Iron chelators modulate the fusogenic properties of Salmonella-containing phagosomes

Abstract

In macrophages, the divalent cations transporter Nramp1 is recruited from the lysosomal compartment to the membrane of phagosomes formed in these cells. Nramp1 mutations cause susceptibility to infection with intracellular pathogens such as Salmonella and Mycobacterium. Intracellular survival of Salmonella involves segregation in an endomembrane compartment (Salmonella-containing vacuole, SCV) that remains negative for the mannose-6-phosphate receptor (M6PR) and that is inaccessible to the endocytic pathway. Expression of Nramp1 at the membrane of SCVs stimulates both acquisition of M6PR and accessibility to newly formed endosomes. The possible role of Nramp1-mediated iron transport on SCV maturation was investigated with membrane-permeant iron chelators. Pretreatment of primary macrophages from Nramp1 mutant mice or of RAW264.7 macrophages (from BALB/c mice bearing an Nramp1^D169-deficient allele) with either desferrioxamine or salicylaldehyde isocotinoyl hydrazone restored recruitment of M6PR and delivery of the fluid phase marker rhodamine dextran to SCVs to levels similar to those seen in macrophages expressing WT Nramp1. The effect was specific and dose-dependent and could be abrogated by preincubation with excess iron. These data suggest that Nramp1-mediated deprivation of iron and possibly of other divalent metals in macrophages antagonizes the ability of Salmonella to alter phagosome maturation.

In mice, mutations at Nramp1 (Natural resistance-associated macrophage protein 1/Slc11a1) cause susceptibility to infections with several intracellular pathogens including Salmonella, Leishmania, and Mycobacterium (1, 2). In humans, polymorphic variants at or near NRAMP1 are associated with susceptibility to tuberculosis and leprosy, both in endemic areas of disease and in the outbreak situation (1, 3, 4). Nramp1 codes for an integral membrane protein, composed of 12 predicted transmembrane domains (5), that is expressed in the Lamp1-positive late endosomal/lysosomal compartment of macrophages (6, 7). Nramp1 is also expressed in the membrane of gelatinase-positive tertiary granules of neutrophils where it copurifies with membrane-bound subunits of the vacuolar H+-ATPase (8). After phagocytosis of inert particles (6) or live bacteria (8–10) by both cell types, Nramp1 is rapidly recruited to the membrane of the phagosome.

Nramp1 defines a large family of membrane proteins that has been highly conserved in evolution from bacteria to humans (11, 12). Studies in Xenopus laevis oocytes (13) and stably transfected Chinese hamster ovary cells (14, 15) have shown that the close mammalian homolog of Nramp1, Nramp2/DMT1/Slc11a2 (16), functions as a pH-dependent divalent metal transporter of broad substrate specificity. Nramp2/DMT1 is expressed at the brush border of the absorptive epithelium of the duodenum, where it is responsible for uptake of dietary iron (17). Nramp2 also colocalizes with transferrin in recycling endosomes of many cell types including reticulocytes, where it transports iron from the acidified lumen of the endosomes into the cytoplasm (18–20). Naturally occurring mutations at Nramp2 (G184R) in mice (mk) and rats (b) cause a severe form of microcytic anemia (20, 21). Finally, mammalian Nramp can functionally complement in vivo loss-of-function mutations in their fly (22) and yeast (23) counterparts.

These studies have suggested a scenario in which Nramp1 transport of divalent cations at the phagosomal membrane may influence microbial survival and/or replication at that site. However, the mechanism, substrate specificity, and direction of transport at the phagosomal membrane with respect to protein topology and polarity of pH gradient are still the subject of controversy (1, 11). Transport studies with radioisotopic ⁵⁵Fe²⁺, using either intact cells or isolated phagosomes containing either latex beads or live Mycobacterium avium, have suggested that Nramp1 may transport iron into phagosomes (24, 25). Independently, it has been reported that injection of Nramp1 mRNA into X. laevis oocytes induces small Zn²⁺-dependent inward currents suggestive of metal uptake. These authors conclude that Nramp1 may transport metals from the cytoplasm into the phagosome by a proton antiporter mechanism (26). It has been proposed that increased phagosomal iron may stimulate oxygen radical formation in situ via the Haber–Weiss reaction (24–26). On the other hand, we have used a metal-sensitive and pH-resistant fluorescent dye (Fura-FF6) chemically coupled to zymosan particles to monitor by microfluorescence imaging the effect of Nramp1 on divalent cation fluxes across the membrane of phagosomes formed in live primary macrophages (27). This study suggests that Nramp1 may function as a pH-dependent and bafilomycin-sensitive divalent cation efflux pump at the phagosomal membrane by a mechanism similar to that of Nramp2-mediated iron transport across the membrane of acidified endosomes, and which is impaired in reticulocytes from mk mice (19).

The mechanism by which Nramp1 metal transport at the phagosomal membrane modulates either pathogen virulence or affects antimicrobial defenses of phagocytes is not understood. In the case of Mycobacterium, electron microscopy (28) and microfluorescence imaging studies (29) have shown that Nramp1 recruitment to phagosomes formed in bone marrow macrophages is associated with increased bacterial cell damage and absence of bacterial replication. In such phagosomes, Nramp1 seems to antagonize the ability of M. avium (28) and Mycobacterium bovis (29) to block recruitment of vacuolar H⁺-ATPase, reduce luminal acidification, and inhibit fusion to secondary lysosomes. In contrast, studies of Salmonella-containing vacuoles (SCVs) formed either in HeLa cells or cultured macrophages indicate that recruitment of Nramp1 to SCVs affects neither acidification nor fusion to dextran rhodamine-labeled, Lamp-1-positive lysosomes (10). However, Nramp1-positive SCVs show increased recruitment of M6PR, and increased delivery of fluid phase markers (rhodamine dextran) from early endosomes (10), both of which are known to be actively suppressed by Salmonella in permissive, Nramp1-negative cells (30). These results suggest that Nramp1 may antagonize expression of pathogen-encoded virulence determinants responsible for intracellular survival, possibly by restricting availability of divalent metals such as iron in the phagosomal lumen. This hypothesis was tested in the present study.

Materials and Methods

Materials.

A rabbit polyclonal antiserum against the cation-independent M6PR was a generous gift from S. Honing and K. von Figura (University of Göttingen, Göttingen, Germany), and a rat anti-mouse Lamp1 mAb was obtained from M. Desjardins (University of Montreal, Montreal). Cy3-labeled secondary anti-rat and anti-rabbit Abs were purchased from Jackson ImmunoResearch. Alexa Fluor 350 and tetramethylrhodamine dextran (10,000 kDa) were from Molecular Probes, and N,N,N′,N′-tetrakis 2-pyrimethylethylenediamine (TPEN) was from Calbiochem. Desferrioxamine (DFO) and salicylaldehyde isocotinoyl hydrazone (SIH) have been described (31) and were obtained from P. Ponka (Lady Davis Institute, Montreal). Hydroxyethyl starch (HES)–DFO consists of HES to which DFO is covalently conjugated (32). The resulting compound (gift of Biomedical Frontiers, Minneapolis) is a high-molecular-mass iron chelator of 80,000 kDa that does not cross cellular membranes.

Cell Culture.

RAW264.7 is a macrophage cell line (TIB 71, American Type Culture Collection) isolated by transformation from BALB/c mice, which are homozygous for the loss-of-function G169D allele of Nramp1 (Nramp1^D169) and are permissive to invasion and replication of Salmonella typhimuriumin vitro (16). Transfection and overexpression of a c-Myc-tagged variant of the WT G169 allele of Nramp1 (Nramp1^G169) in these cells (RAW/Nramp1-cMyc) corrects susceptibility of the parental cell line to Salmonella (9). RAW264.7 cells and RAW/Nramp1-cMyc transfectants were maintained as described (9).

Infection of Macrophages.

129sv mice carry a WT allele at Nramp1 (Nramp1^G169). Mutant 129sv mice bearing a null mutation at the Nramp1 locus (Nramp1^−/−) were produced by homologous recombination (33). Resident peritoneal macrophages were obtained from 6- to 8-week-old 129sv (WT) and Nramp1 mutant [knockout (KO), −/−] mice by peritoneal lavage and were used 24 h after plating for both Salmonella invasion experiments and immunofluorescence assays, exactly as described (10). In parallel, RAW macrophages and RAW/Nramp1-cMyc transfectants seeded at ≈60% confluencey were used for invasion. Briefly, an S. typhimurium strain (14028s) expressing pFPV1, a modified pBR322 plasmid (Tet^r) encoding the GFP (10), was used for fluorescence detection assays. This strain (GFP-S, GFP-expressing Salmonella) was used for invasion of RAW cells and peritoneal macrophages with a multiplicity of infection of 1:100, as described (10). Invasion of cells on coverslips was allowed to proceed for 20 min at 37°C, followed by extensive washing to eliminate extracellular bacteria.

Immunofluorescence.

RAW cells and primary macrophages infected with GFP-S were incubated for fixed periods of time in RPMI medium 1640 at 37°C to allow for maturation of SCVs. After 90 min, maturation was terminated by addition of ice-cold blocking buffer (1% albumin/2% donkey serum in PBS). To detect extracellular bacteria, cells were incubated with anti-Salmonella lipopolysaccharide Abs (1:200 dilution) for 15 min and exposed to secondary Abs labeled with Alexa Fluor 350 (1:200) for 15 min. After washing, cells were fixed in 4% paraformaldehyde for 1 h, permeabilized, and blocked overnight in 1% donkey serum/2% albumin plus 0.1% Triton X-100 in PBS. To label late endosomes/lysosomes, cells were incubated with anti-Lamp1 (1:600) or anti-M6PR (1:800) Abs for 1 h followed by Cy3-conjugated secondary Abs for 60 min. After mounting, cells were visualized by conventional epifluorescence microscopy (×100) or confocal microscopy (×63) to quantify the percentage of SCVs positive for specific markers, as described (10). At least 100 internalized bacteria were counted for each condition. Access of preformed SCVs to endocytic material was studied by using a fluid-phase marker, tetramethyl-rhodamine dextran 10,000, as described (10).

Chelation of Divalent Metals.

Cells were incubated in the presence of the indicated amount of DFO for 24 h before invasion. For experiments involving SIH, the chelator was added to the cells 2 min before invasion. Membrane-impermeant chelators, EDTA and HES–DFO (6DFO/M₁), were added to the media 30 min before invasion.

Results

Results in Figs. 1 and 2 recapitulate the effect of recruitment of Nramp1 to the membrane of SCVs on their fusogenic properties. Primary peritoneal macrophages from either 129sv mice (WT for Nramp1, WT) or 129sv mice that harbor a targeted copy of Nramp1 (KO) were infected with a Salmonella strain that expresses high levels of a GFP-S. The effect of Nramp1 on recruitment of M6PR to GFP-S-containing vesicles was analyzed by immunofluorescence. In these experiments, an Ab directed against Salmonella lipopolysaccharide was used to distinguish between intracellular and extracellular bacteria. Typical images are shown in Fig. 1A, and the magnitude of the Nramp1 effect on M6P-R recruitment is quantitated in histograms shown in Fig. 3B (control groups). Similarly, the effect of Nramp1 on fusion of preformed GFP-S-containing phagosomes with early recycling endosomes was analyzed by labeling recycling endosomes with the fluid phase marker rhodamine dextran. Typical images are shown in Fig. 2A, and quantitation of the Nramp1 effect is shown in histograms of Fig. 4B (control groups). Parallel experiments were conducted in Nramp1-defective RAW264.7 macrophages (RAW; derived from BALB/c, Nramp1^D169) and RAW264.7 cells overexpressing a functional Nramp1-cMyc protein (RAWNramp1; Figs. 3C and 4C, control groups). Results with primary macrophages and with RAW cells were very similar and showed that: (i) in cells lacking functional Nramp1 (129sv-KO, RAW), SCVs segregate from the endocytic pathway, remaining largely negative for M6PR and largely inaccessible to rhodamine dextran; and (ii) expression of Nramp1 increases recruitment of M6PR and rhodamine dextran. The Nramp1 effect was quantitatively stronger in primary macrophages (compare control groups in Fig. 3 B with C for M6PR, and Fig. 4 B with C for dextran). Control experiments show that in the absence of Salmonella infection M6PR staining (Fig. 1B) and rhodamine dextran distribution (Fig. 2B) are similar in primary macrophages and RAW cultured cells (data not shown) positive or negative for Nramp1, indicating that neither overexpression nor deletion of Nramp1 affect the intracellular distribution of these compartments.

Figure 1.

Effect of Nramp1 on recruitment of M6PR to GFP-SCVs formed in peritoneal macrophages. (A) Peritoneal macrophages from Nramp1^−/− (KO, Upper) and Nramp1^+/+ mice (WT, Lower) were exposed to GFP-S for 15 min and, after removal of excess bacteria, were incubated for another 90 min. The cells were immunostained for M6PR (red). Analysis of the distribution of bacteria (green) and M6PR was performed by fluorescence microscopy. An overlay of both red and green staining is shown (Right), and yellow indicates colocalization. (B) Distribution of M6PR was analyzed in peritoneal macrophages from WT and from KO animals in the absence of Salmonella infection.

Figure 2.

Effect of Nramp1 on recruitment of rhodamine dextran to GFP-SCVs formed in peritoneal macrophages. (A) Peritoneal macrophages were infected with GFP-S as described in Fig. 1, and cells were incubated for 60 min with 1 mg/ml of rhodamine dextran. Distribution of bacteria (green) and internalized dextran (red) was analyzed by fluorescence microscopy. An overlay of both red and green staining is shown (Right), and yellow indicates colocalization. (B) Distribution of rhodamine dextran was analyzed in peritoneal macrophages from WT and KO animals in the absence of Salmonella infection.

Figure 3.

Effect of iron chelation on M6PR recruitment to SCVs formed in RAW an RAW-Nramp1 cells and in peritoneal macrophages from 129svWT and Nramp1 mutant mice. The effect of membrane-permeant and membrane-impermeant metal chelators on recruitment of M6PR to SCVs was investigated. (A) The distribution of GFP-S bacteria (green) and M6PR was analyzed by fluorescence microscopy in the absence (Upper) and presence (Lower) of DFO in peritoneal macrophages from Nramp1^−/− mice (KO). An overlay of both red and green staining is shown (Right), and yellow indicates colocalization. (B). The percentage of SCVs positive for M6PR in peritoneal macrophages from Nramp1^−/− mice (KO, open bars) and syngeneic normal controls (WT, filled bars) was quantitated in the presence/absence of each chelator (as indicated). Possible competing effects of iron added (at 40 and 80 μM) to the chelator were examined for DFO. (C) The same experiments were performed on RAW (open bars) and RAW-Nramp1 (filled bars). The effect of DFO on Lamp1 recruitment to SCVs (D) and M6PR recruitment to latex beads containing phagosomes (E) formed in RAW/RAW-Nramp1 cells was also measured. Data are means ± SE of three independent experiments.

Figure 4.

Effect of iron chelation on access of fluid phase markers to SCVs formed in RAW and RAW-Nramp1 cells and in peritoneal macrophages from 129sv WT and Nramp1 mutant mice. The effect of metal chelators on the recruitment of rhodamine dextran was examined as described in Fig. 3. (A) The distribution of GFP-S bacteria (green) and dextran was analyzed by fluorescence microscopy in the absence (Upper) and the presence (Lower) of DFO in peritoneal macrophages from Nramp1^−/− mice (KO). An overlay of both red and green staining is shown (Right). (B) The percentage of SCVs positive for dextran staining is shown for peritoneal macrophages from Nramp1^−/− (KO) and normal syngeneic (WT) controls and (C) for RAW and RAW-Nramp1 cells. Data are means ± SE of three independent experiments

To explore further the potential role of divalent cations in the observed effect of Nramp1 on fusogenic properties of SCVs, we attempted to mimic this effect with divalent cations chelators. In these experiments, Nramp1-negative 129sv (KO) primary macrophages and RAW264.7 cells and their Nramp1-expressing counterpart were exposed to chelators before infection with GFP-S, and the effect of the chelator on SCV maturation was monitored. Our initial studies of fluorescence quenching on Fura-FF6-zymosan in phagosomes demonstrated Mn²⁺ transport by Nramp1 (27). Although a selective and high-affinity chelator for Mn²⁺ would be needed for the proposed studies, such a chelator is currently unavailable. TPEN is a membrane-permeable chelator of broad specificity displaying high affinity (K_d > 10⁻¹⁰ M) for a number of divalent cations, including known Nramp proteins substrates such as Mn²⁺ and Fe²⁺ (34), but also non-Nramp substrates such as Ca²⁺ and Mg²⁺. At 100 μM, TPEN was found to reduce recruitment of M6PR to SCVs and labeling of SCVs by rhodamine dextran in both Nramp1-negative and -positive cells (data not shown). However, TPEN treatment of macrophages also reduced recruitment of the lysosomal marker Lamp-1 to SCVs (known not to be affected by Nramp1; ref. 10) and to phagosomes containing latex beads (data not shown). Thus, TPEN seemed to show a generalized inhibitory effect on intracellular fusion events, probably unrelated to the specific effect of Nramp1 on SCVs but possibly related to modulation of Ca²⁺ and/or Mg²⁺ intracellular levels by the chelator.

Nramp family members from mammals, plants, and prokaryotes can transport Fe²⁺. In addition, we have observed that an Nramp1 mutant with altered trafficking properties is expressed at the plasma membrane of Chinese hamster ovary cells, where it mediates cellular uptake of not only ⁵⁴Mn²⁺ but also ⁵⁵Fe²⁺ (J. R. Forbes and P.G., unpublished data). These and studies by Khun et al. (24, 25) suggest that Nramp1 may also transport iron at the phagosomal membrane. Therefore, we tested the effect of two iron chelators, DFO and SIH, on the maturation of SCVs. DFO diffuses slowly across membranes, whereas SIH is highly hydrophobic and readily lipid-soluble (31). Primary macrophages from 129sv (WT) and 129sv Nramp1 mutant (KO) mice were incubated with increasing concentrations of DFO or SIH for either 24 h before or at the time of invasion, respectively, and the effect of exposure to chelators on recruitment of M6PR to vacuoles containing GFP-S was monitored. Representative images are shown in Fig. 3A, and results are quantitated in the histograms shown in Fig. 3B. Identical experiments were carried out with RAW264.7 and RAW264.7 Nramp1 transfectants and results are shown in Fig. 3C. In RAW264.7 macrophages, DFO caused a marked increase in the fraction of SCVs positive for M6PR (Fig. 3C). DFO also caused an increased, albeit more modest, recruitment of M6PR to SCVs formed in RAW/Nramp1-cMyc transfectants (Fig. 3C). A DFO concentration of 150 μM was sufficient to reproduce the effect of Nramp1, with the proportion of M6PR-positive SCVs being similar in Nramp1-negative and -positive cells (Fig. 3C). Addition of FeSO₄ (40 and 80 μM) to RAW cells pretreated with DFO (100 μM) at the time of invasion reduced the effect of the chelator on recruitment of M6PR to SCV, suggesting that the effect was specific and largely mediated by iron chelation (Fig. 3C). A similar stimulatory effect on M6PR recruitment to GFP-SCVs was observed when highly membrane-permeable SIH was used in the same concentration range (Fig. 3 B and C). In addition, neither the high-molecular-weight membrane-impermeant DFO conjugate HES–DFO nor EGTA appeared to affect the M6PR recruitment to SCVs (Fig. 3C). These results suggest that chelation of intracellular iron by DFO and SIH is responsible for the observed effect on maturation of SCVs. Finally, neither DFO nor SIH had an effect on Lamp1 recruitment (Fig. 3D) or on M6PR recruitment to phagosomes containing inert latex beads (Fig. 3E). This result suggests that the effect of DFO and SIH on SCVs maturation is specific and possibly occurs through impairment of an active Salmonella-mediated process.

To ascertain that the DFO/SIH effect was not restricted to the RAW264.7 cell line, similar experiments were conducted in primary peritoneal macrophages infected with GFP-S (Fig. 3B). DFO (100 μM) caused a marked increase in the number of SCVs positive for M6PR in Nramp1^−/− (KO) macrophages (from 16% in controls to ≈60% in treated cells), a proportion similar to that seen in Nramp1-positive 129sv macrophages treated or not treated with DFO (Fig. 3B). Similarly, SIH increased the number of M6PR-positive SCVs formed in Nramp1-negative primary macrophages. The effect of DFO on M6PR recruitment was due at least in part to chelation of iron, as addition of excess iron at the time of invasion reduced the DFO effect (Fig. 3B).

An effect of DFO and SIH on accessibility of preformed SCVs to early endosomes was tested after labeling the early endocytic compartment with rhodamine dextran, which was used as a fluid phase marker. A possible DFO/SIH effect was monitored both in cultured RAW264.7 macrophages and primary peritoneal macrophages. A set of typical images is shown in Fig. 4A (GFP-S, green; rhodamine dextran, red; overlay, yellow), and colocalization results are quantitated in Fig. 4C (for RAW, RAW/Nramp1) and Fig. 4B (for 129sv, 129sv.KO). When cells were treated with DFO (100 μM), a strong increase in the number of dextran-accessible preformed SCVs was observed both in RAW264.7 macrophages (from 10% dextran-positive SCVs to 71%) and Nramp1 mutant (KO) primary macrophages (from 20% to 78%). The final proportion of dextran-accessible SCVs under these conditions was very similar to that seen in the corresponding cell type expressing Nramp1 (Fig. 4 B and C). SIH treatment produced similar results, whereas the membrane-impermeant HES–DFO had no effect on fusion of SCVs to early endosomes. Addition of FeSO₄ at the time of invasion partially reversed the effect of DFO in Nramp1-negative cells and also reduced the proportion of dextran-accessible SCVs formed in Nramp1-positive cells. These results confirm those obtained with M6PR (Fig. 3) and suggest that intracellular survival of Salmonella in permissive cells requires access to intracellular iron pools, which can be antagonized by iron chelators and/or Nramp1 activity.

Discussion

A large body of published data indicates that iron is a key determinant in onset, progression, and ultimate outcome of a number of infections (35). A delicate balance in body iron must be maintained for resistance to infections. On one hand, insufficient iron levels may impair certain antimicrobial defenses expressed by macrophages and neutrophils, including phagocytosis, cytokine production, respiratory burst, myeloperoxidase activity, and generation of oxygen radicals through the iron-dependent Haber–Weiss or Fenton reactions (35). On the other hand, excess iron (iron overload) may have a detrimental effect on host defenses, either through increased availability of nutritional iron to microbes or direct impairment of phagocyte functions (35, 36). Indeed, conditions associated with primary or secondary iron overload in humans such as E-β Thalassemia, hereditary hemochromatosis, and transfusional iron overload in patients with severe anemia are associated with increased susceptibility to infections with Gram-positive and -negative bacteria, including Salmonella (37, 38). Likewise, studies in mouse models of infection have also shown that induction of iron overload causes increased susceptibility to a lethal challenge with Salmonella (39). On the other hand, deprivation of nutritional iron in mice is associated with prolonged survival and decreased mortality from acute Salmonella infection (40).

These observations have suggested that an adequate supply of iron is absolutely essential for Salmonella virulence in vivo in general and for intracellular replication in macrophages in particular (41, 42). Indeed, Salmonella expresses under different conditions a surprising number of high- or low-affinity, ATP-dependent or proton-coupled (tonB-dependent) iron acquisition systems (Fe²⁺, Fe³⁺), such as fepBCDG, sitAD, FeoABC, CorAD, and the Nramp homolog MntH (43). Single mutations at feoB and sitA-D reduce virulence in vivo, and double mutations at MntH, sitA-D, or feoB completely abrogate virulence in Nramp1^−/− mutant 129sv mice (43, 44). Nonsiderophore metal chelators inhibit cell-free growth of Salmonella in vitro, and dipyridyl, an iron chelator, can significantly reduce the intracellular replication of both WT Salmonella, and particularly, Salmonella mutants partly impaired in iron acquisition, when studied in Nramp1-defective RAW264.7 macrophages (43). Thus, iron acquisition is key to intracellular survival and active replication in permissive mammalian cells.

Our working hypothesis is that Nramp1 functions as a pH-dependent efflux pump at the phagosomal membrane to create an intraphagosomal environment that is limiting for at least Mn²⁺ and Fe²⁺. This theory is based on the observations that: (i) in live macrophages, Nramp1 recruitment to the phagosomal membrane enhances in a pH-dependent fashion the Mn²⁺-dependent quenching of Fura-FF6 covalently attached to zymosan particles (27); (ii) an Nramp1 mutant showing altered intracellular trafficking can transport Fe²⁺ and Mn²⁺ at the plasma membrane with characteristics similar to that of its close homolog Nramp2 (J. R. Forbes and P.G., unpublished data); and (iii) Nramp2 acts as a Fe²⁺ efflux pump at the membrane of acidified endosomes, an activity that is impaired in reticulocytes from anemic Nramp2 mutant mk mice (19). Nramp1-mediated depletion of divalent metals from the phagosomal space seems to create a stressful and growth-limiting environment that is actively sensed by the bacteria within it. This proposition is supported by recent experiments with Salmonella carrying a number of transcriptional LacZ fusions that were used to infect Nramp1-positive and -negative mice in vivo, and corresponding macrophages ex vivo. These studies showed that intracellular Salmonella respond to the presence of Nramp1 by induction of a number of “virulence” genes that map within Salmonella pathogenicity island 2 (SPI2), including ssrA and sseJ (45). The Nramp1 effect on SPI2 gene induction in vivo can also be reproduced in vitro in cell-free conditions after addition of iron chelators (such as dipyridyl) to Salmonella culture medium (45). Similarly, we have noted an effect of Nramp1 on the level of transcriptional induction of the mycobacterial gene MbtB (J. R. Forbes and P.G., unpublished data) that participates in the synthesis of iron siderophores (mycobactins) and that is absolutely required for intracellular survival of Mycobacterium tuberculosis (46). Finally, the intracellular survival strategy of Salmonella involves seclusion in an SCV that remains negative for M6PR, a protein known to regulate the delivery of a subset of lysosomal enzymes from the trans-Golgi network to the prelysosomal compartment (47). Also, SCVs do not fuse with early endosomes in permissive cells. Recruitment of Nramp1 to the SCVs antagonizes both of these effects, likely through depletion of luminal divalent cations (10).

The current study shows that in macrophages chelator-mediated depletion of intracellular iron stimulates recruitment of M6PR and fusion of SCVs to recycling endosomes. Because the chelators used, in particular SIH, are membrane-permeant, the observed effect may be similar to that produced by Nramp1, involving depletion of intraphagosomal iron and resulting in increased bacteriostasis and/or bactericidal activity. Although there is general agreement that iron-limiting conditions may adversely affect intracellular survival of Salmonella, a recent report has suggested that DFO may actually stimulate intracellular replication of Salmonella. Indeed, DFO but not lactoferrin treatment was found to strongly stimulate Salmonella replication in tissues in vivo at 24 h (48). The interpretation of in vivo results with DFO treatment may be complicated by pleiotropic effects of DFO on both the host and the bacterium, including the possibility of acting as an exogenous chelator for iron, increasing bioavailability to bacteria. This is the case for Yersinia enterocolitica, where DFO treatment markedly increases virulence and stimulates growth in mice (49–51). Collins et al. (48) also reported a modest stimulatory effect of DFO, but not lactoferrin, on growth of Salmonella in bone marrow macrophages ex vivo 6 h after infection. These results are somewhat different from those of Boyer et al. (43) who showed that dipyridyl treatment reduced intracellular replication of either WT Salmonella or of Salmonella mutants partly impaired for iron acquisition in RAW267.4 macrophages. The reasons for the apparent discrepancy are unclear and will require further experimentation, but may reflect differences in strain of Salmonella, type of macrophages, type, dose, and duration of exposure to iron chelators used in both studies.

The mechanism by which Nramp1-mediated depletion of divalent cations from the phagosome adversely affects replication of otherwise unrelated intracellular pathogens is not completely understood. Intracellular pathogens under Nramp1 control in vivo have evolved different strategies to block phagosome maturation and survive intracellularly. In the case of Salmonella, SCVs neither acquire M6PR nor the lysosomal enzyme cathepsin L, which is delivered from the trans-Golgi network to endocytic compartments through this receptor, although they acquire other M6PR-independent lysosomal glycoproteins, including Lamp1 (52). This is an active process that is observed only in phagosomes containing live Salmonella but not in those containing heat-inactivated Salmonella (ref. 10 and data not shown). Mycobacterium-containing phagosomes formed in permissive cells also show a maturation block (53). This process is characterized by reduced acidification, reduced fusion to lysosomes, and increased retention of endosomal markers and is observed only in phagosomes containing live Mycobacteria (28, 29). Thus, the active process by which Salmonella or Mycobacterium arrest phagosome maturation may strictly depend on an Mn²⁺- or Fe²⁺-dependent enzymatic activity that may be directly antagonized by Nramp1 transport at the phagosomal membrane. Alternatively, the Nramp1 effect could be more general and secondary to deprivation of nutritional metals required for normal metabolic activity of the pathogen. Finally, Mn²⁺ and Fe²⁺ are key cofactors required for activity of pathogen-encoded antioxidant defenses such as catalase, peroxidase, and superoxide dismutase (54, 55). Nramp1-mediated deprivation of cations from the phagosomal space would impair such protective enzymatic activity, causing a net enhancement of the bactericidal activity of the phagocyte.

Abbreviations

SCV

Salmonella-containing vacuole

M6PR

mannose-6-phosphate receptor

TPEN

N,N,N′,N′-tetrakis 2-pyrimethylethylenediamine

DFO

desferrioxamine

SIH

salicylaldehyde isocotinoyl hydrazone

HES

hydroxyethyl starch

GFP-S

GFP-expressing Salmonella

KO

knockout