ABSTRACT
Iron is an essential micronutrient required for the viability of many organisms. Under oxidizing conditions, ferric iron is highly insoluble (∼10−9 to 10−18 M), yet bacteria typically require ∼10−6 M for survival. To overcome this disparity, many bacteria have adopted the use of extracellular iron-chelating siderophores coupled with specific iron-siderophore uptake systems. In the case of Bacillus subtilis, undomesticated strains produce the siderophore bacillibactin. However, many laboratory strains, e.g., JH642, have lost the ability to produce bacillibactin during the process of domestication. In this work, we identified a novel iron acquisition activity from strain JH642 that accumulates in the growth medium and coordinates the iron response with population density. The molecule(s) responsible for this activity was named elemental Fe(II/III) (Efe) acquisition factor because efeUOB (ywbLMN) is required for its activity. Unlike most iron uptake molecules, including siderophores and iron reductases, Efe acquisition factor is present under iron-replete conditions and is regulated independently of Fur repressor. Restoring bacillibactin production in strain JH642 inhibits the activity of Efe acquisition factor, presumably by sequestering available iron. A similar iron acquisition activity is produced from a mutant of Escherichia coli unable to synthesize the siderophore enterobactin. Given the conservation of efeUOB and its regulation by catecholic siderophores in B. subtilis and E. coli, we speculate that Efe acquisition factor is utilized by many bacteria, serves as an alternative to Fur-mediated iron acquisition systems, and provides cells with biologically available iron that would normally be inaccessible during aerobic growth under iron-replete conditions.
IMPORTANCE Iron is an essential micronutrient required for a variety of biological processes, yet ferric iron is highly insoluble during aerobic growth. In this work, we identified a novel iron acquisition activity that coordinates the iron response with population density in laboratory strains of Bacillus subtilis. We named the molecule(s) responsible for this activity elemental Fe(II/III) (Efe) acquisition factor after the efeUOB (ywbLMN) operon required for its uptake into cells. Unlike most iron uptake systems, Efe acquisition factor is present under iron-replete conditions and is regulated independently of Fur, the master regulator of the iron response. We speculate that Efe acquisition factor is highly conserved among bacteria and serves as a backup to Fur-mediated iron acquisition systems.
INTRODUCTION
Iron is required for the survival of virtually all organisms. Many central metabolic processes require iron, including the tricarboxylic acid (TCA) cycle, electron transport, amino acid and nucleotide biosynthesis, nitrogen fixation, oxidative stress response, DNA replication, and photosynthesis. In addition, iron plays an important role in other processes, including the expression of virulence factors, host colonization, and symbiosis (see reviews in references 1, 2, and 17). Excess iron can be detrimental to cells, leading to the generation of hydrogen peroxide, superoxide anion, and hydroxyl radicals (3–5). These reactive oxygen species cause damage to DNA and proteins, membrane dysfunction, and lipid peroxidation (6–10).
Depending on its oxidation state, iron can exist as oxidized ferric Fe(III) or reduced ferrous Fe(II) iron. In Bacillus subtilis and many other bacteria, iron homeostasis is maintained by the ferric uptake repressor (Fur). Fur acts as a global sensor of intracellular iron by coordinating transcription of iron responsive genes with available iron. Under iron-replete conditions, B. subtilis Fur, along with its corepressor Fe(II), binds to the “Fur box” comprised of two overlapping operator sites, each with a 7-1-7 minimal recognition element (11, 12). Binding of Fur-Fe(II) to its operator sequence represses transcription of 20 operons (39 genes) whose gene products are involved in iron acquisition (13). When iron is limited, Fur dissociates from its operator sequence, enabling the transcriptional machinery access to promoter DNA, and derepression of target genes ensues (11, 12).
Bacteria typically require ∼10−6 M iron for survival (14, 15). Although iron is the fourth most abundant element in the earth's crust, the availability of iron is severely limited in aerobic environments at physiological pH. Under these conditions, ferrous iron is rapidly and spontaneously reduced to ferric iron. In addition, the concentration of available ferric iron is estimated to be ∼10−9 to 10−18 M, with the majority existing as ferric oxides and ferric hydroxides that aggregate into biologically inert polymers (14–16). Ferric iron is even more restricted in a host environment, e.g., ∼10−24 M. Specifically, host iron is sequestered into heme, iron-sulfur proteins, and iron storage proteins that suppress the generation of reactive oxygen species (17, 18).
Not surprisingly, bacteria have evolved sophisticated strategies to acquire iron from the host and from the environment. For example, Legionella spp. and Streptococcus spp. reduce ferric iron into the more soluble ferrous form that is subsequently transported directly into cells (19–21). The pathogenic bacteria Neisseria spp., Helicobacter pylori, Vibrio spp., and Yersinia spp. acquire iron directly from the host iron storage proteins transferrin, lactoferrin, and heme (22–25).
Many bacteria and fungi utilize a common strategy of acquiring iron by secreting extracellular iron-chelating molecules called siderophores (Greek for iron bearer). These low-molecular-mass molecules (<1 kDa) have a high affinity for ferric iron, with dissociation constants ranging from 10−20 to 10−50 M. Specific uptake complexes import Fe(III)-siderophores directly into the cell to satisfy the cell's requirement for iron (2, 18, 26).
B. subtilis can acquire iron from many different sources. B. subtilis has multiple uptake systems for internalizing Fe(III)-xenosiderophore complexes produced by other bacteria and fungi (27). Alternatively, B. subtilis can import elemental ferric and ferrous iron directly into cells via the EfeUOB (YwbLMN) complex, an uptake system conserved among many bacteria and yeasts (28–30). Citrate also acts as an iron chelator, and the yfmCDEF gene products are required for importing Fe(III)-citrate into cells (27). Undomesticated strains of B. subtilis, e.g., NCBI 3610, synthesize the catecholic siderophore bacillibactin (BB; 2,3-dihydroxybenzoyl glycine-threonine). The dhbABCDE operon encodes the biosynthetic machinery for the synthesis of bacillibactin and itoic acid (2,3-dihydroxybenzoyl glycine), the latter a precursor to bacillibactin with weak iron-chelating activity (31–33). Many enteric bacteria, including Escherichia coli, produce siderophores from 2,3-dihydroxybenzoate, e.g., enterobactin (EB), that are structurally similar to bacillibactin (34). In B. subtilis, Fe(III)-bacillibactin and Fe(III)-enterobactin are imported into cells via the FeuABC-YusV complex (27). Interestingly, the B. subtilis laboratory strain JH642 fails to produce bacillibactin due to mutations in the gene encoding phosphopantetheinyl transferase (sfp0) that occurred during the process of domestication (32, 35). Thus, the question remains, how do laboratory strains of B. subtilis acquire iron in the absence of bacillibactin?
In this work, we identified a novel iron acquisition activity produced from cultures of B. subtilis laboratory strain JH642. The molecule(s) responsible for this activity was named elemental Fe(II/III) (Efe) acquisition factor because the efeUOB (ywbLMN) operon is required for its uptake into cells. Efe acquisition factor accumulates in the growth medium as cultures progress through the growth cycle and coordinates the iron response with population density. Efe acquisition factor is produced under iron-replete conditions, and its synthesis is regulated independently of Fur. Restoring bacillibactin production in strain JH642 inhibits the activity of Efe acquisition factor, presumably by sequestering available free iron. A similar Efe acquisition activity is present in cultures of E. coli that are defective for enterobactin production. We speculate that this novel iron acquisition activity is conserved among many bacteria and serves as a backup to Fur-mediated systems.
RESULTS
Rationale and experimental design.The disparity between the amount of available ferric iron present during aerobic growth (10−9 to 10−18 M) and the estimated ∼10−6 M required for viability raises the question of how laboratory strains of B. subtilis, e.g., JH642, acquire iron in the absence of the bacillibactin biosynthetic machinery. We rationalize that if strain JH642 utilizes an unidentified, extracellular molecule(s) to acquire iron, it is predicted to accumulate in the growth medium. Add-back experiments using cell-free conditioned medium could provide insight into the presence of such a factor, should it exist. Specifically, the addition of conditioned medium from a high-density culture containing iron acquisition molecules to an actively growing low-density culture lacking these molecules should, in theory, result in a decrease in transcription of Fur-regulated genes as cells rapidly internalize iron and repress the Fur regulon.
To test this hypothesis, cell-free conditioned medium was prepared by growing a culture of strain JH642 to high cell density (optical density at 600 nm [OD600] ∼ 1.5). Cells were pelleted by centrifugation, and the medium was filter sterilized (see Materials and Methods). Test strain JMS122 (ΔcomA), an isogenic derivative of the laboratory strain JH642, was grown to low cell density (OD600 ∼ 0.2). The culture was separated into two shaker flasks. One flask received an equal volume of conditioned medium, while the second flask received an equal volume of uninoculated medium. Both cultures were allowed to grow for an additional 30 min, after which time cells were harvested, RNA was isolated from each culture, and microarray analyses were performed to measure global changes in gene expression (see Materials and Methods) (Fig. 1). Strain JMS122 (ΔcomA) was used in these studies to avoid the identification of quorum sensing genes that are known to be regulated under these growth conditions by the transcriptional activator ComA (36, 37).
Overview of experimental design. Add-back experiments were performed to identify genes differentially regulated by conditioned medium and uninoculated medium. Specifically, cell-free conditioned medium was prepared from a high-density culture of strain JH642 as described in Materials and Methods. Strain JMS122 (ΔcomA) was grown to low cell density (OD600 ∼ 0.2). The culture was divided into two shaker flasks. One flask received an equal volume of cell-free conditioned medium, while the other flask received an equal volume of uninoculated medium. Cultures were allowed to grow for 30 min at 37°C with vigorous aeration. RNA was isolated from each culture, and microarray analyses were used to measure changes in gene expression (see Materials and Methods).
Transcriptome analysis.A microarray platform, developed by Auchtung et al. (38), was used in this study to identify genes whose mRNA levels differed at least 2.5-fold when treated with conditioned medium or uninoculated medium compared to a reference containing equal amounts of cDNA from both samples (see Materials and Methods). Although changes in mRNA levels can result from transcription or mRNA stability, we assume that changes in mRNA levels reflect actual changes in gene expression. A total of 49 genes were identified whose mRNA levels differed at least 2.5-fold between the two conditions. Three genes were upregulated a modest degree, 2.5- to 3-fold, while 46 genes were downregulated 2.5- to 14-fold (Table 1).
Genes differently regulated by conditioned mediuma
The most striking change in transcription came from genes directly regulated by Fur. Most of the genes in the Fur regulon were downregulated 2.5- to 12-fold in response to conditioned medium (Table 1, bold). With few exceptions (e.g., yfmCDEF operon), every gene in a Fur-regulated operon met the criteria for significance. The remaining genes listed in Table 1 were not considered for further analysis because they lack association with a recognizable pathway, they represent only a small part of a specific pathway, or the effects on transcription were minimal. Instead, we focused our analysis on Fur-regulated genes since cells respond to conditioned medium by repressing transcription of the Fur regulon.
Iron acquisition activity increases with population density.The gene expression profiles from the microarray experiments were generated by normalizing experimental cDNA samples to a reference sample comprised of equal amounts of cDNA from cultures treated with uninoculated and conditioned media. This allows us to more accurately compare gene expression across different but related experiments. However, this analysis results in an underrepresentation of the absolute magnitude of the response. Therefore, quantitative reverse transcription-PCR (qRT-PCR) was used to more accurately measure the relative mRNA abundance of representative Fur-regulated genes following the addition of conditioned medium in the add-back experiments (see Materials and Methods). Consistent with the microarray analyses, a decrease in transcription was observed upon the addition of conditioned medium for all 5 Fur-regulated genes tested (dhbA, fhuD, feuA, ydbN, and ykuN). Depending on the gene, a maximal 10- to 1,000-fold decrease in mRNA levels was observed in the initial 20 min after treatment, followed by a steady increase in mRNA throughout the course of the experiment (Fig. 2A, black bars). In contrast, a 2- to 3-fold increase in mRNA abundance was observed for all 5 genes following the addition of uninoculated medium (Fig. 2A, gray bars).
Effect of cell-free conditioned medium on transcription of Fur-regulated genes. Add-back experiments were performed using conditioned medium and uninoculated medium. Quantitative RT-PCR was used to measure mRNA abundance of Fur-regulated genes (see Materials and Methods). Transcript abundance was normalized to uninoculated medium at 0 min (UM T = 0 min). Experiments were performed in triplicate, and the percent standard error is shown. (A) The relative mRNA abundance was determined for Fur-regulated genes dhbA, fhuD, feuA, ydbN, and ykuN. Black bars represent the addition of conditioned medium from a high-density culture, while gray bars represent uninoculated medium. Asterisks represent P values of <0.01 relative to UM T = 0 min. (B) Abundance of the dhbA gene was determined after the addition of uninoculated medium (UM) (dark gray bars), cell-free conditioned medium from a low-density (LD; OD600 ∼ 0.2) culture of strain JH642 (light gray bars), or conditioned medium from a high-density (HD; OD600 ∼ 1.5) culture of strain JH642 (black bars). Asterisks represent P values of <0.01 between the two conditions.
The density of the culture used to prepare the conditioned medium in the add-back experiments had a profound effect on the magnitude of the response by Fur. Specifically, a modest (3-fold) reduction in dhbA mRNA was observed from strain JH642 following 20 min of incubation in conditioned medium from a low-density culture (Fig. 2B, light gray bars). In contrast, a 100-fold reduction in dhbA mRNA was observed following treatment with conditioned medium from a high-density culture (Fig. 2B, black bars). In both cases, an increase in dhbA mRNA was observed after the initial 20 min of treatment with conditioned media (Fig. 2B, light gray and black bars). We conclude that the extracellular factor(s) responsible for modulating the activity of Fur-regulated genes is present or becomes activated at high cell density but not low density.
The simplest explanation for these results is the accumulation of an iron acquisition activity present in conditioned medium that provides biologically available iron to cells. During the add-back experiment, iron present in conditioned medium is predicted to be internalized by cells, resulting in the observed repression of Fur-regulated genes (Fig. 2A, black bars, 20 min). As iron is consumed by cells, a gradual derepression of iron utilization genes occurs as cells attempt to acquire additional iron to maintain homeostasis (Fig. 2A, black bars, 20 to 120 min). The presence of an iron acquisition activity also explains the derepression of Fur-regulated genes observed when uninoculated medium was used in the add-back experiment. Diluting any preexisting iron acquisition factors present in the growth medium with uninoculated medium is predicted to decrease the concentration of iron available to cells resulting in derepression of Fur-regulated genes as cells attempt to acquire additional iron (Fig. 2A, gray bars).
Fur coordinates transcription of iron-responsive genes with population density.Accumulation of an extracellular iron acquisition activity in the growth medium could provide an effective way to coordinate transcription of the iron response with population density. To test this prediction, promoter fusions to lacZ were created for several Fur-regulated genes and the transcriptional response was monitored throughout the growth cycle by measurement of β-galactosidase activity (see Materials and Methods). For each reporter tested (dhbA, ykuN, and feuA), β-galactosidase specific activity was low at low cell density, increased 2- to 3-fold during exponential growth, and decreased as cultures transitioned into stationary phase (Fig. 3A). To determine the role of Fur repressor in the observed response, a fur deletion was constructed in each of the reporter strains (see Materials and Methods). The density-dependent transcriptional regulation was abolished in the fur mutant strains. Specifically, transcription was derepressed 3- to 6-fold compared to that in the wild type and remained constant throughout the growth cycle, resulting in high β-galactosidase activity even as cells transitioned into stationary phase (Fig. 3B).
Effect of Fur on density-dependent regulation of iron-responsive genes. Cultures containing Fur-regulated promoter fusions to lacZ were grown in minimal medium, aliquots were removed every hour, and β-galactosidase activity was determined (see Materials and Methods). Experiments were repeated in triplicate, with similar results. Results from a single representative experiment are shown. (A) Results with strains NC209 (PdhbA-lacZ), KG46 (PykuN-lacZ), and KG19 (PfeuA-lacZ). (B) Results with strains KG1275 (PdhbA-lacZ Δfur), KG1271 (PykuN-lacZ Δfur), and KG1283 (PfeuA-lacZ Δfur). (C) Differential rate plot using data from panel A. Strains NC209 (PdhbA-lacZ), KG46 (PykuN-lacZ), and KG19 (PfeuA-lacZ) were used. (D) The half-life of β-galactosidase activity was determined from cultures grown to low and high densities. Briefly, cells were grown to the indicated density and left untreated or treated with chloramphenicol to inhibit protein synthesis. Aliquots were removed every 30 min for measurement of β-galactosidase activity, and the half-life was determined. The half-lives of β-galactosidase activity are statistically insignificant between the different conditions. Experiments were performed in triplicate, and standard error is shown. LD, low density (OD600 ∼ 0.2); HD, high density (OD600 ∼ 1.5).
To further characterize the density-dependent regulation of iron responsive genes, a differential rate plot was prepared from the data in Fig. 3A. An increase in β-galactosidase synthesis was observed from all three reporters during exponential growth (1 to 3 h), as shown by the positive slope of the curves (Fig. 3C). This is consistent with cells experiencing iron limitation and attempting to acquire additional iron to maintain homeostasis. As cultures entered into stationary phase (4 to 6 h), the rate of β-galactosidase synthesis was reduced for all three reporter strains and β-galactosidase was actively turned over, as indicated by the negative slope of the curves (Fig. 3C).
Three factors contribute to reduced β-galactosidase activity as the reporter strains enter into stationary phase. First, as the iron acquisition activity accumulates in the growth medium, iron is internalized, and transcription of Fur-regulated genes is repressed, resulting in reduced β-galactosidase synthesis (Table 1 and Fig. 2A). Second, unlike in E. coli, β-galactosidase is unstable in B. subtilis. To demonstrate this effect, strain KG46 (PykuN-lacZ) was grown to low (OD600 ∼ 0.2) and high (OD600 ∼ 1.5) cell densities and treated with the translational inhibitor chloramphenicol, and aliquots were removed for measurement of β-galactosidase activity. Similar reductions in β-galactosidase activity were observed at both cell densities following treatment with chloramphenicol, resulting in half-lives of 145 min at low density and 123 min at high density (Fig. 3D, black and gray bars, respectively). Third, translation of the reporter gene decreases at high cell density, as observed by similar half-lives of β-galactosidase activity in the presence and absence of antibiotic treatment (Fig. 3D, dark gray and light gray bars, respectively). The decrease in translation is likely due to a combination of repressed transcription of PykuN-lacZ and turnover of the mRNA.
Taken together, these results indicate that Fur coordinates the transcription of iron-responsive genes with population density through the action of an extracellular iron acquisition activity. Specifically, transcription is derepressed at low cell density when cells are limited for extracellular iron acquisition molecules. At high cell density, when the culture transitions into stationary phase, sufficient iron acquisition molecules are present in the medium and transcription of target genes is repressed. The decrease in β-galactosidase activity observed from the reporter strains at high cell density is due to a combination of repressed transcription of the reporter by Fur, unstable β-galactosidase activity, and decreased translation presumably due to reduced mRNA levels.
Characterization of conditioned medium.The instability of β-galactosidase in B. subtilis should, in theory, make it an ideal reporter for studying transcriptional repression of Fur-regulated genes. To test this prediction, strain NC209 (PdhbA-lacZ) was grown to low cell density (OD600 = 0.2), treated with uninoculated medium or conditioned medium, and β-galactosidase activity was measured throughout the growth cycle. A 10-fold reduction in β-galactosidase activity was observed after treatment with conditioned medium compared to uninoculated medium, followed by a steady increase in activity throughout exponential growth (Fig. 4A). Although the magnitude of the response was significantly lower than that determined by quantitative RT-PCR (Fig. 2A), we can use the PdhbA-lacZ reporter to monitor repression of transcription.
Biochemical characterization of conditioned medium from strain JH642. (A) Add-back experiments were performed with strain NC209 (PdhbA-lacZ) using conditioned medium from strain JH642 and uninoculated medium. The percent activity was determined from the time point, typically 3 h, with the lowest β-galactosidase activity upon treatment with conditioned medium and normalized to the value for the corresponding time point with uninoculated medium. The results of a single representative experiment are shown to illustrate how percent activity was determined. (B) Conditioned medium was treated with DNase I, RNase A, proteinase K, 75°C for 20 min, storage at 4°C for 3 months, or passage through a Centricon-3 centrifugal device with a 3-kDa-molecular-mass-cutoff membrane (see Materials and Methods). Add-back experiments were performed with strain NC209 (PdhbA-lacZ) using treated conditioned and uninoculated media. Percent activity was determined as described above and normalized to the value for uninoculated medium with the same treatment. The response to each treatment was statistically insignificant compared to no treatment. Experiments were repeated in triplicate, and standard errors are shown. (C) Conditioned medium was prepared from cultures left untreated or sonicated to release their cellular components. The conditioned medium was diluted 1:2, 1:3, or 1:10 with S750 minimal medium as indicated in parentheses. Add-back experiments were performed using strain KG46 (PykuN-lacZ), and β-galactosidase activity was normalized to the value with untreated, uninoculated medium. Differences observed between the two conditioned medium preparations at each dilution were not statistically significant.
To begin to characterize the biomolecule(s) responsible for iron acquisition, conditioned medium from strain JH642 was treated as described below and add-back experiments were performed using strain NC209 (PdhbA-lacZ). Specifically, conditioned medium was treated with DNase I, RNase A, proteinase K, heat (75°C for 20 min), or long-term storage at 4°C (see Materials and Methods). None of the treatments had a significant effect on iron acquisition, indicating that the molecule(s) responsible for the observed activity was not likely a nucleic acid or a heat-labile protein (Fig. 4B). To determine the approximate molecular mass of the molecule(s) responsible for iron acquisition, conditioned medium was filtered using centrifugal concentration devices with a defined molecular mass cutoff (see Materials and Methods). The activity remained in the flowthrough following filtration with a 3-kDa-molecular-mass cutoff membrane, indicating that the molecular mass of the molecule(s) responsible for iron acquisition is less than 3 kDa (Fig. 4B).
It is possible that iron uptake is not due to an extracellular factor but is the result of free iron released from lysing or dead cells during the preparation of conditioned medium. To test this prediction, cells were grown to high density and disrupted by sonication to release their cellular contents, and conditioned medium was prepared as described in Materials and Methods. Add-back experiments were performed using the reporter strain KG46 (PykuN-lacZ), and β-galactosidase activity was normalized to that of untreated, uninoculated medium. As a control, sonication of uninoculated medium had no effect on β-galactosidase activity of strain KG46 (Fig. 4C, first series). An ∼10-fold decrease in β-galactosidase activity was observed with conditioned medium prepared from sonicated cells, similar to conditioned medium prepared from intact cells (Fig. 4C, second series). If the conditioned medium from intact cells was already saturated with iron, releasing additional iron from cells by sonication is not predicted to further repress transcription of PykuN-lacZ. Only if the available iron were diluted below saturation would we observe an effect on transcription. Thus, we prepared dilutions of the two conditioned media with S750 minimal media and repeated the add-back experiments. No significant difference in β-galactosidase activity was observed between untreated and sonicated media at each dilution (Fig. 4C, series 3 to 5). We conclude that the iron acquisition activity observed in strain JH642 was not due to iron released from dying or dead cells.
Effects of Fur and extracellular iron on iron acquisition activity.In many bacteria, transcription of iron acquisition molecules and iron uptake pathways are directly regulated by Fur in response to available iron (39). If transcription of the gene(s) required for the iron acquisition activity from strain JH642 is also directly regulated by Fur, then relieving the negative regulation by Fur is predicted to increase its production. To test this prediction, conditioned medium was harvested from strains JH642 and KG60 (Δfur), add-back experiments were performed on strain JH642, and transcription of the Fur-regulated dhbA gene was measured using quantitative RT-PCR (see Materials and Methods).
Conditioned medium isolated from high-density cultures of strains JH642 and KG60 (Δfur) repressed transcription of dhbA to similar extents (Fig. 5A, top). It is possible that the iron acquisition activity from strain JH642 is saturated at high cell density. If so, a greater effect might be observed if conditioned medium was prepared from low-density cultures when the concentration of iron acquisition molecules is predicted to be low. The add-back experiments were repeated using conditioned medium from low-density cultures. However, no statistical difference in dhbA transcription was observed following treatment with conditioned medium prepared from low-density cultures of either strain (Fig. 5A, bottom).
Effect of Fur repressor and extracellular iron on iron acquisition activity. (A) Add-back experiments were performed on strain JH642 using uninoculated medium (UM), conditioned medium from strain JH642 (wild type [WT]), or conditioned medium from strain KG56 containing a fur deletion (Δfur). Conditioned medium was isolated from cultures grown to high (OD600 ∼ 1.5) and low (OD600 ∼ 0.2) densities (top and bottom graphs, respectively). The abundance of dhbA mRNA was measured using quantitative RT-PCR. Experiments were performed in triplicate, and percent standard error is shown. Asterisks represent P values of <0.01 between the two treatments (WT versus Δfur mutant) at each time point. (B) Strain KG46 (PykuN-lacZ) was grown in minimal medium containing different concentrations of FeCl3, aliquots were removed every hour, and β-galactosidase specific activity was determined. The average generation time of strain KG46 grown in the presence of 5 μM FeCl3 was 50 min, compared to 110 min in the absence of FeCl3. Note that 5 μM FeCl3 was used in the experiments described throughout this article. Experiments were performed in triplicate, with similar results. Results of a single representative experiment are shown.
Alternatively, it is possible that the iron acquisition activity is, indeed, regulated by Fur, but cells are starved for iron in the presence of 5 μM FeCl3, used in these experiments. If this were the case, derepression of Fur-regulated genes would occur under these growth conditions, and relieving the negative regulation by Fur would have minimal impact on the transcription of target genes. To determine whether iron is limited under these growth conditions, the reporter strain KG46 (PykuN-lacZ) was grown in minimal medium supplemented with different amounts of FeCl3, and transcription was monitored throughout the growth cycle by measurement of β-galactosidase activity.
The concentration of FeCl3 in the growth medium had a significant impact on the transcription of PykuN-lacZ. As the concentration of FeCl3 decreased, PykuN-lacZ transcription was derepressed, resulting in increased β-galactosidase activity. Interestingly, the density-dependent regulation of PykuN-lacZ occurred at all concentrations of FeCl3, except 50 μM, which completely repressed transcription (Fig. 5B). Strain KG46 is clearly not limited for iron in the presence of 5 μM FeCl3, as shown by a 5-fold reduction in β-galactosidase activity compared to that in medium lacking FeCl3 (Fig. 5B). The concentration of FeCl3 in the growth medium also had a profound effect on the growth rate of strain KG46. As the concentration of FeCl3 decreased, the generation time of strain KG46 increased. Specifically, a 2-fold increase in the growth rate was observed for cultures grown without FeCl3 (110 min) compared to that at 50 min in medium containing 5 μM FeCl3 (Fig. 5B). Based on the observed growth rate and the transcriptional response of PykuN-lacZ, we conclude that strain KG46 is not limited for iron with 5 μM FeCl3 present in the growth medium, a standard concentration of iron used for growing B. subtilis. To our knowledge, this is the first example of an iron acquisition activity that is regulated independently of Fur repressor and is present under iron-replete conditions.
Effect of receptor mutants on iron acquisition. B. subtilis has multiple acquisition systems for internalizing Fe(III)-bacillibactin and Fe(III)-xenosiderophores produced from a variety of different microbes (27). To determine if any of these pathways are involved in acquiring iron by strain JH642, we obtained genetic deletions in each of the uptake systems (kind gift from J. Helmann). Individual deletions were introduced into the reporter strain KG46 (PykuN-lacZ) and tested for the ability to utilize iron from conditioned medium prepared from strain JH642 using the add-back experiments.
An ∼10-fold decrease in β-galactosidase activity was observed for wild-type strain KG46 following the addition of conditioned medium compared to that in uninoculated medium (Fig. 6A). Similar effects were observed in strains with deletions of yfhA (schizokinen and anthrobactin uptake), fhuB (ferrioxamine and ferrichrome uptake), feuA (bacillibactin uptake), and yfmC [Fe(III)-citrate uptake]. These Fe(III)-xenosiderophore uptake systems are clearly not involved in the utilization of iron from conditioned medium prepared from strain JH642 (Fig. 6A).
Effects of single receptor mutants on iron uptake from conditioned medium. Add-back experiments were performed on reporter strains (PykuN-lacZ) containing single mutations in different Fe(III)-xenosiderophore uptake systems. Experiments were performed in triplicate, with similar results. UM, uninoculated medium; CM, conditioned medium. (A) Strains contain PykuN-lacZ and the following mutations: KG46 (wild type), KG1754 (ΔyfhA), KG1755 (ΔfhuB), KG1752 (ΔfeuA), KG1757 (ΔyfmC), and KG1748 [ΔefeU (ΔywbL)]. Conditioned medium was prepared from strain JH642. Percent activity was determined for each reporter strain as described for Fig. 4A. Similar β-galactosidase activities were observed for each reporter strain when treated with uninoculated medium; the exception was strain KG1748, which exhibited a 20-fold increase in β-galactosidase activity compared to that of the wild type. Experiments were performed in triplicate, and percent standard error is shown. The asterisk represents a P value of <0.01 relative to the value for the wild type. (B) Results of a representative add-back experiment performed on strain KG1748 [PykuN-lacZ ΔefeU (ΔywbL)] are shown. Strain KG1748 was grown to an OD600 of ∼0.1 and treated with the following: uninoculated medium (solid circles), conditioned medium from strain JH642 (squares); conditioned medium from strain KG56 (ΔdhbABCDE) (triangles), and uninoculated medium supplemented with 340 μM citrate (open circles). For comparison, wild-type strain KG46 (PykuN-lacZ) was treated with uninoculated medium (diamonds). β-Galactosidase activities are shown at the top and growth curves at the bottom.
On the other hand, strain KG1748 (PykuN-lacZ ΔefeU), containing a deletion in the first gene of the efeUOB (ywbLMN) operon, exhibited signs of extreme iron starvation. First, an ∼20-fold increase in β-galactosidase activity was observed in strain KG1748 compared to that in the wild type, indicating significant derepression of the Fur-regulated PykuN-lacZ. Second, growth was arrested in strain KG1748 as cultures reached a maximal OD600 of ∼0.2 (Fig. 6B, solid circles). The addition of conditioned medium to strain KG1748 slightly improved the growth defect, enabling cells to reach an OD600 of ∼1, but cells remained starved for iron, as shown by high β-galactosidase activity (Fig. 6B, squares). This was mainly due to the presence in the conditioned medium of itoic acid, which has weak iron-chelating activity and is imported through the FeuABC-YusV complex (31–33). We created a strain unable to synthesize itoic acid by deleting the dhbABCDE operon (see Materials and Methods). Growth of strain KG1748 (ΔefeU) was arrested at an OD600 of 0.1 to 0.2, and cells were starved for iron following the addition of conditioned medium from the itoic acid-deficient strain KG56 (ΔdhbABCE) compared to the wild-type reporter strain KG46 treated with uninoculated medium (Fig. 6B, triangles and diamonds, respectively).
High concentrations of citrate function as an effective iron chelator. Ollinger et al. showed that Fe(III)-citrate is imported into B. subtilis via the YfmCDEF complex (27). Consistent with these findings, the addition of exogenous citrate to the growth medium of strain KG1748 (PykuN-lacZ ΔefeU) repressed transcription of PykuN-lacZ and restored the growth rate to wild type levels (Fig. 6B, open circles). Taken together, we conclude that increased transcription of PykuN-lacZ and the growth defects observed in strain KG1748 (ΔefeU) are consistent with the cell's inability to import sufficient quantities of iron to support cellular processes. The iron acquisition activity present in conditioned medium of strain JH642 appears to work in combination with the EfeUOB complex to acquire biologically available iron under these growth conditions.
Previous work by Meithke et al. demonstrated that free, elemental Fe(II) and Fe(III) are transported into cells via the gene products of the efeUOB operon (30). The presence of an iron reductase could explain the activity observed in conditioned medium. Specifically, Fe(III) could be reduced to Fe(II) for subsequent import into cells. The ferrozine assay, described by Lovely and Phillips (40), was used to measure Fe(II) present in conditioned medium of strain JH642. However, no Fe(II) was detected in conditioned medium even after 100-fold concentration (data not shown). This result was somewhat expected since any soluble Fe(II) would be rapidly and spontaneously oxidized to Fe(III) during aerobic growth. Alternatively, iron reduction could occur at the membrane, followed by immediate transport of Fe(II) or Fe(III) into cells that would be undetectable by the ferrozine assay.
The identity of the molecule(s) responsible for the iron acquisition activity and the form of iron transported into cells remains unknown at this time. It is possible that Fe(II) is produced by an iron reductase with free Fe(II) or rapidly oxidized free Fe(III) transported directly into cells by the EfeUOB complex. Alternatively, Fe(II) or Fe(III) bound by a chelating molecule, e.g., a siderophore, could also account for the observed activity. Future experiments are required to distinguish between these possibilities. Regardless of the molecule's identity, a secreted factor(s) is clearly responsible for providing cells with biologically available iron that would otherwise be inaccessible under these growth conditions. Thus, we named the molecule(s) responsible for the iron acquisition activity elemental Fe(II/III) (Efe) acquisition factor because of the requirement for EfeUOB complex.
Production of Efe acquisition factor in undomesticated and domesticated strains.Undomesticated strains of B. subtilis produce the siderophore bacillibactin (BB) (31–33). To begin to determine whether undomesticated strains also produce Efe acquisition factor, we constructed strain EMR130 (ΔdhbABCDE), an isogenic derivative of the undomesticated strain DS7187 containing a deletion of the dhbABCDE operon that is unable to produce bacillibactin or its precursor itoic acid (see Materials and Methods). Conditioned medium was prepared from strains DS7187 (wild type) and EMR130 (ΔdhbABCDE), which produce bacillibactin (BB+) and lack bacillibactin (BB−), respectively. Add-back experiments were performed with the reporter strain KG46 (PykuN-lacZ) and derivatives of strain KG46 that contain mutations in the uptake systems for bacillibactin (feuA) or Efe acquisition factor (efeU).
Parental strain KG46 (PykuN-lacZ) internalized iron present in conditioned media from both strains DS7187 (BB+) and EMR130 (BB−), repressing transcription of PykuN-lacZ, as shown by a 10-fold reduction in β-galactosidase activity (Fig. 7A, first series). Similarly, conditioned medium from strain EMR130 (BB−) also repressed transcription of PykuN-lacZ in the feuA mutant reporter strain (Fig. 7A, second series, black bar). In contrast, the efeU mutant reporter strain was unable to utilize iron from conditioned medium produced from strain EMR130 (BB−), resulting in high β-galactosidase activity (Fig. 7A, third series, black bar). These results are consistent with strain EMR130 (BB−) producing Efe acquisition factor.
Effects of bacillibactin on the activity of Efe acquisition factor. Add-back experiments were performed on wild-type strain KG46 (PykuN-lacZ) and the reporter strain containing individual deletions in feuA (KG1752) and efeU (KG1748), which are defective for bacillibactin (BB) and Efe acquisition factor (Efe) uptake, respectively. Percent activity was determined for each strain as described for Fig. 4A. Experiments were performed in triplicate, and standard error is shown. Asterisks represent P values of <0.01 between the two conditions for each strain. (A) Conditioned medium from undomesticated bacillibactin-defective (BB−) strain EMR130 (ΔdhbA-E::erm) and from the undomesticated bacillibactin-producing (BB+) strain DS7187 (NCIB 3610 pBS32::ΔcomI). (B) Conditioned medium from bacillibactin-defective laboratory strain JH642 (sfp0) and from bacillibactin-producing laboratory strain EMR135 (sfp0amyE::Psfp-sfp+::Cm). (C) Conditioned medium from enterobactin-defective (EB−) E. coli strain EMR126(ΔentC::kan) and from the wild-type enterobactin-producing strain MG1655.
Conditioned medium from strain DS7187 (BB+) repressed transcription of PykuN-lacZ containing the efeU mutation (Fig. 7A, third series, gray bar). This was expected since previous work demonstrated that Fe(III)-bacillibactin is imported into cells via the FeuABC-YusV complex (27). Interestingly, conditioned medium from strain DS7187 (BB+) failed to repress transcription of PykuN-lacZ in the feuA mutant reporter strain (Fig. 7A, second series, gray bar). Taken together, these results indicate that undomesticated, bacillibactin-producing strains of B. subtilis (e.g., DS7187) fail to produce Efe acquisition factor or that its activity is perturbed by bacillibactin.
We sought to determine whether the regulation of Efe acquisition factor by bacillibactin was specific to undomesticated strains of B. subtilis or common to the domesticated laboratory strain JH642. Laboratory strain JH642 (sfp0) is unable to produce bacillibactin due to mutations in the sfp gene. Replacing sfp0 with a functional sfp allele (sfp+) restores bacillibactin production in strain JH642 (31–33). The wild-type sfp allele, under the control of its own promoter (amyE::Psfp-sfp+), was transformed into strain JH642 (kind gift from D. Kearns). Conditioned medium was prepared from isogenic strains EMR135 (sfp+) and JH642 (sfp0), which produce bacillibactin and lack bacillibactin, respectively. Similar add-back experiments were performed as described above using conditioned medium from strains EMR135 and JH642.
The wild-type reporter strain KG46 (PykuN-lacZ) utilized iron present in conditioned media from both strains EMR135 (BB+) and JH642 (BB−) to repress transcription of PykuN-lacZ, as shown by a 10-fold reduction in β-galactosidase activity compared to that in uninoculated medium (Fig. 7B, first series). Similar to the results shown in Fig. 7A, transcription of PykuN-lacZ was repressed in the feuA mutant with conditioned medium from strain JH642 but not strain EMR135 (Fig. 7B, second series). Likewise, the efeU mutant reporter strain was able to utilize iron present in conditioned medium from strain EMR135 but not from strain JH642 (Fig. 7B, third series). These results indicate that the regulation of Efe acquisition factor by bacillibactin is conserved among domesticated and undomesticated strains of B. subtilis.
Production of Efe acquisition factor from E. coli. E. coli produces the catechol siderophore enterobactin (EB) (34). We determined whether Efe acquisition factor is specific to B. subtilis or conserved among other bacteria, including E. coli. Conditioned medium was isolated from wild-type E. coli (EB+) and strain EMR126 (ΔentC), a mutant deficient for enterobactin production (EB−). Add-back experiments were conducted using the B. subtilis reporter strain KG46 (PykuN-lacZ) and isogenic derivatives containing mutations in feuA or efeU, defective for uptake and utilization of enterobactin and Efe acquisition factor, respectively.
The wild-type B. subtilis reporter strain KG46 (PykuN-lacZ) utilized iron present in conditioned media isolated from both strains of E. coli, as shown by a 8- to 9-fold reduction in β-galactosidase activity compared to that in uninoculated medium (Fig. 7C, first series). The feuA mutant reporter strain internalized the iron from conditioned medium from E. coli strain EMR126 (EB−), but not from wild-type E. coli (EB+), to repress transcription of PykuN-lacZ (Fig. 7C, second series). Similarly, transcription of PykuN-lacZ was repressed in the efeU mutant reporter strain when treated with conditioned medium from wild-type E. coli but not the enterobactin-deficient strain EMR126 (Fig. 7C, third series). Taken together, the results lead us to conclude that similar Efe acquisition activities are present in E. coli and B. subtilis, iron uptake is dependent on the EfeUOB complex, and the activity or production of Efe acquisition factor is inhibited by catecholic siderophores enterobactin and bacillibactin.
Regulation of Efe acquisition factor by bacillibactin and enterobactin.Bacillibactin and enterobactin have high affinities for iron (34). It is possible that these catecholic siderophores modulate the activity of Efe acquisition factor by sequestering available iron from the medium. To test this prediction, conditioned medium was prepared from the bacillibactin-producing strain EMR135 (BB+) and strain JH642 (BB−), which is unable to synthesize bacillibactin. Combinations of the two media were used in add-back experiments with the reporter strain KG1752 (PkyuN-lacZ ΔfeuA), which is defective for uptake of Fe(III)-bacillibactin but not Efe acquisition factor (see Materials and Methods).
As observed previously, the addition of conditioned medium from strain JH642 (BB−) but not strain EMR135 (BB+) repressed transcription of PykuN-lacZ, resulting in reduced β-galactosidase activity compared to that of uninoculated medium (Fig. 8A, series 1 to 3). Interestingly, the reporter strain KG1752 (PykuN-lacZ ΔfeuA) was unable to utilize iron from conditioned medium combined from strains JH642 and EMR135 at a 1:1 ratio, resulting in high β-galactosidase activity, similar to treatment with conditioned medium from strain EMR135 (Fig. 8A, series 4). As the amount of conditioned medium derived from strain EMR135 was reduced compared to that with strain JH642, transcription of PykuN-lacZ was repressed, resulting in reduced β-galactosidase activity (Fig. 8A, series 5 to 7). Thus, we conclude that some factor, most likely bacillibactin, produced from strain EMR135 inhibits the activity of Efe acquisition factor.
Regulation of Efe acquisition factor by bacillibactin and enterobactin. (A) Add-back experiments were performed with strain KG1752 (PykuN-lacZ ΔfeuA) using conditioned medium isolated from isogenic strains JH642 (BB−) and EMR135 (BB+). Different ratios of conditioned medium were prepared from the two strains where “1” represents a volume equal to that of the reporter culture. Percent activity was determined as described for Fig. 4A. Experiments were repeated in triplicate, and standard error is shown. (B) Add-back experiments were performed with strains KG1748 (PykuN-lacZ ΔefeU) and KG1752 (PykuN-lacZ ΔfeuA). The minimal concentration of purified enterobactin that repressed transcription of PykuN-lacZ from strain KG1748 was determined empirically to be 0.6 μM. CM (BB−), conditioned medium from bacillibactin-deficient strain JH642; CM (BB+), conditioned medium from bacillibactin-producing strain EMR135; EB, purified enterobactin purchased from Sigma used at the indicated concentration. Experiments were repeated in triplicate, with similar results. Results from a representative experiment are shown.
It is possible that a factor other than bacillibactin affects the activity of Efe acquisition factor. To test this prediction, the add-back experiments were repeated using purified enterobactin from E. coli. We rationalize that enterobactin serves a good substitute for bacillibactin, since the two molecules are structurally similar, have similar binding affinities for Fe(III), and are imported into B. subtilis via the FeuABC-YusV complex (27, 31–33). Purified enterobactin was added to uninoculated medium in an add-back experiment using strain KG1748 (PykuN-lacZ ΔefeU). Iron acquisition was monitored by measurement of β-galactosidase activity (see Materials and Methods). A concentration of 0.6 μM enterobactin was sufficient to restore growth of strain KG1748 and maximally repress transcription of PykuN-lacZ, similar to treatment with conditioned medium from the bacillibactin-producing strain EMR135 (Fig. 8B, left graph, open circles and solid triangles, respectively). As a control, the same concentration of enterobactin had no effect on transcription of PykuN-lacZ in strain KG1752 containing a deletion of feuA (Fig. 8B, right graph, solid squares). This is consistent with previously published results by Ollinger et al. indicating that FeuABC-YusV is required for Fe(III)-enterobactin utilization by cells (27).
Interestingly, strain KG1752 (PykuN-lacZ ΔfeuA) was able to utilize Efe acquisition factor from conditioned medium to repress PykuN-lacZ, despite the presence of 0.6 μM enterobactin (Fig. 8B, right graph, open squares). However, concentrations of enterobactin greater than 0.6 μM inhibit Efe acquisition factor and alter the cell's ability to acquire iron in the feuA mutant, resulting in derepression of PykuN-lacZ. Specifically, a 5-fold increase in β-galactosidase activity was observed at an OD600 of ∼1 when 1.2 μM enterobactin was added to conditioned medium from strain JH642 compared to conditioned medium treated with 0.6 μM enterobactin (Fig. 8B, right graph, open triangles versus open squares). Cells exhibit signs of iron starvation at even higher concentrations of enterobactin (6 μM), as shown by increased β-galactosidase activity and reduced growth rate (Fig. 8B, right graph, open and solid diamonds).
Taken together, these results indicate that physiological concentrations of bacillibactin inhibit iron uptake by Efe acquisition factor, presumably by sequestering the limited Fe(III) available to cells. Only when the synthesis of bacillibactin is perturbed, e.g., in strains JH642 (sfp0dhbABCDE+) and EMR130 (sfp+ ΔdhbABCDE), does Efe acquisition factor function in combination with the EfeUOB complex to satisfy the cell's requirement for iron.
DISCUSSION
In this work, we identified a novel iron acquisition activity produced by B. subtilis. We named the factor responsible for this activity elemental Fe(II/III) (Efe) acquisition factor because the EfeUOB complex is required for iron acquisition under these conditions. Unlike most iron utilization systems that are regulated by Fur in response to available iron, Efe acquisition factor is present under iron-replete conditions and is regulated independently of Fur. This work provides new insight into the reciprocal regulation of Fur-dependent and Fur-independent iron acquisition systems. We expand on the model of iron acquisition in B. subtilis to include Efe acquisition factor and its regulation by bacillibactin. The implications of dual iron acquisition systems, and the conservation of efeUOB and Efe acquisition factor among bacteria and fungi, are discussed.
Efe acquisition factor could be a siderophore or an iron reductase.In B. subtilis, elemental Fe(II) and Fe(III) are transported directly into cells via the EfeUOB complex (28, 30). The use of iron reductases to reduce Fe(III) to Fe(II) and iron-chelating siderophores to sequester available Fe(III) are two strategies bacteria have evolved to acquire iron (2, 18, 26). Siderophores are low-molecular-mass molecules (<1 kDa) that are synthesized nonribosomally from chemical precursors, e.g., 2,3-dihydroxybenzoic acid, with a high affinity for Fe(III) (31–33). In contrast, iron reductases are typically larger proteins or protein complexes (41). Biochemical characterization of the iron acquisition activity present in conditioned medium indicates that the molecule(s) responsible for the observed activity has a low molecular weight (<3 kDa), is nonproteinaceous, and is resistant to heat (Fig. 4B). These physical properties, coupled with the biologically available form of iron present during aerobic growth conditions, e.g., Fe(III), are consistent with an iron-chelating siderophore.
Alternatively, an iron reductase would also explain the observed iron acquisition activity present in conditioned medium of strain JH642. Shewanella spp. secrete small flavin molecules (∼350 Da) into the growth medium that have been shown to reduce Fe(III)-oxides for subsequent uptake into cells (42). Although no detectable Fe(II) was observed in conditioned medium of strain JH642, it is possible that Fe(III) is reduced, but the concentration of Fe(II) is below the level of detection of the assay (0.5 μM). Alternatively, iron reduction could occur at the membrane, followed by immediate transport of elemental Fe(II) and/or Fe(III) into cells via the EfeUOB complex that would be missed with the ferrozine assay.
At this time, we are unable to determine whether the source of iron entering cells is free, elemental Fe(II) produced from an iron reductase or elemental Fe(III) that spontaneously oxidized as a result of iron reduction. Moreover, we are unable to determine whether the observed activity is due to iron bound to a chelator, e.g., a Fe(III)-siderophore. Interestingly, Efe acquisition factor and the catecholic siderophores appear to utilize a common form of iron, e.g., Fe(III), since physiological concentrations of bacillibactin and enterobactin inhibit iron uptake via the EfeUOB complex, presumably by sequestering away available iron (Fig. 8). Regardless of the oxidation state of iron or the identity of the molecule(s) responsible for iron import into cells, it is clear this novel iron acquisition activity plays an important role in providing a biologically active form of iron that can be utilized by cells to maintain iron homeostasis.
Model for regulation of Efe acquisition factor by bacillibactin.Undomesticated strains of B. subtilis produce the siderophore bacillibactin (2,3-dihydroxybenzoyl glycine-threonine). Many laboratory strains, including JH642 used in these studies, fail to produce bacillibactin because of mutations in sfp. However, strain JH642 produces itoic acid (2,3-dihydroxybenzoyl glycine), a precursor to bacillibactin with weak iron-chelating activity (32, 35). Bacillibactin, but not itoic acid, inhibits the activity of Efe acquisition factor, presumably by sequestering available iron (Fig. 8 and 9A, left). Efe acquisition factor acts as a fail-safe for successful iron acquisition should the Fur-mediated system become inactivated due to genetic mutation. This is most obvious in strain KG1748 [sfp0 ΔefeU (ΔywbL)], which is unable to synthesize bacillibactin but also lacks the ability to import Efe acquisition factor. Strain KG1748 exhibits classic signs of iron starvation, even in the presence of 5 μM FeCl3 in the growth medium. First, cells have a severe growth defect, with complete growth arrest occurring around an OD600 of ∼0.2. Second, transcription of Fur-regulated genes is maximally derepressed as cells attempt to acquire additional iron to maintain cellular processes (Fig. 6B, solid circles). Itoic acid has reported weak iron-chelating activity (31–33). Although strain KG1748 produces itoic acid, it is insufficient to sustain culture growth (Fig. 6B, solid circles). The iron-deprived state of strain KG1748 can be rescued by treating cells with exogenous iron chelators, e.g., citrate and enterobactin. Immediately following the addition of chelator to the growth medium, cells resume growth and Fur-regulated genes are repressed as the cell's requirement for iron is satisfied (Fig. 6B and 8).
Model for regulation of Efe acquisition factor by bacillibactin and density-dependent regulation of the iron response. (A) Bacillibactin acts as the primary siderophore used by B. subtilis, while Efe acquisition factor serves as a backup. The activity of Efe acquisition factor is inhibited by bacillibactin, presumably by sequestering available iron. In the absence of bacillibactin, Efe acquisition factor serves as the sole iron acquisition system. (B) In the absence of bacillibactin, secretion of Efe acquisition factor coordinates the iron response with population density. At low cell density when the concentration of Efe acquisition factor is low, transcription of Fur-regulated genes is derepressed as cells attempt to acquire additional iron to maintain homeostasis. Efe acquisition factor accumulates in the growth medium as the culture progresses through the growth cycle. At high cell density, sufficient Efe acquisition factor is present and transcription of Fur-regulated genes is repressed. Among the genes repressed by Fur is the efeUOB operon, whose gene products are required for the import of Efe acquisition factor. Dashed lines represent transcriptional control, while solid lines represent posttranslational control.
In B. subtilis, bacillibactin serves as the primary siderophore for iron acquisition. Fur directly regulates transcription of the dhbABCDE operon, which encodes the bacillibactin biosynthetic machinery. In addition, Fur regulates the transcription of the feuABC and yusV genes, whose protein products comprise the Fe(III)-bacillibactin uptake complex (11, 43). By controlling bacillibactin production and its uptake with a common regulatory factor, i.e., Fur, cells can effectively coordinate iron uptake with available iron to maintain iron homeostasis. In the absence of bacillibactin, iron availability and uptake are decoupled as Efe acquisition factor is produced under iron-replete conditions (Fig. 8 and 9A, right). At first glance, this would appear to be energetically costly to the cell, but without this activity, cells are starved for iron and unable to maintain basic cellular processes even with the presence of 5 μM FeCl3 in the growth medium (Fig. 6B, solid circles).
Density-dependent regulation of the iron response by Efe acquisition factor.In Vibrio vulnificus, quorum sensing and Fur repressor work in concert to coordinate the iron response with available iron and population density. The result is increased expression of the siderophore vulnibactin at low cell density when cells are actively growing and require additional iron, followed by reduced expression of vulnibactin at high cell density when cells enter into stationary phase and the requirement for iron is less critical (44).
A similar strategy is used by laboratory strains of B. subtilis to coordinate the iron response with population density. However, instead of utilizing the quorum response, accumulation of Efe acquisition factor in the growth medium provides a way to coordinate transcription of iron-responsive genes with population density. At low cell density, when the concentration of Efe acquisition factor is low, cells are limited for iron, and transcription of Fur-regulated genes is derepressed as actively growing cells attempt to acquire additional iron (Fig. 3A and 9B, left). As the culture progresses through the growth cycle, Efe acquisition factor accumulates in the growth medium. At high cell density, when cells are less metabolically active, the concentration of Efe acquisition factor is high and transcription of Fur-regulated genes is repressed (Fig. 3A and 9B, right). Although the production of Efe acquisition factor is not directly regulated by Fur per se (Fig. 5A), transcription of the efeUOB operon is directly regulated by Fur (13). Repression of efeUOB by Fur is predicted to limit the amount of iron imported into cells, enabling cells to prevent the accumulation of excess iron and leading to the formation of reactive oxygen species and cellular damage.
Conservation of Efe acquisition factor among bacteria and yeasts.The efeUOB operon is highly conserved among many bacteria and yeasts (28–30). Until now, the role of an associated iron acquisition factor has been overlooked. Efe acquisition factor is produced in B. subtilis and E. coli, and its reciprocal regulation by Fur-mediated catecholic siderophores, e.g., bacillibactin and enterobactin, is common to both species of bacteria (Fig. 7). Given the conservation of the efeUOB operon and Fur-regulated iron acquisition systems among bacteria, it is interesting to consider that a similar Efe acquisition factor is produced by other microbes. Since Fur-regulated iron acquisition systems modulate the activity of Efe acquisition factor, at least in B. subtilis and E. coli, this might explain why the observed activity has eluded identification in other bacteria. Many bacteria and fungi lack Fur-mediated, or equivalent, iron acquisition systems. Perhaps these microbes rely solely on Efe acquisition factor to acquire iron.
MATERIALS AND METHODS
Growth media.Liquid cultures were grown in S7 defined minimal medium salts (45) containing 50 mM 4-morpholinepropanesulfonic acid instead of 100 mM (S750) and supplemented with 1% glucose, 0.1% glutamate, and the following trace metals: 2 μM MgCl2, 0.7 μM CaCl2, 50 μM MnCl2, 1 μM ZnCl2, 1 μg/ml of thiamine, and 5 μM FeCl3. Solid medium plates contained Spizizen salts (46) supplemented with 1% glucose and 0.1% glutamate. For growth of B. subtilis and E. coli, minimal medium was supplemented with tryptophan (40 μg/ml), phenylalanine (40 μg/ml), and threonine (200 μg/ml), where appropriate. LB agar plates were used for routine cloning and growth of B. subtilis and E. coli. The following antibiotics were used: ampicillin (100 μg/ml), neomycin (2.5 μg/ml), tetracycline (12 μg/ml), phleomycin (8 μg/ml), chloramphenicol (5 μg/ml), spectinomycin (100 μg/ml), and erythromycin (0.5 μg/ml) and lincomycin (12.5 μg/ml) together to select for macrolidelincosamide-streptogramin B resistance.
Growth conditions.Overnight cultures grown as light lawns on Spizizen minimal medium plates (46) were used to inoculate shaker flasks containing S750 minimal medium at a final OD600 of ∼0.02. Cultures were incubated at 37°C with vigorous aeration. For the experiments whose results are shown in Fig. 5B, S750 minimal medium was first prepared without FeCl3. The appropriate amount of FeCl3 was then added to the medium. Note that some iron was still present in the medium without FeCl3, as care was not taken to completely remove all residual iron from the glassware.
To prepare conditioned medium, cultures were grown to the appropriate cell density (determined by OD600) and cells were pelleted by centrifugation at 5,000 × g for 10 min at 4°C. Typically, low-density cultures were grown to an OD600 of ∼0.2, while high-density cultures were grown to an OD600 of ∼1 to 2. The liquid medium was filtered into a sterile bottle using a 0.2-μm-pore-size surfactant-free cellulose acetate membrane bottle top filter (Thermo Scientific). Cell-free conditioned medium and uninoculated medium, treated the same way, were stored at 4°C until further use.
Add-back experiments using conditioned medium were performed by inoculating an overnight test culture in freshly prepared S750 minimal medium at a final OD600 of ∼0.02. The culture was allowed to grow at 37°C with vigorous aeration until an OD600 of ∼0.2 was reached. An equal volume of culture was added to two separate flasks that were preequilibrated to 37°C with either conditioned medium or uninoculated medium. Growth was allowed to resume by incubation at 37°C with vigorous aeration. Samples were taken at the specified times for RNA isolation or assay of β-galactosidase activity.
Plasmids, strains, and alleles. Escherichia coli strain MG1655 (K-12 F− λ−ilvG rfb-50 rph-1) was used in the add-back experiments whose results are shown in Fig. 7C. Bacillus subtilis strains in Table 2 were derived from the parental laboratory strain JH642 (trpC2 pheA1) (47) or the undomesticated, prototrophic strain DS7187, a derivative of NCIB 3610 that contains a deletion of comI on plasmid pBS32 (48). E. coli strains DH5α and AG1111, an MC1061 derivative [F′ (lacIq) lacZΔM15 Tn10], were used for routine cloning.
Strains used in this study
Promoter-lacZ fusions.Promoter fusions to lacZ were constructed and integrated into the amyE locus of B. subtilis by homologous recombination. Briefly, primers were used to amplify by PCR the first ∼500 bp upstream of the beginning of the coding sequence and into the coding sequences of the genes fhuD, ykuN, and feuA. A termination codon (TAA) was engineered after the first 10 amino acid codons of each gene. EcoRI and BamHI restriction enzyme recognition sites were incorporated into the forward and reverse primers, respectively. PCR products were digested with EcoRI and BamHI restriction enzymes (NEB) and ligated with T4 DNA Ligase (NEB) into plasmid pKS2 (49). Ligation reactions were transformed into E. coli strain DH5α or AG1111 and the strain was plated on LB solid medium with ampicillin. Clones were verified by DNA sequencing (UMASS Core Facility). Plasmid DNA was linearized by digestion with NcoI restriction enzyme (NEB) and transformed into B. subtilis strain JH642 by a two-step procedure described by Boguslawski et al. (50), and the strain was plated on LB solid medium with neomycin.
Gene disruptions.Insertion disruptions of fur and dhbABCDE were constructed and integrated into the B. subtilis chromosome. Briefly, pGEMcat::erm was created by amplifying the promoter region and erythromycin resistance gene from pCAL215 (51). The DNA fragment was directionally cloned into pGEMcat by digestion with restriction enzymes BamHI and XbaI, followed by ligation with T4 DNA ligase (NEB). Ligation reactions were transformed into E. coli strain DH5α or AG1111, and the strain was plated on LB solid medium with ampicillin. Clones were verified by DNA sequencing (UMASS Core Facility). The construct was designated pGEMcat::erm.
Plasmid pGEMcat::erm was used to create gene disruptions in fur and dhbABCDE. Briefly, primers were used to amplify ∼500 bp upstream and downstream of each coding sequence. Restriction sites were engineered into the primers for directional cloning of the DNA fragments into pGEMcat::erm. Ligation reactions were transformed into E. coli, and the organism was plated on LB solid medium with ampicillin. Clones were verified by DNA sequencing (UMASS Core Facility). Plasmid DNA was linearized by digestion with NcoI restriction enzyme (NEB) and transformed into B. subtilis strain JH642, and the strain was plated on LB solid medium with erythromycin.
Oligonucleotides.All oligonucleotides used in this study were synthesized and desalted by Integrated DNA Technologies. Sequences are available upon request. Primers for quantitative PCR (qPCR) were designed using Primer 3 (52, 53). We used an optimum size of 20 nucleotides, a melting temperature of 60°C, a PCR product size of 75 to 200 bp, and the remaining default settings.
Computational analyses.Statistical analyses were performed using Microsoft Excel and the suite of statistical tools from In silico (http://www.in-silico.net ). Double-sample Student's t tests were performed, and a P value of 0.01 was used as a cutoff for statistical significance.
Microarray analyses.Add-back experiments using strain JMS122 (ΔcomA::cat) and conditioned medium from strain JH642 or uninoculated medium were conducted as described above, with minor modifications. Briefly, an overnight culture of strain JMS122 (ΔcomA::cat) was grown with vigorous aeration at 37°C until OD600 of ∼0.2 was reached. An equal volume of culture was added to two separate flasks that were preequilibrated to 37°C with either conditioned medium or uninoculated medium. The cultures were incubated at 37°C with vigorous aeration for 30 min. A 25-ml aliquot of culture was removed and added to 25 ml of cold methanol. Cells were harvested and total RNA was prepared as previously described (54). Briefly, cells were pelleted by centrifugation at 12,000 × g for 20 min at 4°C, and the supernatant was discarded. The cell pellets were resuspended in cold methanol, pelleted by centrifugation at 12,000 × g for 5 min at 4°C, and stored at −20°C until further use. Total RNA was isolated using an RNAeasy kit as described by the manufacturer (Qiagen).
RNA from each sample was reverse transcribed and labeled as previously described (38). A reference cDNA sample was prepared by combining equal amounts of the two labeled cDNA pools prepared from uninoculated and conditioned media. Labeled cDNA from each experimental sample was hybridized with the labeled reference cDNA to 65-mer oligonucleotides on the microarray slide that correspond to each of the ∼4,200 genes in B. subtilis. Microarray slides were scanned and analyzed as previously described (38).
The Statistical Analysis of Microarrays (SAM) program was used to identify genes whose expression changed 2.5-fold or greater with 95% or greater confidence (http://statweb.stanford.edu/~tibs/SAM/ ). The average ratios were determined for three independent experiments (Table 1). The genes were arranged into known or putative operons based on the predicted co-orientation of transcription and the absence of predicted Rho-independent terminators as annotated by the Subtilist web server (http://genolist.pasteur.fr/SubtiList/ ). Genes that are part of a known or predicted operon but with changes less than 2.5-fold with a 95% or greater confidence are also included in Table 1.
Quantitative RT-PCR.Cultures were grown in S750 minimal medium as described above, 5- to 10-ml aliquots were removed at the specified times, and an equal volume of cold methanol was added to each aliquot. Cells were harvested by centrifugation at 12,000 × g for 5 min at 4°C, and total RNA was prepared as described previously (54). Briefly, cells were pelleted by centrifugation at 14,000 × g for 20 min at 4°C, and the supernatant was discarded. The cell pellet was resuspended in cold methanol and stored at −20°C until further use. RNA was isolated using RNeasy minicolumns according to the manufacturer (Qiagen). RNA quality was assessed by monitoring the integrity of rRNA using gel electrophoresis. Genomic DNA was removed from the RNA sample by treatment with DNase I (Roche) for 30 min at 37°C, followed by heat inactivation 75°C for 20 min. The SuperScript III first-strand cDNA synthesis kit was used along with 0.5 to 1 μg of total RNA and random hexamers to prepare cDNA according to the manufacturer (Invitrogen). Real-time PCR was used with Brilliant III Ultra Fast qPCR SYBR green according to the manufacturer (Agilent Technologies) along with an Mx3000PQPCR thermocycler and MxProQPCR software (Stratagene). The following conditions were used for PCR: 1 cycle of 95°C for 3 min to denature DNA, 40 cycles of 95°C for 20 s followed by 60°C for 20 s for amplification, and 1 cycle of 95°C for 1 min, 55°C for 30 s, and 95°C for 30 s to generate the dissociation curve. The data were calibrated to divIC mRNA since the abundance of divIC mRNA remained constant throughout the growth cycle during the conditions used in these experiments.
β-Galactosidase assays.Liquid cultures were grown in shaker flasks at 37°C with vigorous aeration. One-milliliter aliquots were removed and placed in a 2.2-ml 96-well polypropylene block that was stored at −20°C until it was time to assay β-galactosidase activity. A second aliquot was taken to determine OD600. β-Galactosidase activity was determined as previously described by Griffith and Grossman (55). Briefly, cells were prepared by thawing to room temperature, adding 20 μl of toluene to each well, and permeabilizing cells directly in the block with vigorous pipetting up and down using a multichannel pipettor. The assay was initiated with the addition of 20 μl of freshly prepared ortho-nitrophenyl-β-d-galactopyranoside (4 mg/ml) and terminated with the addition of 40 μl of 1 M Na2CO3. Cell debris was pelleted in a microtiter plate by centrifugation at 3,000 × g for 10 min. A 100-μl aliquot of each supernatant was transferred to a new plate using a multichannel pipettor. A420 was determined using a SpectraMax plate reader (Molecular Dynamics), and data analysis was performed using Microsoft Excel. β-Galactosidase activity per milliliter was calculated as 1,000 × (ΔA420/minute/milliliter) and β-galactosidase specific activity as 1,000 ×(ΔA420/minute/milliliter)/OD600 of culture.
Characterization of conditioned medium.To begin to characterize the molecule(s) responsible for the iron acquisition activity present in the growth medium of strain JH642, conditioned medium was treated as described below and add-back experiments were performed using strain NC209 (PdhbA-lacZ).
Physical characterization.The minimal concentrations of DNase I, RNase A, and proteinase K were empirically determined to degrade plasmid DNA, rRNA, and bovine serum albumin (BSA), respectively, as determined by gel electrophoresis. Specifically, conditioned medium was treated individually with each of the following: 10 Kunitz units/ml of DNase I (Sigma) for 1 h at 37°C, 1 Kunitz unit/ml of RNase A (Sigma) for 1 h at 37°C, and 150 μAu/ml (Qiagen) for 3 h at 37°C, followed by heat inactivation at 75°C for 20 min, heat (75°C for 20 min), long-term cold storage (4°C for at least 3 months), and filtration with a Centricon-3 (5,000 × g at 4°C).
Assay of ferrous iron.Iron standards were prepared using 100 μl of known concentrations (0 to 30 nmol) of Fe(NH4)2(SO4)2 (Sigma) in a sealed and flushed serum bottle containing 1.9 ml of degassed oxalate solution (28 g/liter of ammonium oxalate, 15 g/liter of oxalic acid) (Sigma). Conditioned medium samples (100 to 300 μl) were also prepared with degassed oxalate solution. An aliquot (20 μl) of the sample in oxalate solution was added to 1 ml of ferrozine solution [1 g/liter of ferrozine in 50 mM HEPES (pH 7.0) and 20 mM Fe(NH4)2(SO4)2 in 0.5 N HCl]. Samples were vortexed, and A562 was determined using a spectrophotometer (Milton Roy; Spectronic 601). Alternatively, undiluted conditioned medium or conditioned medium that had been concentrated 100-fold by evaporation using a speed vacuum was added directly to the ferrozine solution and assayed as described above.
ACKNOWLEDGMENTS
We thank Rajat Pal for performing some of the preliminary quantitative RT-PCR experiments, Patrick Hill and Melanie Barker Berkmen for stimulating discussion and for critical reading of the manuscript, Alan Grossman, John Helmann, and Daniel Kearns for kindly sharing strains, Alan Grossman for use of the microarray platform, and Jennifer Lin for assistance with the ferrozine assay.
This work was supported by a grant from the Healey Endowment Foundation awarded to K.L.G. and funding from the Microbiology Department and the College of Natural Sciences at the University of Massachusetts, Amherst.
FOOTNOTES
- Received 20 June 2016.
- Accepted 24 September 2016.
- Accepted manuscript posted online 17 October 2016.
- Copyright © 2016 American Society for Microbiology.