Iron, manganese, and cobalt transport by Nramp1 (Slc11a1) and Nramp2 (Slc11a2) expressed at the plasma membrane

Mutations in the mouse Nramp1 gene (natural resistance associated macrophage protein 1, also known as Slc11a1; OMIM [Online Mendelian Inheritance in Man] #600266) cause susceptibility to infection by several intracellular pathogens including Mycobacterium, Leishmania, and Salmonella.¹  Likewise, polymorphic variants of NRAMP1 are associated with human susceptibility to tuberculosis and leprosy in endemic areas of disease.^2,3 Nramp1 mRNA is abundant in mouse macrophages and human neutrophils, where it encodes a 90- to 100-kDa integral membrane phosphoglycoprotein⁴  present in the Lamp1-positive late endosomes and lysosomes⁵  and in gelatinase-positive tertiary granules,⁶  respectively. Following phagocytosis, Nramp1 is rapidly recruited to and remains associated with the membrane of phagosomes containing either inert particles⁵  or live bacteria/parasites.^7-9  Recruitment of Nramp1 to the membrane of Mycobacteria containing phagosomes is associated with bacteriostasis, bacterial damage, and appears to antagonize the ability of Mycobacterium to block phagolysosomal fusion and acidification.^10,11  Similarly, Nramp1 appears to impair the ability of Salmonella to shelter in a vacuole that does not fuse to early endosomes and that remains negative for the mannose-6-phosphate receptor.⁹ 

Nramp1 has a close mammalian paralog Nramp2 (OMIM #600523; also known as Dmt1, Dct1, Slc11a2)¹²  that has been functionally characterized. Transport studies using Xenopus laevis oocytes¹³  and mammalian cell lines^14,15  have demonstrated that Nramp2 is a broad specificity divalent-metal transporter that functions in a pH-dependent fashion, stimulated by acidic pH, which is suggestive of a proton/metal-symport mechanism.¹³  In mice, Nramp2 is expressed at the duodenum brush border, where it is responsible for transferrin-independent uptake of dietary iron from the intestinal lumen.¹⁶  Nramp2 also colocalizes with transferrin in the recycling endosomes of many cell types, including reticulocytes,^17,18  where it transports iron from the acidified lumen of the endosomes into the cytoplasm.^19-21  A mutation in Nramp2 impairs both of these aspects of iron homeostasis and causes microcytic anemia in mk mice and Belgrade rats.^17,20,22 

These studies of Nramp2 have suggested a role for Nramp1 in metal transport at the phagosomal membrane. However, controversy regarding the mechanism of Nramp1 metal transport with respect to protein topology and the direction of metal transport in relation to the proton gradient has persisted.²³  Previously, we used a metal-sensitive fluorophor (Fura-FF6) chemically coupled to zymosan particles to monitor divalent-metal flux by means of real-time microfluorescence imaging across the membrane of single phagosomes formed in live Nramp1^+/+ and Nramp1^–/– primary macrophages.²⁴  Nramp1-positive phagosomes exhibited reduced intraphagosomal accumulation of externally added Mn²⁺ ions. Likewise, Nramp1-positive phagosomes showed increased release of Mn²⁺ ions from preloaded Fura-FF6–zymosan particles. Mn²⁺ transport by Nramp1 was abrogated by bafilomycin (vacuolar H⁺/ATPase inhibitor), suggesting that Nramp1 functions to efflux Mn²⁺ ions in a pH-dependent fashion from acidified phagosomes down the proton gradient. This mechanism is similar to Fe²⁺ transport by Nramp2 across the membrane of acidified endosomes, implying that Nramp1/2 share a common transport mechanism and suggests that Nramp1 exerts its antimicrobial activity through depletion of divalent metals from the phagosomal space. This hypothesis is in agreement with results from other independent studies.^25-28 

In contrast, increased Nramp1-dependent accumulation/binding of isotopic Fe²⁺ into isolated phagosomes containing either Latex beads or M avium has been reported.^29-31  Increased Fe²⁺ accumulation was blocked by anti-Nramp1 antibodies, suggesting that Nramp1 may transport cytoplasmic Fe²⁺into phagosomes. In an independent study, injection of Nramp1 mRNA into X laevis oocytes induced small Zn²⁺-dependent inward currents suggestive of metal uptake.³²  Additional studies of pH-dependent transport of isotopic Zn²⁺ led these authors to conclude that Nramp1 may transport cytoplasmic metals into the phagosome by a proton/divalent-metal antiport mechanism. These authors proposed that increased phagosomal Fe²⁺ would stimulate oxygen radical production in situ via the Fenton reaction resulting in increased bactericidal activity.^29-32  However, phagosomal metal influx mediated by Nramp1 requires that Nramp1 would have to be mechanistically distinct from Nramp2, with respect to direction of transport, use of the transmembrane pH gradient, and/or membrane topology of the proteins.

In order to address these differing conclusions regarding the transport function of Nramp1, we sought to directly compare Nramp1 and Nramp2 proteins. Should Nramp1/2 have the same membrane organization and work by the same mechanism, ectopic expression of Nramp1 at the plasma membrane of whole cells would, like Nramp2, be expected to result in the pH-dependent uptake of metals from the extracellular milieu. To test this proposal, independent Chinese hamster ovary (CHO) transfectants expressing HA-tagged Nramp1 protein at the plasma membrane were created. Comparative analysis of Nramp1 and Nramp2 activity was performed with respect to transport function, pH dependence, and metal ion selectivity. These studies show that the 2 proteins are mechanistically indistinguishable but may have different substrate selectivity.

A mammalian expression plasmid (Nramp1HA-pCB6) containing a neomycin resistance gene together with the complete Nramp1 coding sequence modified by the in-frame insertion of a hemagglutinin (HA) epitope (YPYDVPDYAS) at amino-acid position 330 was constructed as was previously described for the corresponding HA-tagged Nramp2 construct (Nramp2HA-pCB6).¹⁵  CHO LR73 cells³³  were cultured as previously described¹⁵  and transfected with Nramp1HA-pCB6 using a calcium-phosphate coprecipitation method.³⁴  Stably transfected clones were selected (geneticin, 1 mg/mL; Invitrogen, Burlington, ON, Canada) for 10 days, followed by isolation and expansion of individual clones. Total protein extracts were prepared from CHO transfectants, and Nramp1HA protein expression was analyzed by immunoblotting with anti-HA monoclonal antibodies. The CHO transfectant N2-310a stably expressing Nramp2HA was previously reported,¹⁵  as were the CHO transfectants stably expressing Nramp1/2 proteins modified with a C-terminal c-Myc epitope (Nramp1Myc, Nramp2Myc).^5,18  All chemicals were purchased from Sigma Chemical (Oakville, ON, Canada) unless otherwise noted.

Total cell protein extracts were prepared by solubilizing cell pellets in Tris [tris(hydroxymethyl)aminomethane]-buffered saline (TBS; 100 mM Tris-HCl pH 7.5, 150 mM NaCl) containing 1% Triton X-100, 1 mM PMSF (phenylmethanesulfonyl fluoride), 2 μg/mL leupeptin, 2 μg/mL aprotinin, 1 μg/mL pepstatin, 2 mM EDTA (ethylenediaminetetraacetic acid), and 20% glycerol (20 minutes, on ice), and the insoluble material was removed by centrifugation (16 000 g, 10 minutes, 4°C). Crude membrane fractions were prepared as previously described.³⁵  Protein concentrations were determined using the Bradford assay (BioRad, Missisauga, ON, Canada). Discontinuous sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE) was done with a 4% polyacrylamide stacking gel and a 10% separating gel.³⁶  Proteins were mixed with sample buffer³⁶  and denatured (20°C, 20-30 minutes) prior to loading onto gels. Prestained molecular mass markers (BioRad) were included in all SDS-PAGE experiments.

For immunoblotting, proteins separated by SDS-PAGE were transferred electrophoretically to polyvinylidine difluoride membranes (4°C, 450 V/hr; Schleicher and Schuell, Keene, NH) in buffer containing 20% methanol and 0.01% SDS.³⁷  Membranes were blocked (30 minutes-1 hour, 20°C) in TBS containing 0.25% Tween-20 (TBST) and 5% nonfat skim milk. Membranes were then incubated in blocking buffer for either 3 hours at 20°C or 16 hours at 4°C with one of the following primary antibodies: mouse monoclonal anti–HA 16B12 or anti–c-Myc 9E10 antibodies (both used at 1/1000; Covance, Princeton, NJ), or affinity purified polyclonal rabbit anti-Nramp2NT or anti-Nramp1NT antibodies (used at 1/1000), which are directed against the amino-terminus of Nramp2 and Nramp1, respectively.^4,18  Membranes were washed in TBST (3 × 5 minutes, 20°C) prior to incubation with horseradish peroxidase–conjugated goat anti–rabbit IgG or anti–mouse IgG (1/10 000; Perkin-Elmer, Woodbridge, ON, Canada) secondary antibodies (1 hour, 20°C) in blocking buffer. Membranes were washed in TBST (4 × 5 minutes, 20°C), and specific immune complexes were revealed by enhanced chemiluminescence (ECL; Perkin-Elmer) and autoradiography (Kodak, Rochester, NY).

Immunofluorescence on nonpermeabilized whole cells was done as previously described.¹⁵  Live cells grown on coverslips were blocked (15 minutes, 4°C) and then incubated (1 hour, 4°C) with mouse anti–HA monoclonal antibodies (diluted 1/50) or with affinity-purified rabbit polyclonal anti–Nramp1NT antiserum (diluted 1/50). Cells were then washed extensively, fixed with 4% paraformaldehyde (20 minutes, on ice), and incubated (1 hour, 20°C) with either goat anti–mouse IgG Cy3 (1/200; Jackson Laboratories, West Grove, PA) or goat anti–rabbit IgG rhodamine (1/100; Jackson Laboratories) secondary antibodies. Following extensive washing, the coverslips were mounted onto glass slides using Permafluor antifade reagent (Shandon, Pittsburgh, PA). The cells were photographed by epifluorescence microscopy using a Nikon (Tokyo, Japan) Eclipse E800 microscope mounted with a 60 × oil-immersion objective lens and a Nikon DXM1200 digital camera.

Attached cells were rinsed twice with ice-cold phosphate-buffered saline (PBS) and once with ice-cold borate buffer (10.0 mM boric acid, 154 mM NaCl, 7.2 mM KCl, 1.8 mM CaCl₂, pH 9.0), and then incubated (15 minutes, on ice) in the same buffer containing Sulfo-NHS-LC-biotin (0.5 mg/mL; Pierce, Milwaukee, WI). Unreacted biotin was removed by 3 washes with TBS containing 0.2 M glycine and one wash with TBS. Total cell protein extracts were prepared from the biotinylated cells. Biotinylated proteins were captured by incubation (16 hours, 4°C) of 100 μg of total cell extracts with 50 μL of a 50% (wt/vol) slurry of streptavidin-agarose beads (Pierce) in 500 μL (total) of TBS buffer containing 1% Triton X-100 and protease inhibitors. Beads were washed 5 times with TBS/1% Triton X-100, and bound proteins were eluted in 50 μL of sample buffer³⁶  containing 5% β-mercaptoethanol (30 minutes, 20°C). Captured proteins were analyzed by SDS-PAGE and immunoblotted with either anti–HA or anti–cMyc monoclonal antibodies, or with anti–Nramp1NT or anti–Nramp2NT affinity-purified antisera.

Cells were detached using PBS/citrate (5-10 minutes, 37°C) and resuspended in loading medium (α-MEM [minimum essential media], 0.5 mg/mL BSA [bovine serum albumin], 25 mM HEPES [N-2-hydroxyethylpiperazine-N′-2-ethanesulfonic acid] pH 7.4). Cells (1 × 10⁶ per assay) were centrifuged and resuspended in prewarmed (37°C) loading medium containing either 0.25 μM calcein-AM (acetoxy-methylester) or 2 μM Fura2-AM (1 mM stock solutions in 100% DMSO [dimethyl sulfoxide]; Molecular Probes, Eugene, OR) followed by incubation at 37°C for 10 minutes or 1 hour, respectively. After this loading period, cells were washed and resuspended in loading medium. Transport assays were performed with a Perkin-Elmer LS-50B fluorometer equipped with a stirring and water-jacketed cuvette holder. To measure metal-dependent calcein/Fura-2 quenching, the fluorometer settings were excitation λ = 488/360 nm and emission λ = 517/510 nm with 5/7.5 micron bandpass slit widths. Immediately prior to each transport assay, an aliquot of cells was washed once in PBS (37°C), resuspended in 500 μL transport buffer (37°C; 150 mM NaCl, MES [2-(N-morpholino)ethanesulfonic acid] pH 5-6 or HEPES pH 7-8), transferred into a prewarmed cuvette, and thereafter cell-associated fluorescence was recorded continuously (0.5-second intervals). After a stabilization period (60 seconds), divalent-metal was added and cell-associated fluorescence was recorded for an additional 120 seconds. Divalent-metals MnCl₂ and CoCl₂ were prepared as 2 mM stock solutions in water. Iron was prepared fresh as a 2 mM FeNH₄SO₄ stock solution in transport buffer with a 25:1 molar ratio of sodium ascorbate (50 mM) to maintain the metal in its reduced form. Fluorescence quenching data were normalized for individual samples to the fluorescence value taken at 70 seconds to facilitate visual presentation of the data.

Cells were harvested as described for the fluorescent assay and resuspended (10⁷ cells per assay) in 1.5 mL of transport buffer (25 mM Tris, 25 mM MES, 140 mM NaCl, 5.4 mM KCl, 5 mM glucose, 1.8 mM CaCl₂, pH 5.5). Transport was initiated by addition of 1 mL of radioisotope buffer, followed by incubation at 20°C. Mn²⁺ radioisotope buffer was transport buffer containing 0.45 μM ⁵⁴Mn (⁵⁴MnCl₂, 13.4 Ci/mmol [495.8 GBq/mmol]; Perkin-Elmer) and 22.05 μM MnCl₂ (22.5 μM total Mn, 49:1 molar ratio of cold MnCl₂:⁵⁴Mn), giving a final concentration of 9 μM Mn²⁺ in each transport reaction. Fe²⁺ radioisotope buffer contained 1.125 μM ⁵⁵Fe (⁵⁵FeCl₃, 3.012 Ci/mmol [111.4 GBq/mmol]; Perkin-Elmer) and 21.375 μM FeNH₄SO₄ (22.5 μM total Fe, 19:1 molar ratio of cold FeNH₄SO₄/⁵⁵Fe) together with 1.125 mM sodium ascorbate (50:1 molar ratio ascorbate–Fe) giving a final concentration of 9 μM Fe²⁺. At predetermined time intervals (0, 5, 15, 30 minutes), 500 μL cell aliquots were transferred to microcentrifuge tubes containing a 200-μL oil cushion (4:1 silicon oil-mineral oil). Cells were pelleted by centrifugation (12 000 g, 10 seconds) through the oil cushion, the aqueous phase was removed, and the walls of the tube were washed with transport buffer. The oil cushion was removed, and the cell pellets were digested with 0.1N NaOH. Lysates were neutralized by addition of an equal volume of 0.1N HCl, and cell-associated radioactivity was determined by liquid scintillation counting. The protein content of each lysate was determined using the Bradford assay (BioRad). Background radioisotope binding was determined by parallel control transport experiments performed on ice. For metal ion selectivity studies, cells (3 × 10⁶) were resuspended in 375 μL transport buffer, and transport was initiated by addition of an equal volume of radioisotope buffer. Stock Mn²⁺ radioisotope buffer contained 0.4 μM ⁵⁴Mn and 19.6 μM MnCl₂ (20 μM total Mn), and stock Fe²⁺ radioisotope buffer contained 1 μM ⁵⁵Fe and 19 μM FeNH₄SO₄ (20 μM total Fe) in 1 mM sodium ascorbate. Cells were incubated for 10 minutes with serial 2-fold dilutions (0.3125-10 μM final) of these stocks, and the amount of cell-associated radioactivity was determined as described above.

The aim of the present study was to express Nramp1 at the plasma membrane (PM) of CHO cells, where its transport properties could be studied and compared to Nramp2. Previous studies done in primary macrophage, together with RAW264.7 and/or CHO cells transfected with Nramp1 and Nramp2 constructs modified by the addition of a C-terminal c-Myc tag (Nramp1Myc, Nramp2Myc), indicated different subcellular distributions of the 2 proteins.^5,18  Nramp2Myc was found to be expressed at the PM and in recycling endosomes, whereas Nramp1Myc was not found at the PM but was detected in lysosomes/late endosomes. Likewise, an Nramp2 protein modified by the insertion of an HA epitope (Nramp2HA) into the predicted loop delineated by putative transmembrane (TM) domains 7 and 8 was shown to be functional and expressed at the PM of CHO cells with the HA epitope extracellularly located and accessible to antibodies added to the external medium.¹⁵  In an attempt to identify cell lines expressing Nramp1 at the PM, Nramp1 cDNA was similarly modified by insertion of HA epitope in the TM7/8 loop, followed by transfection into CHO cells. Several transfectants stably expressing Nramp1HA protein were identified (Figures 1, 2, 3; N1-1816, N1-94, N1-116, and N1-123). The possibility of PM localized Nramp1HA protein was examined by immunofluorescent detection of the HA epitope in nonpermeabilized cells, and representative images are shown in Figure 1. In nonpermeabilized N1-1816 cells, Nramp1HA staining was detected at the cell periphery by extracellular anti–HA monoclonal antibodies (Figure 1B), revealing a staining pattern similar to that observed in control N2-310a cells expressing Nramp2HA (Figure 1C). PM staining of N1-1816 cells was Nramp1HA-specific, as it was absent in untransfected CHO cell controls (Figure 1A) and in N1-1816 cells stained with the secondary antibody alone (Figure 1D). This demonstrates that the predicted TM7/8 loop of Nramp1HA is extracellular in N1-1816 cells. Additionally, Nramp1HA was detected by an anti–Nramp1NT rabbit polyclonal antiserum directed against the amino terminus of the protein in Triton X-100 permeabilized N1-1816 cells (Figure 1F), but not in intact N1-1816 cells (Figure 1E), demonstrating that the N-terminus of Nramp1HA in N1-1816 cells was intracellular. Together, these results indicate that Nramp1HA was expressed at the PM, and that it has a membrane topology similar to that of Nramp2HA.

Cell-surface biotinylation was used to verify PM expression of Nramp1/2HA proteins. Live CHO, N1-1816 (Nramp1HA), and N2-310a (Nramp2HA) cells were reacted with a membrane-impermeant biotinylating reagent (Sulfo-NHS-LC-biotin), and biotinylated proteins were captured using streptavidin-conjugated agarose beads. Proteins present in the streptavidin-captured fraction (the entire eluate was loaded; “B” in Figure 2A) were analyzed together with proteins remaining in the supernatant fraction after streptavidin-capture (10% [vol] of the postcapture supernatant was loaded; “S” in Figure 2A) by immunoblotting with anti-HA antibody. Both Nramp1HA and Nramp2HA proteins were biotinylated and subsequently detected in the streptavidin-captured fraction (“B”) of the corresponding CHO transfectants (Figure 2A). Nramp1/2HA protein detection was specific, as it was absent in both biotinylated untransfected CHO controls and in unbiotinylated Nramp1/2HA cell extracts incubated with streptavidin beads (Figure 2A). Neither the abundant cytosolic protein tubulin nor the endo/lysosomal mannose-6-phosphate receptor proteins were detected in biotinylated fractions of the cell lines examined (data not shown). These data demonstrate that biotinylated Nramp1HA and Nramp2HA proteins in the streptavidin-captured fraction were due to surface labeling of PM-localized Nramp1/2HAproteins, as opposed to labeling of intracellular protein pools by the biotinylating reagent.

In previous studies of CHO cells expressing Nramp1/2Myc proteins, Nramp2Myc was detected by immunofluorescence at the PM, while Nramp1Myc was not.^5,18  Thus, the surface biotinylation of Nramp1/2Myc (Figure 2B) proteins expressed in CHO cells was compared to that of Nramp1/2HA (Figure 2A). Both Nramp1Myc and Nramp2Myc (Figure 2B) were biotinylated and detected in the streptavidin-captured fraction (“B”) by immunoblotting with an anti-Myc antibody. However, the relative proportion of Nramp1Myc in this fraction compared to that of Nramp1HA was considerably lower (Figure 2A-B, compare “B” to “S”), whereas surface-biotinylated Nramp2HA and Nramp2Myc proteins appeared to be detected in streptavidin-captured fractions in similar amounts under the same conditions (Figure 2A-B, compare “B” to “S”). Direct comparison of Nramp1HA and Nramp1Myc proteins in streptavidin-captured fractions by immunoblotting with anti–Nramp1NT polyclonal antiserum (Figure 2C, compare “B” to “S”) further illustrates that Nramp1Myc was detectable in surface-biotinylated fractions but in a significantly lower proportion compared to Nramp1HA. This suggests that insertion of the HA tag in the predicted TM7/8 loop of Nramp1 was responsible for the increased PM localization of Nramp1HA protein.

Immunoblot analysis also showed that in contrast to Nramp1Myc, which is expressed as a glycoprotein of ∼ 90 to 100 kDa in CHO cells, Nramp1HA is present as 2 prominent bands of ∼ 90 to 100 kDa and of ∼ 50 kDa. The ∼ 50 kDa was particularly apparent when the proteins were immunoblotted against anti–Nramp1NT polyclonal antibodies. Pulse-chase metabolic labeling of Nramp1HA protein with ³⁵S-methionine followed by immunoprecipitation suggested that the lower ∼ 50 kDa band consisted of a mixture of newly synthesized polypeptide as well as degradation products and/or partially glycosylated Nramp1HA (data not shown).

Three additional stably transfected CHO clones expressing significant amounts of Nramp1HA (N1-94, N1-116, N1-123) were identified (Figure 3). Crude membrane preparations from these clones were analyzed by immunoblotting with anti-Nramp1NT, anti-Nramp2NT, or anti-HA antibodies (Figure 3A). The results confirmed the presence of a ∼ 50 kDa and ∼ 90 to 100 kDa immunoreactive species in all Nramp1HA transfectants. These clones were analyzed for presence of Nramp1HA protein at the PM by surface biotinylation, followed by analysis of the streptavidin-captured proteins by immunoblotting (Figure 3B). N2-310a (Nramp2HA) and N1-1816 (Nramp1HA) transfectants were used as positive controls and CHO cells as negative controls. These experiments confirmed results of transfectant N1-1816 (Figure 2) by showing surface labeling/PM localization of Nramp1HA in transfectants N1-94, N1-116, and N1-123. Interestingly, the ∼ 90 to 100 kDa species was predominantly biotinylated, compared to the 50 kDa species, indicating that the former comprises most of the Nramp1HA protein present at the cell surface.

The possibility of Nramp1HA metal transport activity at the PM was investigated using transfectants N1-94, N1-1816, N1-116, N1-123 that express different amounts of Nramp1HA. Nramp2HA expressing transfectant N2-310a and CHO cells were used as positive and negative controls, respectively (Figure 4). Metal transport was investigated in whole cells using a fluorescence quenching assay.¹⁵  Cells were loaded with the membrane-permeant acetoxymethylester (AM) form of the dyes calcein or Fura-2, which are subsequently de-esterified in the cytosol releasing the fluorescent, membrane-impermeant, metal-sensitive probe. The effect of extracellularly added Fe²⁺ (Figure 4A) or Co²⁺ (Figure 4C) on intracellular calcein fluorescence was measured. Similarly, the effect of extracellularly added Mn²⁺ on intracellular Fura-2 fluorescence (Figure 4B) was monitored for 200 seconds. Experiments were carried out at pH 6.0, a pH known to be optimal for Nramp2 activity in this assay.^15,38  Typical fluorescence quenching traces are shown in Figure 4A-C, and the rate of quenching was calculated from the initial linear slope of individual traces generated in 3 to 6 independent experiments (Figure 4D-F). For each of the metal/fluorophor combinations tested, Nramp1HA-expressing cells demonstrated substantial, rapid, time-dependent, and statistically significant (Figure 4) fluorescence quenching compared to CHO controls. Background fluorescent quenching in CHO-negative control cells was identical in 5 independent isolates (data not shown). Examination of the rate of quenching measured in independent Nramp1HAtransfectants for each divalent-metal/fluorophor combination (Figure 4D-F) suggested that divalent-metal uptake was proportional to the amount of Nramp1HA protein expressed in each cell line and to the amount of Nramp1HA protein localized to the PM as determined by immunoblot analysis of the corresponding total cell membrane or surface biotinylated fraction (Figure 2). Clone N1-94, which expresses the largest amount of Nramp1HA, exhibited initial rates of Fe²⁺,Mn²⁺, and Co²⁺ uptake that were respectively 2.1 ± 0.1-, 2.7 ± 0.2-, and 2.1 ± 0.1-fold greater than background measured in CHO cells. These results demonstrate that Nramp1HAprotein is capable of Fe²⁺,Mn²⁺, and Co²⁺ uptake at the PM.

A hallmark of metal transport by eukaryotic and prokaryotic members of the Nramp protein superfamily is that transport is pH dependent, suggesting a proton-symport mechanism.²³  Therefore, the pH dependence of divalent-metal transport by Nramp1HA at the PM was investigated. For this, CHO and N2-310a (Nramp2HA) controls together with N1-94 cells (Nramp1HA) were loaded with Fura-2 and Nramp1/2HA-dependent Mn²⁺ uptake was monitored by fluorescence quenching at different extracellular pH (Figure 5). Typical fluorescence quenching traces for each cell line are shown for pH 5.0 (Figure 5A), 6.0 (Figure 5B), and 7.0 (Figure 5C). The rate of quenching at each pH was calculated from the initial linear slope of individual traces generated in 3 to 5 independent experiments (Figure 5D). These rate data are also expressed relative to the CHO cell quench rate at each pH (Figure 5E). Nramp1HA-dependent quenching of Fura-2 fluorescence by Mn²⁺ was strongly pH dependent: Transport compared to CHO controls was substantial at acidic pH 5.0 and 6.0, but was minimal at neutral pH 7.0 (Figure 5E). This behavior was identical to that seen for Nramp2HA tested under the same experimental conditions (Figure 5). Similar results were obtained when Co²⁺ and Fe²⁺ were used as transport substrates (data not shown). Therefore, results from fluorescence quenching assays demonstrate that Nramp1HA is transport competent at the PM and indicate that metal transport by Nramp1HA is pH dependent and mechanistically similar to Nramp2HA.

Divalent-metal transport activity for Nramp1HA also was investigated using radioisotopic ⁵⁴Mn or ⁵⁵Fe. For these studies, control CHO cells, N1-94 (Nramp1HA) and N2-310a (Nramp2HA) transfectants were used. In the first set of experiments, cells were incubated with tracer amounts of radioisotopic ⁵⁵Fe or ⁵⁴Mn in a total metal concentration of 9 μM at pH 5.5. At predetermined time intervals over a 30-minute period, cells were pelleted by centrifugation through an oil cushion to remove unincorporated metal, and the cell-associated radioactivity was determined. Results from 3 to 5 independent experiments are shown in Figure 6A (⁵⁵Fe²⁺) and 6B (⁵⁴Mn²⁺). Expression of Nramp1HA or Nramp2HA strongly stimulated accumulation of ⁵⁵Fe²⁺ and ⁵⁴Mn²⁺ into transfected CHO cells. For both proteins, metal uptake was time dependent and increased steadily over the 30-minute incubation period. Metal transport was temperature dependent, being abrogated at 4°C (Figure 6A-B). Subtraction of the nonspecific metal binding to cells measured at 4°C gives the net amount of metal uptake, which is shown in Figure 6C (⁵⁵Fe²⁺) and Figure 6D (⁵⁴Mn²⁺). Over 30 minutes, Nramp1HA caused a 4- and 10-fold stimulation of ⁵⁵Fe²⁺ and ⁵⁴Mn²⁺ uptake, respectively, compared to 3- and 4-fold stimulation for Nramp2HA-expressing cells. Thus, Nramp1HA appeared to transport Mn²⁺ to a greater degree than Nramp2HA, while both proteins showed similar uptake of Fe²⁺.

The substrate selectivity of the Nramp1/2HA transporters for Fe²⁺ and Mn²⁺ was characterized in dose-response experiments (Figure 7). In these experiments, the total concentration of Fe²⁺ and Mn²⁺ in the transport buffer was varied between 0.31 μM and 10 μM. Cell-associated radioactivity was determined (after 10 minutes) at each divalent-metal concentration. Parallel transport assays were carried out at 20°C and at 4°C. The 4°C binding data were subtracted from each corresponding 20°C data point. In the case of Fe²⁺ (Figure 7A), results were very similar for Nramp1HA and Nramp2HA with transport reaching a plateau at approximately 1 μM, closely approximating that previously determined for Nramp2 in fluorescence quenching transport studies.¹⁵  In the case of Mn²⁺ (Figure 7B), transport appeared to plateau at approximately 2.5 μM for both Nramp1HA and Nramp2HA. However, Nramp1HA appeared to be a more efficient transporter for Mn²⁺ than Nramp2HA, with a minimum of 2-fold increases in total cellular accumulation of the metal over all concentrations tested, in agreement with results shown in Figure 6B. Together, these results confirm and extend those obtained in fluorescence quenching studies (Figures 4, 5) and establish that Nramp1HA (1) is active at the plasma membrane, (2) can transport Fe²⁺, Mn²⁺, and Co²⁺, (3) is acid-pH dependent, (4) is mechanistically indistinguishable from Nramp2HA, and (5) appears to be a more efficient transporter of Mn²⁺ than is Nramp2HA.

Both the mechanism of transport and the substrate specificity of Nramp1 at the phagosomal membrane have proven difficult to study. In order to overcome this difficulty, we have inserted an HA epitope into the TM7/8 loop of Nramp1 (Nramp1HA) and expressed the recombinant protein at the plasma membrane of CHO cells. PM expression in CHO transfectants was demonstrated by direct extracellular accessibility of the HA tag to antibodies in whole cells using immunofluorescence (Figure 1), as well as by cell-surface biotinylation experiments (Figures 2, 3). These studies provide topologic information for Nramp1 and establish that the TM7/8 loop, which bears a number of predicted N-linked glycosylation sites,³⁹  is extracellular when Nramp1HA is expressed at the PM. By inference, the TM7/8 loop of Nramp1 would be found in the lumen of phagosomal vesicles. We show also that the N-terminus of Nramp1HA, which is only accessible to anti–Nramp1NT antibodies in permeabilized but not in whole cells, is cytoplasmic. Thus, the membrane organization of Nramp1HA at the PM is identical to that established previously for Nramp2HA,¹⁵  in agreement with the high degree of shared sequence similarity (78%) and identity (64%).¹² 

The PM localization of Nramp1HA in transfected CHO cells is unique and different from that of the endogenous protein found in either primary macrophages or neutrophils, where Nramp1 was not at the PM but was found in lysosomes or tertiary granules, respectively.^5-7  Likewise, recombinant Nramp1Myc expression in CHO cells or in RAW264.7 macrophages was restricted to the late endosomes and lysosomes and was not detected at the PM.⁵  Unsurprisingly, expression of Nramp1Myc in CHO cells did not stimulate uptake of extracellular divalent-metal.¹⁵  This suggests that insertion of the HA epitope into the TM7/8 loop may directly alter targeting, maturation, or processing of Nramp1HA, resulting in its accumulation at the PM. Several endosomal/lysosomal proteins transiently pass through the PM, prior to final localization via endocytotic retrieval.^40,41  The relatively high level of Nramp1HA PM accumulation compared to Nramp1Myc detected in transfected CHO cells may reflect partial or complete uncoupling of this retrieval process due to the presence of HA tag. Also, the TM7/8 loop is predicted to have several N-glycosylation sites.³⁹  The structure or processing of these sites may be altered by the inserted HA tag, thereby contributing to increased PM Nramp1HA accumulation. Indeed, apical PM targeting of Nramp2 in polarized MDCK (Madin Darby canine kidney) cells has been shown to depend on N-glycosylation.⁴²  Likewise, the effect of N-glycosylation on lysosomal protein targeting recently has been demonstrated for the protein endolyn.⁴³  Disruption of N-glycosylation in MDCK cells was shown to cause redistribution of endolyn from an apical PM lysosomal sorting pathway to the basolateral PM.

Our results from fluorescence quenching transport assays (Figures 4, 5) and transport studies with isotopic ⁵⁵Fe²⁺ and ⁵⁴Mn²⁺ (Figures 6, 7) show that Nramp1HA expressed at the PM is transport competent and functions as a multispecific divalent-metal transporter. Divalent-metal transport by Nramp1HA was time and temperature dependent and required an acidic pH. These Nramp1HA transport characteristics are identical to those demonstrated for Nramp2HA in our study, and they closely parallel results from independent transport studies of Nramp2 expressed either in Xenopus laevis oocytes¹³  or transfected mammalian cells,^14,15  as well as those reported for other eukaryotic and bacterial Nramp homologs.²³  PM expression of Nramp1HA has permitted the analysis of its substrate selectivity compared to that of Nramp2HA. Our experiments have shown that like other Nramp family transporters,^13,44-46  Nramp1HA can transport both Fe²⁺ and Mn²⁺ with apparent affinity in the low micromolar range. These values are in the physiologically relevant concentration range and are consistent with the known concentration of chelatable Fe²⁺ in mammalian cells (0.2-1 μM),⁴⁷  as well as cellular and tissue concentrations of Mn²⁺ (< 4 μM).⁴⁸  In addition, Nramp1HA appears to have a preference for Mn²⁺ ions when compared to Fe²⁺, suggesting that the former may be a preferred substrate at the phagosomal membrane. More detailed transport studies in membrane vesicles will be required to fully characterize the ion selectivity of Nramp1 and Nramp2.

Despite this potential difference in substrate preference, our results indicate that Nramp1 and Nramp2 are functionally equivalent multispecific divalent-metal transporters. The major physiologic difference between the 2 proteins appears to be at the level of the cell type–specific expression and subcellular site of transport. Nramp2 is expressed in recycling endosomes, where it transports transferrin-delivered iron from the acidified endosomal lumen into the cytoplasm down a proton gradient.^17-21  This endosomal Fe²⁺ efflux is impaired in reticulocytes from microcytic anemia mk mice, which bear a loss-of-function mutation at Nramp2.^17,22  It is very likely that Nramp1 functions in an analogous fashion to remove divalent metals from the phagosome. Indeed, microfluorescence imaging studies of Nramp1-positive phagosomes containing zymosan particles labeled with the metal-sensitive fluorophor Fura-FF6 have demonstrated pH-dependent metal efflux by Nramp1 at the phagosomal membrane within live primary macrophage.²⁴  For both Nramp1 and Nramp2, luminal acidification is achieved by recruitment of the vacuolar H⁺/ATPase, which provides the proton gradient necessary for Nramp protein function.^11,24,49  Altogether, these results strongly support a model in which Nramp1 transports Fe²⁺, Mn²⁺, and likely other divalent metals from the lumen of acidified phagosomes and into the cytoplasm.

The mechanism by which Nramp1-mediated depletion of divalent metals from the phagosomal space affects the survival and replication of unrelated intracellular parasites is not fully understood. Several lines of evidence indicate that an adequate supply of divalent metals such as Fe²⁺, Mn²⁺, and Mg²⁺ is critical for successful intracellular parasitism.^50,51  First, bacteria such as Salmonella express, under different conditions, a surprising number of diverse high- or low-affinity adenosine triphosphate (ATP)–dependent or proton-coupled iron (Fe²⁺, Fe³⁺) and/or manganese transporters such as fepBCDG, sitAD, FeoABC, CorAD,^52-56  and the Nramp homolog MntH.^45,57,58  Second, mutation of several of these transporter genes abrogate virulence and impair intracellular replication in vivo.^53,59-61  Third, intracellular replication of Salmonella within permissive Nramp1-negative RAW264.7 macrophage was partly abrogated by incubation with the metal chelator dipiridyl.⁵⁹  Interestingly, recent studies using Salmonella-bearing gene-specific reporter constructs have established that phagosomal divalent-metal depletion by Nramp1 creates a stressful environment, which is sensed by bacteria within macrophage. The intracellular bacteria responded to the presence of Nramp1 by the transcriptional induction of a number of “virulence” genes that map within Salmonella pathogenicity island 2 (SPI2) including ssrA and sseJ.⁶²  Similarly, infection of human macrophage by M tuberculosis results in induction of several mycobacterial genes required for siderophore-mediated iron uptake.⁶³  More specifically, preliminary comparative transcriptional profiling studies of M bovis–infected RAW264.7 macrophages (Nramp1-negative) and RAW264.7-Nramp1 transfectants suggest that Nramp1 has a direct effect on the level of transcriptional induction of the mycobacterial MbtB gene involved in iron acquisition (J.R.F. and P.G., unpublished data, 2002). Together, these results suggest that Nramp1 can act as an important antagonist of bacterial metal acquisition systems in the microenvironment of the phagosome.

We acknowledge Dr Francois Canonne-Hergaux and Dr Samantha Gruenheid for the preparation of anti-Nramp1NT and Nramp2NT polyclonal antibodies, and Dr Virginie Picard for the kind gift of cell lines and cDNA. We thank Steven Lam-Yuk-Tseung for advice with the fluorescent quenching assays.