Pyrophosphate-Mediated Iron Acquisition from Transferrin in Neisseria meningitidis Does Not Require TonB Activity

The ability to acquire iron from various sources has been demonstrated to be a major determinant in the pathogenesis of Neisseria meningitidis. Outside the cells, iron is bound to transferrin in serum, or to lactoferrin in mucosal secretions. Meningococci can extract iron from iron-loaded human transferrin by the TbpA/TbpB outer membrane complex. Moreover, N. meningitidis expresses the LbpA/LbpB outer membrane complex, which can extract iron from iron-loaded human lactoferrin. Iron transport through the outer membrane requires energy provided by the ExbB-ExbD-TonB complex. After transportation through the outer membrane, iron is bound by periplasmic protein FbpA and is addressed to the FbpBC inner membrane transporter. Iron-complexing compounds like citrate and pyrophosphate have been shown to support meningococcal growth ex vivo. The use of iron pyrophosphate as an iron source by N. meningitidis was previously described, but has not been investigated. Pyrophosphate was shown to participate in iron transfer from transferrin to ferritin. In this report, we investigated the use of ferric pyrophosphate as an iron source by N. meningitidis both ex vivo and in a mouse model. We showed that pyrophosphate was able to sustain N. meningitidis growth when desferal was used as an iron chelator. Addition of a pyrophosphate analogue to bacterial suspension at millimolar concentrations supported N. meningitidis survival in the mouse model. Finally, we show that pyrophosphate enabled TonB-independent ex vivo use of iron-loaded human or bovine transferrin as an iron source by N. meningitidis. Our data suggest that, in addition to acquiring iron through sophisticated systems, N. meningitidis is able to use simple strategies to acquire iron from a wide range of sources so as to sustain bacterial survival.


Introduction
Neisseria meningitidis (Nm) is found exclusively in humans, and although it is frequently present in the nasopharynx of asymptomatic carriers, it may be the causative agent of life-threatening invasive infections such as septicemia and meningitis [1]. Ability to acquire iron from various sources has been demonstrated to be a major determinant in the pathogenesis of Nm [2]. In mammals, iron sequestration is the main form of nutritional immunity [3], [4]. Obtaining iron required for bacterial growth is a challenge, since 99.9% of total body iron is sequestered inside the cells [5]. Outside the cells, iron is bound to transferrin in the serum or to lactoferrin in mucosal secretions [2]. Another iron source in mammals is heme, mainly contained in hemoproteins like hemoglobin. When freed after erythrocyte lysis, most hemoglobin is bound by haptoglobin. Hemoglobin degradation allows the release of heme that is sequestered by hemopexin to prevent its toxicity [5]. Bacterial acquisition of iron in mammals requires the activity of transport systems allowing uptake of iron and/or heme bound to proteins. In Nm, the HmbR [6] and HpuAB outer membrane transport systems [7] allow the bacteria to use hemeloaded proteins as a heme source. HmbR and HpuAB systems differ according to their substrate specificity. HmbR can obtain heme from hemoglobin with better efficiency for human hemoglobin [6]. In contrast, HpuAB not does not exhibit specificity toward the human forms of its two substrates, characterized as hemoglobin and haptoglobin-hemoglobin complexes [8]. Nm strains express HmbR, HpuAB or both systems [9]. Most invasive strains express HmbR alone or both heme uptake systems, as reported in isolates of the hyperinvasive genotype ST-11 [9]. Strains expressing only the HpuAB heme transport system were mostly described as carriage strains [9]. The periplasmic heme binding protein and the inner membrane heme transporter are not yet identified. Inside the cytoplasm, heme is degraded by HemO, a bacterial heme oxygenase, thus allowing the release of iron [10].
The main source of iron in blood is iron-loaded transferrin. Iron is extracted from iron-loaded human transferrin by the TbpA/ TbpB outer membrane complex [11]. Also, Nm expresses the LbpA/LbpB outer membrane complex, which can extract iron from iron-loaded human lactoferrin [12]. After transportation through the outer membrane, iron is bound by the periplasmic protein FbpA and directed to the FbpBC inner membrane transporter [13]. Most of heme and iron outer membrane transport systems require energy provided by the ExbB-ExbD-TonB system [14]. TonB independent iron transport processes were also reported. [15], [16]. Alongside the two systems allowing the obtaining of iron contained in human protein, Neisseriae genomes encode systems enabling uptake of free iron. The transport of iron-loaded xenosiderophores has been investigated in Neisseria gonorrhoeae [17]. Iron-loaded xenosiderophores are transported by the TonB-dependent outer membrane transporter FetA [18], sent by FbpA to the inner membrane FbpBC transporter and degraded inside the cytoplasm to allow iron release [15]. TonB-independent transport of xenosiderophores through the outer membrane has been described in N. gonorrhoeae, but the mechanism remains hypothetical [17]. In contrast, the role of the FbpABC inner membrane ABC transporter in TonB-independent use of enterobactin, salmochelin and other xenosiderophores has been clearly demonstrated [15]. The absence of siderophore biosynthesis was reported for Nm [19]. Only the use of a ferrated form of three dihydroxamate siderophores (schizokinen, arthrobactin, aerobactin) can stimulate growth of Nm [20]. Recently, the binding of Ferric enterobactin by the factor H binding protein was described [21].
Iron-complexing compounds like citrate and pyrophosphate have been shown to support Nm growth ex vivo [19]. The use of iron pyrophosphate as an iron source by N. meningitis was described, but not investigated. Pyrophosphate-dependent use of iron was investigated in Escherichia coli [22], [23]. In that bacterium, pyrophosphate facilitates the enterobactin-dependent iron uptake process [24]. In the absence of enterobactin, pyrophosphate acts as an iron chelator and strongly inhibits E. coli growth [25]. Also, pyrophosphate was shown to participate in iron transfer from transferrin to ferritin [26]. This report aimed to investigate the mechanism that allows use of ferric pyrophosphate as an iron source and its impact on meningococcal virulence.

Ethics statement
This study was carried out in strict accordance with the European Union Directive 2010/63/EU (and its revision 86/609/ EEC) on the protection of animals used for scientific purposes. Our laboratory has the administrative authorization for animal experimentation (Permit Number 75-1554) and the protocol was approved by the Institut Pasteur Review Board that is part of in the Regional Committee of Ethics of Animal Experiments of the Paris region (CETEA 2013-0190).

Bacterial strains and plasmids
Bacterial strains and plasmids used in this study are listed in Table 1. in water at a 10 mM final concentration, filter-sterilized and stored at room temperature. Desferal (Sigma; ref: D9533) was prepared in water at 15 mM, filter-sterilized and stored at 220uC. All solutions were filter-sterilized using 0.20 mm Millipore filters. Nm strains were grown on GCB agar plates supplemented with Kellogg supplement solution [27]. To create iron depletion, supplement S2 was substituted for desferal (30 mM final concentration). When required, kanamycin (Kan) and erythromycin (Ery) were added at 50 mg/ml, and 2 mg/ml respectively. Nm strains were grown at 37uC under a 5% CO 2 atmosphere. E. coli strains were grown on LB medium [28] at 37uC. Solid media agar contained 1.5% agar.

Use of iron source assays
To evaluate the effect of mutation of the Nm capacity to use various iron sources, strains were first isolated on GCB plates supplemented with S1 and S2 complements and grown for 18 h at 37uC in the presence of 5% CO 2 . Bacteria were isolated on the test plates and incubated for 18 h at 37uC in the presence of 5% CO 2 . Iron-depleted GCB plates (see above) were supplemented with the tested iron sources.

Invasion assays in mice
Nm tested strains were grown on GCB plates for 18 h at 37uC under a 5% CO 2 atmosphere. Bacteria collected from one plate were suspended in physiological serum and the density of the cell suspension was adjusted to 2.5610 6 bacteria/ml. Four-hundred ml of the bacterial suspension were supplemented with 100-ml of the tested iron source, and the mixture was inoculated intraperitoneally into 7-week-old BalbC mice (Janvier). The number of viable bacteria before inoculation was then determined by plating serial dilutions on GCB plates. At t = 6 h, blood and intraperitoneal samples were collected, diluted in physiological serum and serial dilutions were plated on GCB plates supplemented with S1 and S2 and kanamycin (50 mg/ml). After 18 h incubation at 37uC under a 5% CO 2 atmosphere, colonies were counted.

Imaging of bioluminescence from animals
Mice were then anesthetized with a constant flow of 2.5% isoflurane mixed with oxygen, using an XGI-8 anesthesia induction chamber (Xenogen Corp.). The mice were maintained for at least 5 min. Bacterial infection images were acquired using an IVIS spectrum system (Xenogen Corp., Alameda, CA) according to instructions from the manufacturer. Analysis and acquisition were performed using Living Image 3.1 software (Xenogen Corp.). Images were acquired using a 1 min integration time with a binning of 16. All other parameters were held constant. Quantifying was performed using the photons per second emitted by each mouse.

Genetic techniques
Nm was transformed using linear 3-partner PCR fragments obtained as described below. Nm strains were grown on GCB plates for 18 h at 37uC under a 5% CO 2 atmosphere. Bacteria collected from one plate were suspended in GCB medium completed with S1 and S2 supplements and MgCl2 at a 5 mM final concentration (GCBMg medium). Bacterial density was adjusted at OD 600 :1. Three hundred microliters of the bacterial suspension were placed inside a well of a 24-well multiwell plate (Falcon), supplemented with a PCR fragment (100 to 500 ng) and incubated for 30 min at 37uC under a 5% CO 2 atmosphere. The mixture was supplemented with 700 ml of GCBMg medium and incubated for 5 h at 37uC under a 5% CO 2 atmosphere. One-hundred and 500 ml samples of the mixture were plated on GCB complete medium supplemented with selective antibiotic and incubated for 18 h at 37uC under a 5% CO 2 atmosphere. Six clones were isolated on selective medium, screened using PCR and positive clones were stored at 280uC in complete GCB medium supplemented with glycerol (20% final concentration). For fbpABC mutants, GCB supplemented with S1 Kellogg supplement solution [27], bovine hemoglobin (10 26 M) and erythromycin was used as selective medium.

DNA manipulations
DNA fragments were amplified from chromosomal Nm strain 2C4.3 in a Hybaid PCR thermocycler using Phusion DNA polymerase (Finnzymes). Restriction, modification and ligation were carried out according to the manufacturer's recommendations. Purification of DNA fragments from the PCR reaction, the restriction reaction and agarose gels was performed using the Macherey-Nagel NucleoSpin Extract II kit.

Construction of the 2C4.3::lux strain
Plasmid pXen-13 (Xenogen Corp., Alameda, CA) containing the Photorhabdus luminescens luxCDABE operon was modified by insertion of an Nm-specific promoter sequence. To express the luxCDABE operon under the PproB meningococcal promoter Nm, a 600 bp promoter sequence of the porB gene (PporB) from strain 2C4.3 was amplified using primers PorB3 and PorB4 (Table 2) and cloned into a BamHI site upstream of the luxCDABE operon after Kleenow filling. The generated plasmid was named pDG33. The fragment encompassing the luxCDABE cassette and the porB promoter was extracted by digesting pDG33 with KpnI and SacI restriction enzymes and inserted into the BamHI site of plasmid pTE-KM [29], upstream from the kanamycin resistance cassette aph3'. In the resulting vector, named pDG34, the PporB-luxCDABE-aph3' was flanked by the meningococcal pilE gene and, 120 bp downstream, by the pilE gene to facilitate chromosomal integration upon transformation.

Construction of knockout mutants in Nm
Non-polar mutations that delete entire genes were created by allelic exchange with the non-polar Ery gene cassette. For knockout genes in Nm, the methods already described require the use of E. coli to clone, in plasmids, Nm DNA fragments containing a gene of interest, disrupted by insertion of a cartridge expressing antibiotic resistance [30]. These methods require cloning steps and are subordinated to the stability of the recombinant plasmids and their absence of toxicity when introduced into E. coli. To avoid the use of cloning steps, we directly introduced into Nm the disrupted genes contained in the DNA fragment obtained using a two-step PCR procedure. The two-step PCR procedure was used to produce a PCR product in which the Ery gene cassette is flanked by arms of about 500 to 1,000 bp, corresponding to sequences upstream from the start codon and downstream from the stop codon of the gene of interest. The erythromycin cartridge was amplified from plasmid pMGC20 [31] using Eram1 and Eram3 as primers ( Table 2). The primers used for tonB were TonBAmtAmt and TonBAmtAvlEry for the upstream region and TonBAvlAmtEry and TonBAvlAvl for the downstream region (Table 2). For porA, the primers used were PorAAmtAmt and PorAAmtAvlEry for the upstream region and PorAAvlAmtEry and PorAAvlAvl for the downstream region ( Table 2). For porB, the primers used were PorBAmtAmt and PorBAmtAvlEry for the upstream region and PorBAvlAmtEry and PorBAvlAvl for the downstream region (Table 2). To delete fbpABC, the primers used were FbpABCAmtAmt and FbpAB-CAmtAvlEry for the upstream region and FbpABCAvlAmtEry and FbpABCAvlAvl for the downstream region (Table 2). For each gene of interest, the sequence of the 59 end of the reverse primer used to amplify the upstream region was anti-parallel to the 59end of Eram1 primer and the sequence of the 59 end of the forward primer used to amplify the downstream region was antiparallel to the 59 Eram3 primer. For each gene of interest, a 1 ml sample of upstream and downstream regions was mixed with 1 ml of the erythromycin cartridge and the mixture was amplified using primers TonBAmtAmt and TonBAvlAvl for tonB, PorAAmtAmt

Iron binding assay
The ability of desferal, pyrophosphate and its structural analogues to bind iron Fe 3+ was visualized with a classical assay used to quantify siderophores in solution [32].

Statistical analysis
Data are expressed as the mean 6 SD of 5 samples, and the reproducibility was confirmed at least in three separate experiments. Statistical analysis were performed using two-way unpaired Student's t-test and considered significant if P,0.05.

Ex vivo use of ferric pyrophosphate as an iron source
In a first set of experiments, we investigated the ex vivo use of ferric pyrophosphate as an iron source by Nm strain 2C4.3. The tested strain was cultured on GCB medium supplemented with S1 complement and desferal 15 mM or 30 mM to create iron depletion. On this medium, no growth of the Nm 2C4.3 strain was observed. The addition of iron pyrophosphate led to growth restoration ( Table 3). The minimal concentration of iron pyrophosphate required for growth on GCB iron-depleted medium was 15 mM (Table 3). In iron pyrophosphate, the iron content was about one-tenth of the iron pyrophosphate compound in weight. In spite of the presence of desferal used as a chelator, Nm was able to use iron pyrophosphate as an iron source. This suggested that the affinity of pyrophosphate for iron was higher than that of desferal. This hypothesis was strengthened by comparing the ability of pyrophosphate and desferal to induce a color change in an iron dye complex used to detect and quantify siderophores [32]. This ability was related to the capacity to bind iron and release free dye [32]. As seen in Figure 1, pyrophosphate induced a strong color change at 630 nm, reflecting its ability to bind iron [24]. In contrast, with desferal, the free dye release occurred much more slowly (Figure 1).

The iron pyrophosphate transport pathway in Nm
In order to be used by the bacteria, the iron source must be transported through the outer membrane, the periplasm and the inner membrane. Outer membrane transport of iron and heme primarily involves transporters requiring the presence of the ExbB-ExbD-TonB complex as an energy provider [33]. We first checked for the effect of tonB disruption upon the ability of Nm to use iron pyrophosphate as an iron source. As shown in Figure 2, tonB disruption did not impair the use of iron pyrophosphate as an iron source. In contrast, the use of iron-loaded human transferrin and hemoglobin as an iron source was abolished in the Nm tonB mutant. Similarly, disruption of porA or porB structural genes encoding for the Nm major porins [34] had no effect on the use of iron pyrophosphate as an iron source (Figure 2). The inner membrane FbpABC transporter was shown to be required for the use of transferrin and xenosiderophores as iron sources [35], [15]. We thus tested the effect of fbpABC disruption of the capacity of Nm to use iron pyrophosphate as an iron source. As seen in  Exogenous pyrophosphate allows iron utilization in the presence of desferal GCB medium can support Nm growth in the presence of supplement S1 and in the absence of supplement S2. Thus, iron traces present in the medium are sufficient for sustaining Nm growth. Addition of desferal at 15 mM or 30 mM abolished Nm growth on this medium. Addition of pyrophosphate, at 5 mM or higher, restored the growth of Nm on GCB S1 medium supplemented with desferal (Table 4). This result is in good agreement with the high affinity of pyrophosphate for iron [24]. The use of two structural analogues of pyrophosphate strengthened this conclusion. Imidodiphosphate and methylenediphosphonic acid were added to iron depleted GCB medium and growth of Nm was investigated. Our results demonstrated that methylenediphosphonic acid, similarly to pyrophosphate, allowed Nm growth on iron-depleted medium when added at a 5 mM final concentration (Table 4). In contrast, addition of imidodiphosphate did not support Nm growth on the same medium (Table 4). These results are in accordance with our results demonstrating that pyrophosphate and methylenediphosphonic acid, in contrast to imidodiphosphate, bind iron with higher affinity than desferal ( Figure 1). Pyrophosphate and methylenediphosphonic-acid-dependent use of iron did not require TonB activity, but was abolished when fbpABC genes were disrupted (data not shown).

Pyrophosphate enables TonB-independent use of transferrin as an iron source
In Nm, the use of transferrin as an iron source is restricted to human transferrin [2]. Other transferrins are not used by Nm as an iron source. Transportation of iron from human transferrin requires the activity of the TonB-dependent outer membrane transporter TbpAB [2]. In vitro experiments demonstrated the role of pyrophosphate as a mediator of iron transfer from transferrin to ferritin [36]. According to these results, pyrophosphate can bind iron loaded on transferrin and deliver it to ferritin [26]. Other authors demonstrated transfer from iron-loaded transferrin to pyrophosphate [37], [38]. The results described in these reports prompted us to check for the effect of pyrophosphate on the use of iron-loaded transferrin as an iron source. In a first set of experiments, we investigated the effect of pyrophosphate on the use of human transferrin and bovine transferrin as an iron source. As shown in Figure 3, in the absence of pyrophosphate, Nm used only human transferrin as an iron source. In contrast, in the presence of pyrophosphate, both bovine and human transferrins were iron sources for Nm. This cannot be explained by solubilization of contaminating iron, since the concentration of pyrophosphate used in this assay (1 mM) was not sufficient for supporting Nm growth in the presence of 30 mM desferal (Table 4). Since the TbpAB transport system exhibits absolute specificity for human transferrin, we hypothesized that the transport pathway used in the presence of pyrophosphate was independent of TbpAB activity. As a consequence, tonB disruption would not have an effect on the use of human or bovine transferrin in the presence of pyrophosphate. This was shown to be the case (Figure 3). Human and bovine transferrin as an iron source was also used in the presence of the two pyrophosphate analogues already tested in this report. As seen in Figure 3, imidodiphosphate addition did not alter the phenotype observed with wild type and tonB mutant strains. In contrast, similarly to pyrophosphate, methylenediphosphonic acid allowed TonB-independent use of human and bovine transferrin as iron sources (Figure 3).

Effect of methylenediphosphonic addition upon Nm survival in the mouse model
In the absence of a usable iron source, Nm is cleared from mice very rapidly after intraperitoneal injection [39]. Addition of human transferrin to the bacterial suspension allows Nm to survive in mice [39]. Since the above results demonstrated that addition of pyrophosphate and methylenediphosphonic acid enabled the use of non-human transferrin as an iron source, we hypothesized that the addition of pyrophosphate and its structural analog, methylenediphosphonic acid, could promote the use of mouse transferrin as an iron source and enhance the survival of Nm in the mouse model. The addition of iron pyrophosphate (50 mM final concentration), pyrophosphate (5 mM final concentration) or imidodiphosphate (5 mM final concentration) had no effect on survival of Nm in mice (data not shown). In contrast, addition of 5 mM methylenediphosphonic acid, which is not degraded by inorganic pyrophosphatase [24], [40], increased significantly the ability of wild-type Nm to survive in the mice compared to control untreated mice (p = 0.026) (Figure 4). However, this effect of methylenediphosphonic acid on Nm growth was less prominent than that obtained by the addition of human transferrin (p = 0.0002) compared to control untreated mice ( Figure 4). As evidenced in ex vivo assays, the effects of methylenediphosphonic acid were not abolished by tonB disruption. We therefore tested, in the mouse model (in vivo), the impact of tonB disruption on bacterial survival. As shown in figure 4, dynamic imaging result showed that tonB Nm mutant still showed significant better survival in mice treated with PcP compared to untreated control (p = 0.034). At the opposite, no more difference of survival tonB Nm mutant in human transferrin-treated mice compared to untreated control (p = 0.1) (Figure 4). We further study the survival of the wild type and the tonB mutant during the experimental infection in mice by bacterial counting from the peritoneal cavity and from the blood. The bacterial counts in blood and peritoneal cavity corroborated the results of dynamic imaging obtained with the wild type strain (Figure 4). For the tonB mutant, the results of the bacterial count in blood and peritoneal cavity are in a good accordance with the results of dynamic imaging when human transferrin was added (Figure 4). When methylenediphosphonic acid was added, a non significant trend for higher bacterial counts in the blood was observed (Figure 4). These data suggest that methylenediphosphonic acid enables a wide range of iron acquisition during experimental infection. Table 3. Use of iron pyrophosphate (FePPi), FeNo3 and FeCl3 as iron sources by Nm.

Discussion
Iron acquisition by pathogenic Neisseria within the host is a major virulence trait. Bacteria employ specific receptors to obtain this transition metal from iron-containing proteins (transferrin, lactoferrin) in a TonB-dependent manner. However, tonBindependent pathways have been described. The mechanisms and significance of these pathways are not yet understood. We describe here a TonB independent iron transport process in Nm. This TonB-independent process allows ex vivo transportation of the iron pyrophosphate complex through the outer membrane. In vitro, pyrophosphate and methylenediphosphonic acid (a structural analogue of pyrophosphate) bind iron with higher affinity than desferal, and rescue Nm growth on plates in the presence of desferal in a TonB-independent manner. Ironcomplexing compounds like citrate and pyrophosphate have been shown to support Nm growth ex vivo on culture plates [19], but their transport pathways have not been investigated. Using a rapid method, we built various mutants that enabled demonstrating the TonB-independent mechanism responsible for transport of iron pyrophosphate. porA or porB inactivation did not abolish the ability to use iron pyrophosphate as an iron source. Iron-loaded pyrophosphate could pass the outer membrane through both PorA and PorB porins. This hypothesis is in good agreement with identification of phosphate and ATP as PorB ligands [41]. Also, iron pyrophosphate can pass the outer membrane through a porin hypothesized to be responsible for the TonB independent use of xenosiderophores [17]. Thus, the transport pathway of iron pyrophosphate through the Nm outer membrane remains to be elucidated. FbpABC was shown to be responsible for the transport of iron pyrophosphate through the inner membrane. This complex was already shown to be required for iron transport through the inner membrane in Nm [35] and N. gonorrhoeae [13].
Pyrophosphate was shown to have a siderophore-like activity when ferritin was used as an iron source [19]. Moreover, pyrophosphate was shown to transfer iron from transferrin to ferritin [26]. Accordingly, our data obtained on plates demonstrate that pyrophosphate permits TonB-independent use of iron that is loaded from both human and bovine transferrin. In contrast, the acquisition of iron from transferrin through the TbpAB transporter is highly specific to human transferrin [11]. Indeed, it was previously shown that transgenic mice expressing human transferrin, or injection of iron-loaded human transferrin mice, leads to meningococcal growth in these animal models [39], [42]. We therefore explored, in the mouse model (in vivo), the significance of our finding concerning the role of iron pyrophosphate, pyrophosphate and its analogues on plates (ex vivo). Addition of pyrophosphate did not increase survival capacity in the mice in the absence of added human transferrin. Pyrophosphate degradation by inorganic pyrophosphatase [43][44][45] can explain this result. Addition of methylenediphosphonic acid increases survival of Nm in mice (Figure 4). This effect, also observed on a tonB mutant, suggests that TonB-independent transport of iron bound to methylenediphosphonic acid can support the growth of Nm in mice. Bacterial CFU counting revealed that tonB disruption decreased the ability of N. meningitidis to survive in the mice model in the presence of both PcP and human transferrin ( Figure 4). The effect may be due to a decreased use of murine hemoglobin as an iron source in a TonB-dependent manner [6]. However, the decrease was prominent in the presence of human transferrin. Since iron-loaded transferrin is the main iron source in mice, we propose that methylenediphosphonic acid is able to obtain iron from mouse transferrin, as from bovine and human transferrin, and to form a ferric complex that can be transported through the outer membrane. According to the results obtained with dynamic imaging method, the use of mouse transferrin in the presence of methylenediphosphonic acid not requires the TonB activity. The effect of methylenediphosphonic acid addition on Nm tonB mutant survival inside the mice cannot be related to tonB reversion, since bacteria recovered from the intraperitoneal cavity and from blood were unable to use human transferrin and hemoglobin on plates. Taken together, the data in this report demonstrate ex vivo and in vivo pyrophosphate-mediated use of iron-loaded transferrin as iron sources. Ex vivo data clearly demonstrate that the pyrophosphate-mediated use of iron-loaded transferrin as an iron source not requires the TonB activity. Table 4. Effect of pyrophosphate (PPi), methylenediphosphonic acid (PcP) and imidodiphosphate (PnP) on Nm growth on iron-depleted medium.  . TonB-independent use of transferrin as an iron source. The tested strains were isolated on a GCB plate supplemented with S1 and S2 complements and grown for 18 h at 37uC in the presence of 5% CO 2 . GCB plates depleted for iron by addition of desferal were supplemented with human or bovine transferrin added at a 5 mM final concentration. When specified, PPi, PcP and PnP were added at a 1 mM final concentration. Bacteria were isolated on test plates and incubated for 18 h at 37uC in the presence of 5% CO 2 . The experiment was repeated three times. Representative results are presented. doi:10.1371/journal.pone.0107612.g003 Similarly to pyrophosphate-dependent iron uptake, other TonBindependent iron uptake processes have been described ex vivo in Neisseria [46], [15].
In Escherichia coli, pyrophosphate acts as an iron chelator in an entF strain that is unable to synthesize enterobactin, but is still able to produce dihydroxybenzoic acid [25]. This demonstrates that iron pyrophosphate cannot be used as an iron source in the absence of enterobactin in E. coli. In Nm, which was not demonstrated to produce siderophore, pyrophosphate addition counteracts the iron chelating of desferal, and iron pyrophosphate can be used as an iron source. The pore sizes of PorA (1.4 nm) [47] and PorB (1.6 nm) [48], [49] porins from Nm are close to The tested strains were isolated on a GCB plate supplemented with S1 and S2 complements and grown for 18 h at 37uC in the presence of 5% CO 2 . Bacteria were suspended in sterile physiological serum to obtain a cell density of 2.5610 6 bacteria/ml. When specified, 100 ml of the tested iron source were added to 400 ml of the bacterial suspension to obtain 0.05 mM for human transferrin and 5 mM for PcP. For the control experiment, 100 ml of physiological serum were added. For each experiment, the mixtures were injected intraperitoneally into five mice and bioluminescence was measured 30 min, and 360 min after injection, as described in Materials and methods. At t = 360 min, blood and peritoneal washes samples were taken, diluted in physiological serum and plated on GCB solid medium. After 18 h incubation at 37uC in the presence of 5% CO 2 , the colonies were counted. Data represent the means 6 SD from 3 independent experiments of groups of five mice per time point in each experiment. Student's t-test results were included in the figure and in the table. CFU: colony-forming unit. doi:10.1371/journal.pone.0107612.g004 those of OmpC (1,3 nm) and OmpF (1,4 nm) porins from E. coli [50], suggesting similar transport of iron pyrophosphate across the outer membrane. In Neisseria, the transport system responsible for transportation of iron pyrophosphate through the inner membrane was identified as FbpABC. This inner membrane transport system, was already demonstrated to be required for transport of iron from transferrin [35] and exogenous siderophores [15]. In FbpA, phosphate was identified as a synergistic anion allowing tight sequestration of iron [51]. Similarly to phosphate, pyrophosphate or methylenediphosphonic acid could play the role of a synergistic anion. This was suggested for pyrophosphate, and phosphatase activity was hypothesized for FbpA [51]. According to results obtained with methylenediphosphonic acid, this phosphatase activity is not required for the synergistic activity of pyrophosphate. In E. coli, that synthesizes siderophore, two periplasmic binding proteins and inner membrane transporters facilitate the transport of ferric-siderophore complexes. FhuD, the periplasmic protein responsible for directing ferric hydroxamate to the inner membrane FhuBC 2 ABC transporter, also facilitates the transport of ferrichrome, coprogen, ferrioxamine B and aerobactin [52]. FepB binds ferric enterobactin and enterobactin in the periplasm and directs it to the inner membrane ABC transporter FepC 2 D 2 [53]. Moreover, E. coli synthesizes another ABC transporter, responsible for the transport of iron citrate through the inner membrane [54]. Similarly to FbpABCD from Nm, the FecBCDE inner membrane transport system transports iron Fe 3+ but not the iron citrate complex [54]. In E. coli, FecBCDE can be hypothesized to also transport iron pyrophosphate. In E. coli, expression of the fec operon containing the structural genes of this inner membrane transport system is repressed by iron-loaded Fur and induced in the presence of iron-loaded citrate [55]. In the absence of iron-loaded citrate, basal expression of the fecABCDE operon would not be sufficient to promote a speculated FecBCDEdependent transport of iron pyrophosphate through the inner membrane. Within the cytoplasm, intracellular pyrophosphatase can degrade pyrophosphate and facilitate iron release. In addition, reduction of iron by a ferric reductase [56], [57] could provoke its release from pyrophosphate and methylene diphosphonic acid. Iron reduction by a reductase was reported to be responsible for iron release from siderophores like coprogen and ferrioxamine [58].
Our work opens up new insights into iron acquisition in Nm. Indeed, Nm seems to preferentially use iron among the transition metals. Several systems have been selected to allow highly efficient iron acquisition in the natural habitat of Nm. Pyrophosphate could permit iron acquisition from a wide range of iron sources like lactoferrin at sites such as the nasopharynx, the natural habitat of Nm, and might support meningococcal growth when in competition with other microbial species that produce siderophores, exhibiting lower affinity for iron than enterobactin. Thus, the presence of pyrophosphate enables Nm to obtain iron using a simple, highly competitive pathway.