistent with the hypothesis that the inability of Fur to alleviate repression during Fe limitation resulted in the growth defect (16). All four strains grew equally well with chelator when plated on solid media and incubated anaerobically to force fermentation (Figure 5B). The Δfpa strain did not have a growth defect when cultured anaerobically in liquid media with the DIP or EDDHA concentrations utilized (Figure 5D). These data are consistent with the hypothesis that the essentiality of Fpa in Fe-deplete conditions can be bypassed by fermentative growth.

Figure 5. Fermentative growth bypasses the need for Fpa in Fe-deplete conditions.

Panels A and B: The proC::Tn (JMB10675), Δfpa::tetM proC::Tn (JMB10677), fur* proC::Tn (JMB10676), and Δfpa::tetM fur* proC::Tn (JMB10678) strains were cultured overnight in TSB before serially diluting and spot plating on TSA with and without 700 μM 2,2-dipyridyl (DIP). The cultures displayed in panel A were incubated aerobically, and the cultures in panel B were incubated anaerobically in the absence of a terminal electron acceptor to force fermentative growth. The experiment was performed in triplicates and photographs of a representative experiment are shown. Panels C and D: The WT (JMB1100) and Δfur::tetM (JMB1432) strains were inoculated into TSB medium with and without 120 μM DIP or 100 μM ethylenediamine-N,N’-bis(2-hydroxyphenylacetic acid) (EDDHA) before culturing overnight aerobically or anaerobically in the absence of a terminal electron acceptor. Culture optical densities were recorded after 18 hours of growth. The data displayed represent the average of biological triplicates, and standard deviations are displayed. The data were analyzed using Two-Way ANOVA, and ad hoc Tukey tests were conducted for pairwise comparisons. The * denotes a p-value ≤ 0.05, and ns denotes no significant difference.

Cells cultured fermentatively have lower Fe loads.

We tested the hypothesis that S. aureus accumulated less Fe when cultured in the absence of a terminal electron acceptor. We cultured the WT strain in either TSB or Chelex-treated TSB aerobically or anaerobically and quantified cell-associated Fe. Chelex resin binds to divalent metal ions and is subsequently physically separated from the medium, decreasing the amount of Fe in the medium. Cellular iron measurements were normalized to wet cell weight or magnesium (Mg) levels. The WT cultured aerobically contained approximately two times more Fe per gram of cells than cells cultured in Chelex-treated TSB (Figure 6A). This trend remained with the amount of Fe was standardized to the amount of Mg (Figure 6B). The aerobically cultured WT had approximately two times more cell-associated Fe than the anaerobically cultured cells (Figure 6A and 6B). The anaerobic medium was not supplemented with a terminal electron acceptor to force fermentation. Growth in Chelex-treated TSB did not result in a significant difference in Fe loads in the aerobically and anaerobically cultured cells, demonstrating that fermentative growth lessens S. aureus Fe burden.

Figure 6. Fermentative growth decreases cell-associated Fe.

Panels A and B. The wild type (WT: JMB1100) strain was cultured anaerobically (fermentatively) or aerobically in TSB or Chelex-treated TSB before cells were isolated and cell-associated ⁵⁶Fe and ²⁴Mg ions were quantified using Inductively Coupled Plasma Mass Spectrometry (ICP-MS). Panel A displays the parts of a billion Fe standardized to wet cell pellet weight. Panel B displays the ratio of ⁵⁶Fe to ²⁴Mg ions. The data displayed represent the average of biological quadruplicates with standard deviations shown. The data were further analyzed using Two-Way ANOVA, and ad hoc Tukey tests were conducted for pairwise comparisons. The * and ** denote a p-values of ≤ 0.05 and ≤ 0.01, respectively. The ns denotes no significant difference.

Conclusions

In a previous study, we observed that an Δfpa mutant had a growth defect when cultured in Fe-starved media, which was recovered upon introducing a null fur mutation (16). Fpa interacts with Fur and alleviates Fur repression; therefore, in the absence of fpa, the Fur regulon remained repressed upon Fe depletion, leading to poor growth. Most genes directly regulated by Fur participate in Fe uptake, suggesting that the growth defect of a Δfpa mutant was the result of an inability to alleviate repression of Fe uptake genes. Surprisingly, the increased growth of the fpa fur* strain was independent of the three high-affinity Fe uptake systems (fhuC::Tn). Herein, we demonstrate that the inactivation of fur also restores the growth of a fhuC::Tn mutant upon Fe starvation. We began this work to investigate how the alleviation of Fur repression aided growth during Fe-starved conditions.

Transcriptomic and metabolomic analyses suggested that during aerobic growth, introducing a fur mutation or culturing the WT with DIP increased the usage of fermentation pathways for redox balance while decreasing dioxygen respiration. Consistent with this, the fur mutant or culture of the WT with DIP reduced the rate of dioxygen consumption and decreased the generation of an electrochemical potential. Physiological and genetic experimentation demonstrated that Fe-deplete medium decreases the growth of S. aureus, but this phenotype is reversed if cells are cultured anaerobically. Similar findings were reported by Ledala et al., who observed that limiting aeration obscured the effects of iron limitation, suggesting fermentative metabolism rescues the growth defect of Fe-starved cells (58). Our experiments were conducted under well-aerated conditions so as not to cause an additive effect on respiration/TCA cycle as a result of low dioxygen tension (59).

If glucose and dioxygen are present and cultures are aerated, exponentially growing S. aureus consumes glucose and balances redox with a mixture of fermentative and respiratory metabolism (58). During the post-exponential growth phase, S. aureus shifts to a fully respiratory metabolism. The transition into the respiration-reliant post-exponential and stationary growth phases coincides with increased cell-associated Fe (58, 60). The increase in Fe suggests that respiratory metabolism increases the cellular demand for iron. Collectively, these findings are consistent with a working model where, in Fe-deplete growth conditions, Fur repression is alleviated, resulting in transcription of the Fur regulon. Increased transcription of the Fur regulon decreases the rate of dioxygen respiration to decrease Fe usage and increases the use of fermentative pathways to balance redox (Figure 7).

Figure 7. Working model for respiration and fermentation during growth in Fe-deplete conditions.

The data presented were used to develop a working model for the effects of Fe-replete (left panel) and Fe-deplete (right panel) conditions on redox balance in S. aureus aerobically cultured in TSB medium. During Fe-deplete growth conditions (right panel), Fur repression is alleviated with the aid of Fpa, resulting in increased transcription of the Fur regulon. Altered expression of the Fur regulon results in decreased dioxygen respiration and increased use of fermentative pathways to balance redox.

Friedman et al. conducted proteomics analyses on S. aureus Newman strain and an isogenic Δfur mutant after culture in Fe-replete and Fe-deplete media (18). Both fur inactivation and Fe-chelation caused an increase in proteins used for converting glucose to pyruvate, and as we have demonstrated, a fur mutant has decreased expression of TCA cycle enzymes (19). There was a concomitant increase in the expression of enzymes involved in lactate and formate generation through fermentation, which aligns with the transcriptomics findings and metabolic modeling analyses presented. The increased lactate production increased media acidification, and they hypothesized that this was a strategy to decrease local pH and help acquire Fe from host Fe-ligating proteins (18). Consistent with this, conditioned culture medium from a fur mutant or WT cultured with DIP had decreased pH and promoted the release of Fe from transferrin. They also propose that metabolism redirection causes a decrease in the oxidation-reduction potential, which could impact the oxidation state of Fe ions. The conversion of the insoluble ferric ion to the more bioavailable ferrous ion would help relieve Fe limitation stress. Collectively, these studies suggest that in response to Fe depletion, S. aureus upregulates fermentative metabolism to 1) increase Fe availability, and 2) sustain a metabolism less reliant on Fe that maintains redox homeostasis.

The phenomenon that iron limitation induces a shift in metabolism towards fermentative growth has been previously observed in Escherichia coli (61). During growth in low Fe, E. coli mimics microaerobic growth by decreasing respiration and increasing flux through glycolysis and the pentose phosphate pathways (61). Supporting the increase in fermentation, Fe-starved E. coli accumulates the fermentative metabolites formic and acetic acid (61).

The decrease in respiration in response to Fe limitation has been described to be mediated by Fur-regulated non-coding RNAs (sRNAs). Known as the iron-sparing response, this mechanism was initially revealed by the discovery of RyhB in E. coli (62). Under iron limitation, ryhB is expressed and impacts the translation of mRNAs coding for non-essential Fe-utilizing proteins such as aconitase and respiratory complexes (62, 63). By repressing translation of expendable Fe-utilizing proteins, it allows the cell to ‘spare’ Fe for essential iron-dependent functions. Fe-sparing sRNAs have been characterized in other bacteria and share similar targets to RyhB. Additional iron-sparing sRNAs include FrsA in Bacillus subtilis, PrrF1/PrrF2 in Pseudomonas aeruginosa, and most recently, IsrR in S. aureus (19, 21, 64, 65).

In a recently accepted companion manuscript, we demonstrate that IsrR directly regulates and decreases the expression of terminal oxidases. Terminal oxidases require heme, and we also demonstrate that IsrR directly regulates and decreases heme biosynthesis. This concerted action may explain the decreased dioxygen consumption and decreased membrane potential in the Δfur mutant. Surprisingly, the introduction of an ΔisrR mutation does not alter the ability of a null fur mutation to suppress the growth defect of the Δfpa strain on Fe-chelated medium (Figure S13). These data suggest that introducing a null fur mutation into the Δfpa strain allows for growth on Fe-chelated medium independent of the IsrR regulon or the high-affinity Fe uptake systems. Lack of respiration requires another mechanism to balance redox to continue glucose catabolism. Proteomics studies (16) and the transcriptomic studies herein suggest that Fur positively increases the expression of genes that function in fermentation during Fe-depletion. Future studies will be required to determine whether this altered expression directly results from Fur regulation.

This work provides insight into the dynamic metabolism of S. aureus that enables overcoming challenges encountered in diverse host niches upon Fe limitation. We noted an alteration of N-containing metabolites (histidine, o-phospho-serine, ornithine) that could serve as precursors for Fe uptake metallophores (55). The fur mutant has increased purine synthesis metabolites such as AICAR, guanine, and adenine (Tables 2 and 3). Although it is unclear how or why purine biosynthesis is altered in a Δfur or upon Fe limitation, de novo purine biosynthesis has been shown to play a role in persistent infections and is required for full S. aureus virulence (66–68).

The transcriptomic findings presented align with previous studies, demonstrating that low Fe accessibility decreased expression of genes coding for immunomodulatory proteins (ecb, scn, sssA) and increased cytolysins and proteases (hla, hlgC, splA, aur, gehA) (Table S2) (69). The impact of Fur on virulence factors is thought to be indirect, as Fur has been shown to impact virulence regulators such as Agr, Sae, and Rot (18, 70, 71). We propose that upon iron scarcity, Fpa mediates Fur derepression to 1) increase Fe uptake, 2) decrease expression of Fe-using proteins through IsrR, 3) induce fermentative metabolism, and 4) impact other cellular processes relevant to staphylococcal pathogenesis. Future studies will uncover the mechanism behind how Fur regulates processes like fermentative metabolism and virulence factor expression.

Highlights.

1. Report the transcriptome of a wild type S. aureus and isogenic Δfur mutant upon divalent metal limitation.

2. Report the metabolite profile of wild type S. aureus and Δfur mutant upon divalent metal limitation.

3. Demonstrates that fermentative growth of S. aureus demands less Fe.