ABSTRACT
Innate resistance to mycobacterial growth is mediated by a gene,Nramp1. We have previously reported that Nramp1mRNA from macrophages of Mycobacterium bovis BCG-resistant (Bcgr ) mice is more stable thanNramp1 mRNA from macrophages of BCG-susceptible (Bcgs ) mice. Based on these observations and on reports that show that the closely related Nramp2 gene is a metal ion transporter, we evaluated the effect of iron on the growth ofMycobacterium avium within macrophages as well as on the stability of Nramp1 mRNA. The addition of iron to macrophages from Bcgs mice resulted in a stimulation of mycobacterial growth. In contrast, iron increased the capacity of macrophages from Bcgr mice to control the growth of M. avium. When we treated recombinant gamma interferon (IFN-γ)-activated macrophages with iron, we found that iron abrogated the growth inhibitory effect of IFN-γ-activated macrophages from Bcgs mice but that it did not affect the capacity of macrophages from Bcgr mice to control microbial growth. A more detailed examination of the effect of iron on microbial growth showed that the addition of small quantities of iron to resident macrophages fromBcgr mice stimulated antimicrobial activity within a very narrow dose range. The effect of iron on the growth inhibitory activity of macrophages from Bcgr mice was abrogated by the addition of catalase or mannitol to the culture medium. These results are consistent with an Fe(II)-mediated stimulation of the Fenton/Haber-Weiss reaction and hydroxyl radical-mediated inhibition of mycobacterial growth.
The Bcg gene confers resistance to mycobacterial growth during the early nonimmune phase of infection (7, 19, 22, 23). Typing of recombinant inbred mouse strains for resistance and susceptibility to Mycobacterium bovis (strain BCG) as well as Mycobacterium avium and other mycobacterial species, combined with linkage analyses and dissection of a 30-centimorgan segment on murine chromosome 1, led to the cloning of the cDNA for the Bcg gene, designatedNramp1 (natural resistance-associated macrophage protein) (8, 33, 46). Sequence analysis of Nramp1 revealed a 1.4-kb open reading frame encoding a 484-amino-acid transmembrane protein with structural features common to eukaryotic transporters (17, 26, 43, 45). Susceptibility to mycobacterial growth in macrophages is associated with a glycine to aspartic acid substitution at position 169 within the fourth transmembrane domain ofNramp1 (5, 6, 21). However, the mechanism ofNramp1-mediated resistance to mycobacterial infection and the functional defect in Nramp1 mutants are still unknown.
Previous studies in our laboratory have shown that a variety of mRNA species, including Nramp1, from macrophages of BCG-susceptible (Bcgs ) mice are less stable than mRNAs from macrophages of BCG-resistant (Bcgr ) mice (11, 12). The reduced stability of mRNA from macrophages of mice expressing theNramp1 Asp169-susceptible allele was also observed in mRNA species induced by recombinant gamma interferon (rIFN-γ). mRNA stability is controlled by differences in sequences that facilitate protein binding (29, 40, 42), and this phenomenon promotes protection of mRNA from nucleases. Furthermore, protein binding to some mRNAs is controlled by Fe levels (29, 31,42). For example, the transferrin receptor mRNA is stabilized by the binding of a cytoplasmic aconitase (iron response element binding protein) to stem loop structures in the 3′ untranslated region of the transferrin receptor mRNA when intracellular iron is low. An increase in intracellular iron results in the binding of iron to the aconitase, releasing it from the mRNA, and the subsequent degradation of transferrin receptor mRNA. Additionally, iron serves as an important trace element for the generation of reactive intermediates of oxygen and nitrogen (15, 30, 31, 48, 50) and is an important nutrient for the growth of mycobacteria (20, 51).
Since iron affects mRNA stability and mycobacterial growth, we evaluated the effect of iron on the capacity of macrophages to inhibit the growth of M. avium and on the stability ofNramp1 mRNA. We found that iron stimulated the capacity of macrophages from mice expressing theNramp1 Gly169-resistant allele to limit the growth of M. avium. In contrast, iron stimulated mycobacterial growth in macrophages from mice expressingNramp1 Asp169. The antimycobacterial activity of IFN-γ-activated macrophages from mice expressingNramp1 Asp169 was abrogated by iron, but iron did not affect the capacity of IFN-γ-activated macrophages from mice expressing Nramp1 Gly169 to control the growth of the mycobacteria. The effect of Fe on the growth inhibitory effects of macrophages from these Bcgr mice occurred over a very narrow dose range and was abrogated by the addition of hydroxyl radical scavengers. These data are consistent with a model in whichNramp1 acts to decrease cellular iron levels by transporting iron into phagosomes, where it serves as a catalyst for the Fenton/Haber-Weiss reaction to generate hydroxyl radicals and at the same time stabilizes mRNA.
MATERIALS AND METHODS
Mice.Male BALB/c Bcgs (Nramp1 Asp169) mice were obtained from the Charles River Laboratory at 6 weeks of age. The animals were housed in groups of five in microisolation cages and given food and water ad libitum. Male BALB/c congenic C.D2Idh b-Ity r-Pep-3bmice, which are Bcgr (Nramp1 Gly169), were initially provided by Michael Potter (NCI) and bred in our facilities (11).
Reagents. M. avium (ATCC 35713) was obtained from the American Type Culture Collection (Manassas, Va.) and cultured in our laboratory as previously described (11). rIFN-γ was obtained from Gibco BRL (Grand Island, N.Y.). Actinomycin D, desferrioxamine, ferric ammonium sulfate, NG-monomethyl-l-arginine (MMLA), superoxide dismutase (SOD), mannitol, catalase, and dimethyl sulfoxide (DMSO) were purchased from Sigma (St. Louis, Mo.). The cDNA probe forNramp1 was produced by reverse transcriptase PCR as previously described by us (12). The β-actin cDNA probe was obtained by screening a macrophage cDNA library with actin-specific oligonucleotides. [32P]dCTP (specific activity, 3,000 Ci/mmol) was purchased from Amersham (Arlington Heights, Ill.).55Fe-labeled ferric chloride (specific activity, 17 mCi/mg) was obtained from NEN (Boston, Mass.).
Isolation of macrophages.Resident peritoneal macrophages or splenic macrophages from both Bcgr andBcgs mice were enriched by overnight adherence to plastic culture dishes (Falcon, Franklin Lakes, N.J.) by using Iscove’s modified Dulbecco’s medium (IMDM) (Gibco/BRL, Gaithersburg, Md.) supplemented with 20% defined fetal bovine serum (FBS) (HyClone, Logan, Utah) containing less than 0.03 endotoxin unit (EU) of endotoxin/ml. Nonadherent cells were removed by gentle washing with Hanks balanced salt solution (Gibco/BRL). The splenic macrophages were removed by using a cell scraper as previously described by us (11) and cultured again in 35-mm-diameter culture dishes, at a concentration of 5 × 106 macrophages per dish, in IMDM containing 20% FBS. Following the second incubation for 16 h at 37°C, the purified splenic macrophages (as determined by both differential and nonspecific esterase staining) were then washed with Hanks balanced salt solution, and the medium was replenished with IMDM containing 100 U of rIFN-γ/ml and subsequently treated as described. Macrophage cultures were incubated at 37°C in 5% CO2 for the times indicated for each experiment. Peritoneal macrophages were obtained by lavage and purified by culturing directly in 96-well plates.
Macrophage-mediated mycobacterial growth inhibition.The growth inhibitory capacity of the macrophage populations was assessed as previously described by us (11). Briefly, macrophages were added to 96-well microtiter plates so that a concentration of 2 × 105 macrophages per well was attained. The purified macrophages (>90% macrophages as determined by nonspecific esterase staining) were then infected with 8 × 105CFU of M. avium suspended in IMDM without serum and antibiotics and incubated overnight to allow for phagocytosis. The cultures were then washed to remove unphagocytized bacteria and incubated for 5 days to allow the intracellular growth of the bacteria. The macrophages were then lysed, and bacteria were pulsed overnight by incubating in media containing a mixture (1:1) of 7H9 (Difco) and IMDM with [3H]uracil (5 μCi/ml) (Amersham, Chicago, Ill.) (specific activity, 40 to 60 Ci/mmol) and 0.2% saponin. The bacteria were harvested onto glass fiber filter strips with a PHD cell harvester (Cambridge Technology, Inc., Watertown, Mass.). Radioactivity incorporated by the bacteria was quantitated by liquid scintillation spectrometry.
To induce increased antimycobacterial activity of the macrophages, the cells were treated with 100 U of rIFN-γ/ml (Gibco) in IMDM for 24 h prior to infection with M. avium. Iron was added to the macrophage cultures following removal of unphagocytized bacteria. In other experiments, resident macrophages were treated with iron but were not treated with rIFN-γ.
Isolation of RNA and Northern blot analysis.Splenic macrophages from Bcgr andBcgs mice were stimulated with 100 U of rIFN-γ/ml in the presence or absence of 100 μM desferrioxamine or 50 μM ferric ammonium sulfate. After 20 h the macrophage cultures were treated with 20 μg of actinomycin D or medium over a 48-h course prior to extraction and isolation of total RNA. After treatment, the macrophage monolayers were washed with ice-cold phosphate-buffered saline (PBS) (Gibco/BRL) and incubated on ice for 20 min. The PBS was removed and replaced with lysing buffer containing 8 M guanidine hydrochloride, 0.3 M sodium acetate, and 10% sarcosyl. Total RNA was extracted by the procedure as modified by Evans and Kamdar (16).
Equal amounts of RNA isolated from each treatment were size fractionated by gel electrophoresis (1.5% formaldehyde agarose), and RNAs were transferred to a Hybond-N+ membrane (Amersham) by capillary blotting. A separate lane containing a 0.24- to 9.5-kb RNA marker ladder (Gibco/BRL) was stained following electrophoresis with ethidium bromide and used to determine RNA size. Northern hybridization was carried out by using the protocol as previously described by us (12). Gel purified insert cDNAs were 32P labeled by random priming. Blots initially hybridized to Nramp1 cDNA were stripped with 0.1% sodium dodecyl sulfate and rehybridized to a32P-labeled β-actin cDNA probe. Autoradiographs were quantified by using the UVP ImageStore 5000 gel documentation program (San Gabriel, Calif.) and NIH Image version 1.58.
Iron transport into M. avium-containing phagosomes.In order to determine if iron was differentially taken up by phagosomes from macrophages expressing wild-type or mutantNramp1, we used the RAW264.7 macrophage cell line that had been stably transfected with Nramp1 Gly169(resistant) or Nramp1 Asp169 (susceptible) alleles. The cell lines were kindly provided by Jenefer Blackwell (5). The cells were labeled with 55Fe by first growing the cells for 24 h in IMDM without FBS and then growing them in complete medium supplemented with 5 μM55Fe-labeled iron citrate for 24 h to label endogenous Fe pools. The amounts of 55Fe taken up by the cells did not differ. Mycobacteria were added at a bacteria to macrophage ratio of 10/1 for 1 h at 37°C. The cells were then washed two times with PBS to remove unphagocytized bacteria and then incubated in complete IMDM at 37°C for an additional hour. The cells were removed from monolayers by using a cell scraper, and the phagosomes were isolated by a modification of the method described by Sturgill-Koszycki et al. (25, 44). Briefly, the cells were suspended in lysis buffer containing 250 mM sucrose, 20 mM HEPES (pH 7.0), 0.5 M EGTA, and 0.1% gelatin. Lysis of the cells was accomplished by repeated passage of the cells through a 21-gauge needle until at least 95% of the cells were lysed. The resulting solution was diluted threefold with PBS and centrifuged at 200 × g for 5 min to remove unbroken cells and nuclei. The supernatant was filtered through a 5-μm-pore-size Nucleopore filter (diameter, 25 mm; Corning, Corning, N.Y.), and the filter was rinsed with an equal volume of PBS. The filtrate was layered on top of a 50 to 12% sucrose step gradient and centrifuged at 800 × g for 45 min. The fraction at the 50 to 12% sucrose interface was removed and diluted threefold with PBS. This fraction was then layered onto a 15% Ficoll (low-molecular-weight) cushion, and the sample was centrifuged at 1,300 × g for 45 min. The M. avium-containing phagosomes were removed and resuspended in IMDM, and the amount of radioactivity incorporated into the phagosomes was determined. Equal amounts of phagosomes were isolated from theNramp1 Gly169- orNramp1 Asp169-transfected cell lines. The data are expressed as picomoles of iron per milligram of protein based on the amount of 55Fe added. This value includes the bacteria within the phagosome. However, in preliminary experiments we found that the amounts of bacteria taken up by the cell lines did not differ under the conditions we used to infect the cells.
RESULTS
Iron stimulates the antimycobacterial activity of resident macrophages.The results of experiments performed to determine the effect of iron on mycobacterial growth are presented in Fig.1. The addition of Fe to cultures of infected macrophages from Bcgr mice resulted in an inhibition of mycobacterial growth. The addition of 0.05 μM iron to cultures of macrophages from Bcgr mice resulted in the greatest level of growth inhibition. In contrast to the inhibitory effect of iron on mycobacterial growth in macrophages fromBcgr mice, increasing concentrations of iron resulted in increased growth of M. avium in macrophages fromBcgs mice. The stimulatory effect of iron was observed at concentrations of Fe of 5 μM or higher. Results shown in Fig. 2 indicate that iron stimulated the antimycobacterial activity of the macrophages fromBcgr mice within a very narrow dose range. Maximal inhibition was obtained at 0.05 μM Fe. The inhibitory capacity of the cells was comparable to that attained following treatment of the macrophages with 100 U of rIFN-γ in the absence of added Fe.
Effect of iron on the growth of M. avium in macrophages from Bcgr andBcgs mice. Resident peritoneal macrophages were infected with M. avium and then treated with ferric ammonium sulfate. After 5 days of culture, the macrophages were lysed and the bacteria were labeled with [3H]uracil. The radioactivity incorporated by the bacteria is an indication of the number of metabolically active bacteria remaining in the culture. The data are expressed as the percentages of inhibition of growth compared to growth of cultures of macrophages not treated with iron. The amount of radioactivity (mean ± standard deviation) incorporated byM. avium from Bcgr macrophages was 18,627 ± 2,899 cpm, while that incorporated by M. avium from macrophages from Bcgs mice was 31,095 ± 2,602 cpm. There were no differences noted in the numbers of bacteria phagocytized by the macrophages at the beginning of the experiment. The effect of iron on the growth of the bacteria in macrophages from Bcgs mice was significant as determined by analysis of variance (ANOVA). Similarly, the effect of iron on the inhibition of mycobacterial growth in macrophages fromBcgr mice was also significant as determined by ANOVA. Thus, Fe stimulated the growth of the mycobacteria in macrophages from Bcgs mice. In contrast, at low levels iron suppressed the growth of the mycobacteria in macrophages from Bcgr mice.
The addition of iron to cultures of macrophages infected with M. avium results in the inhibition of mycobacterial growth. For comparison with macrophages treated with rIFN-γ, resident peritoneal macrophages were treated with 100 U of rIFN-γ/ml overnight prior to infection with M. avium. Other cultures were infected with M. avium and treated with iron but not treated with rIFN-γ. After 5 days, the macrophages were lysed and the bacteria were labeled with [3H]uracil. The amount of radioactivity is an indication of the amount of metabolically active bacteria in the cultures. The data are expressed as the percentages of inhibition of bacterial growth compared to growth of cultures not treated with rIFN-γ or iron. The effect of rIFN-γ treatment was significant, as was the effect of iron, on mycobacterial growth as determined by analysis of variance.
Iron abrogates the antimycobacterial activity of IFN-γ-activated macrophages from Bcgs mice.In contrast to the stimulatory effect of Fe on the inhibition of mycobacterial growth of resident macrophages from Bcgr mice, iron did not affect the capacity of rIFN-γ-activated macrophages to control mycobacterial growth (Fig. 3). At the same time, however, Fe stimulated the growth of M. avium in rIFN-γ-activated macrophages from Bcgs mice.
Iron differentially affects the capacity of IFN-γ-activated macrophages from Bcgr orBcgs mice to inhibit the growth of mycobacteria. Macrophages were activated by treatment with 100 U of rIFN-γ/ml prior to infection with M. avium. The radioactivity incorporated by the bacteria was determined after 5 days in culture. The data are expressed as the percentages of inhibition of mycobacterial growth based on the growth of the mycobacteria in macrophages not treated with rIFN-γ. The mycobacteria isolated from macrophages ofBcgs mice that had not been treated with rIFN-γ incorporated 42,025 ± 2,561 cpm (mean ± standard deviation), while those isolated from macrophages ofBcgr mice incorporated 31,689 ± 2,882 cpm. The effect of iron on the growth of M. avium in macrophages from Bcgs mice was significant. Iron did not affect the capacity of the rIFN-γ-treated macrophages fromBcgr mice to inhibit the growth of the bacterium.
IFN-γ lowers intracellular iron levels.Preliminary reports by Hentze and Kuhn (30) and others (31) showed that IFN-γ treatment lowered intracellular iron levels in macrophages from Bcgr mice. The results presented in Table1 show that macrophages fromBcgr mice contain 344 pg of intracellular iron per μg of cell protein; this concentration is similar to that found in macrophages from susceptible mice (291 pg/μg). Treatment of the cells with rIFN-γ lowered the intracellular iron content of macrophages from Bcgr mice to 200 pg/μg, while the intracellular iron levels of macrophages fromBcgs mice remained unaffected.
The effect of IFN-γ on the iron content of macrophages from Bcgr and Bcgs mice a
Inhibitors of hydroxyl radical generation prevent increased antimicrobial activity induced by iron.Iron catalyzes the generation of hydroxyl radicals via the Fenton/Haber-Weiss reaction. The addition of the hydroxyl radical inhibitors catalase, mannitol, and DMSO resulted in an amelioration of the growth inhibitory capacity of macrophages from Bcgr mice that had been treated with iron (Fig. 4). The addition of neither SOD nor MMLA affected the growth inhibitory activity induced by iron.
Inhibitors of hydroxyl radical generation prevent the increased antimycobacterial activity induced by iron. Resident peritoneal macrophages from Bcgr mice were infected with M. avium and cultured with 0.05 μM Fe in the presence or absence of the following: SOD, 1 mg/ml; MMLA, 5 mM; mannitol, 50 mM; catalase, 2 mg/ml; and DMSO, 150 mM. Several concentrations were used for each of the inhibitors; the effect of the optimal concentration is shown. The effects of mannitol, catalase, and DMSO were significant at a P of <0.05 or lower. The effect of SOD or MMLA was not significant.
Manipulation of intracellular iron affects Nramp1 mRNA stability.We have already reported that Nramp1 mRNA expression is inducible with rIFN-γ (11) and thatNramp1 mRNA in macrophages from Bcgr mice is more stable than that in macrophages fromBcgs mice (12). As shown in Table 1, treatment with rIFN-γ also lowered cellular iron content in macrophages from Bcgr mice. We tested the possibility that the decrease in intracellular iron is causally related to the enhanced mRNA stability by artificially altering cellular iron levels by using desferrioxamine, an iron chelator, or by adding iron. The results, shown in Fig. 5, show that chelation of iron with desferrioxamine resulted in increased half-life of Nramp1 mRNA in macrophages fromBcgs mice. The half-life of Nramp1mRNA increased from 6 to 16 h in the presence of desferrioxamine, as contrasted to an increase of less than 2 h observed in macrophages from Bcgr mice. To determine if excess iron affected the expression of Nramp1 mRNA, rIFN-γ-stimulated macrophages from Bcgr andBcgs mice were incubated in the presence or absence of ferric ammonium sulfate, added to artificially raise iron levels. The results presented in Fig. 5 also show that incubation of macrophages from Bcgr mice with iron resulted in a decrease in Nramp1 mRNA half-life from 20 h to about 16 h, while that for Nramp1 from macrophages fromBcgs mice decreased from about 10 to 5 h. Thus, manipulation of intracellular iron levels had a greater effect on mRNA stability in macrophages from Bcgs mice.
Manipulation of intracellular iron affects mRNA stability. Splenic macrophages from Bcgr orBcgs mice were stimulated with 100 U of rIFN-γ/ml in the presence of 100 μM desferrioxamine (Desferox) or 50 μM ferric ammonium sulfate. After 20 h, the macrophage cultures were treated with 20 μg of actinomycin D (Act D) or medium for 2, 4, 8, 16, 32, or 48 h prior to extraction and isolation of total RNA. Control cultures (lane 1) were treated with rIFN-γ only. The corresponding decay curve of Nramp1 mRNA expression, representing the percentages of mRNA remaining following actinomycin D treatment, is derived from densitometric analysis of Northern blots. ■, Mutant (Bcgs ) plus desferrioxamine or Fe; □, mutant (Bcgs ) without desferrioxamine or Fe; ▴, wild-type (Bcgr ) plus desferrioxamine or Fe; ▵, wild-type (Bcgr ) without desferrioxamine or Fe.
Iron transport by phagosomes isolated from M. avium-infected RAW264.7 macrophage cell lines transfected withNramp1 Gly169 allele differs from that by phagosomes isolated from lines transfected with theNramp1 Asp169 allele.The results of this investigation suggest that iron is transported into phagosomes and stimulates hydroxyl radical formation. To test this model, we measured iron transport into phagolysosomes from M. avium-infected cells. The results, presented in Fig. 6, show that phagosomes isolated from cells transfected with theNramp1 Gly169 (resistance) allele and infected with M. avium imported more iron then did phagosomes isolated from infected cells transfected with theNramp1 Asp169 (susceptibility) allele. We did not observe any differences in the amounts of iron imported by the intact cells, nor did we observe that iron was exported differently by the cell lines or by the phagosomes (32a).
Iron transport by phagosomes isolated from the RAW264.7 macrophage cell lines, transfected withNramp1 Gly169 (resistance) orNramp1 Asp169 (susceptibility) alleles and infected with M. avium. Macrophage cell lines were labeled with 55Fe-labeled iron citrate prior to infection withM. avium. Phagosomes were isolated following lysis of the cells and differential centrifugation on a sucrose gradient. The radioactivity incorporated into the phagosomes is adjusted for total protein, which includes bacterial protein, and the data are expressed as picomoles of Fe per milligram of protein. The number of moles of Fe was calculated based on the specific activity of the isotope and does not take into account endogenous Fe levels.
DISCUSSION
The results of this investigation show that macrophages fromBcgr mice and Bcgs mice respond differently to iron. Iron stimulated the growth inhibitory capacity of macrophages from resistant mice. In contrast, iron stimulated mycobacterial growth in macrophages from susceptible mice. Our observation that treatment of macrophages fromBcgr mice with inhibitors of hydroxyl radical generation prevented the enhancing effect of iron on the growth inhibitory capacity of the macrophages strongly suggests that the presence of iron stimulates hydroxy radical formation via the Fe-catalyzed Fenton/Haber-Weiss reactions (28, 32, 35, 37). The generation of OH · from H2O2 (Fenton’s reagent) requires the availability of Fe2+. Our results (Fig. 6) suggest that theNramp1 gene product regulates this transport. The iron-catalyzed interaction of H2O2 and O2 ·− is the Haber-Weiss reaction, and the Fe3+ formed by Fenton’s reagent is reduced by O2 ·−. When levels of Fe2+ are sufficient it can react directly with H2O2 to form OH·. The reactions are inhibited by catalase and mannitol but are not affected by SOD and may further require the interaction with iodine (32). This possibility is currently being explored by us.
The lack of effect of MMLA, a nonmetabolizable form ofl-arginine and thus an inhibitor of nitric oxide (NO) generation, in preventing the Fe-mediated growth inhibition reinforces our previous observation that inhibition of arginine metabolism does not affect Nramp1-mediated control of mycobacterial growth (11). At the same time, as we have previously reported, the addition of MMLA to cultures of IFN-γ-activated macrophages inhibited their capacity to limit the growth of M. avium. Thus, the antimycobacterial activity mediated by rIFN-γ-activated macrophages and the growth inhibitory capacity mediated by Nramp1 are different.
The stimulatory effect of iron on the growth inhibitory capacity of macrophages from Bcgr mice occurred over a very narrow dose range. The inability of high concentrations of iron to stimulate the growth of mycobacteria in IFN-γ-activated macrophages may indicate that iron transport into the phagosome was limited and that only limited quantities of iron were available to stimulate the growth of the microorganism. Alternatively, it is possible that the increased production of NO by the rIFN-γ-stimulated cells and the formation of peroxynitrite ion contributed to the increased capacity to limit the growth of the bacterium, even in the presence of increasing concentrations of iron in vitro (15, 32, 48, 50, 51). Nitric oxide has been reported to cause an efflux of nonheme iron from infected cells (47, 49), and the addition of MMLA to these cultures reduces their growth inhibitory capacity (11).
We have reported, in preliminary studies, that the level of intracellular iron in macrophages from Bcgr mice decreases following treatment of the cells with rIFN-γ (52). The results presented here confirm our earlier report and that by Atkinson and Barton (4). The mechanisms that account for the differences in iron content remain unknown, especially since whole cells were used to make these measurements andNramp1 is expressed in phagosomal membranes and not in the plasma membrane (25). We do not think that iron flux is responsible for the changes in iron content because iron does not appear to be differentially taken up by the cells but is taken up only by phagosomes where Nramp1 is expressed (32a). However, the formation of NO· induces the loss of iron in part by an attack on iron-sulfur centers with associated inhibition of cell metabolism and growth (28, 32, 35, 37, 47,49). While the generation of these reactive nitrogen intermediates does not appear to play an important role inNramp1-mediated growth inhibition, it is possible that this mechanism may account for the loss of iron from the IFN-γ-treated macrophages from Bcgr mice, which have been reported to produce more NO than macrophages fromBcgs mice (5, 36).
What is the direction of Fe transport? Blackwell and Searle have suggested that iron is transported from the phagosome to the cytoplasm (9). This possibility would account for the scavenger function of macrophages, i.e., the removal of effete erythrocytes by phagocytosis and the recycling of iron. Removal of iron would limit its availability to microorganisms, thus accounting for the role ofNramp1 in limiting mycobacterial growth. This possibility points to a rather passive role for Nramp1. The results described by de Chastellier et al. (14) instead suggest a more active role for Nramp1. They showed, by ultrastructural analysis, that macrophages expressing the wild-type Nramp1allele contained more-damaged bacilli, especially at a very early stage of infection. This observation argues instead for a more active role for Nramp1 than just the limitation of iron. Our results suggest that limited quantities of iron transported into the phagosome catalyze the generation of antimicrobial hydroxyl radicals via the Fenton/Haber-Weiss reactions. The observation that limited quantities of iron result in increased antimicrobial activity is analogous to that reported by Alford et al. (2), who showed that differences in the iron content of macrophages correlated with differences in antilisterial activity. High iron levels were synonymous with increasedListeria growth, while lower, but sufficient, levels of iron resulted in increased listericidal activity.
Transferrin receptor-mediated iron uptake could also result in higher levels of iron within the phagosome. However, a reduction in the net uptake of iron occurs in infected host cells following the down regulation of transferrin receptors that occurs following infection (47) or upon interaction with macrophage-activating cytokines (13, 27, 39). This, together with the generation of NO, serves to lower the availability of iron. The transport of redox-active iron into phagosomes by Nramp1 may be required to supply sufficient catalyst for the Fenton/Haber-Weiss reactions when the availability of cellular iron is limited.
The observation that chelation of iron results in increased stability of Nramp1 mRNA supports the observations of Atkinson et al. (3), who showed that chelation of iron resulted in an increased expression of Nramp1 protein in cells that have been treated with rIFN-γ. Changes in mRNA stability can have profound effects on the quantity of translated proteins (34, 41). Thus, the differences we have observed in the stability of Nramp1 mRNA and in the stability of mRNAs of other IFN-γ-activated genes can significantly alter the levels of antimicrobial effector molecules and cytokines produced by the cells. The results of this investigation show that changes in cytoplasmic Fe alter the stability of Nramp1mRNA as well as other mRNAs (unpublished observations). The increased stability of mRNA from macrophages from Bcgr mice may, in part, be the result of the decreased levels of Fe associated with activation (13, 27, 39) as well as the transport of iron from the cytoplasm to the phagosome. This could account for the many pleiotropic effects of the Nramp1 gene, even though some of the differences that have been reported are the result of different genes whose induction requires different signal transduction pathways.
One important mechanism by which Fe regulates mRNA stability is by regulating the binding of an iron response protein to a consensus iron response element sequence in the 3′ untranslated regions of some mRNAs (e.g., transferrin receptor) (39). It has been proposed by Blackwell and Searle that Nramp1 mRNA stability may be similarly regulated, based on the observation that the consensus sequence for the iron response element is found in the 3′ untranslated region of rat Nramp2 mRNA (9). However, we have not been able to identify this sequence in the 3′ untranslated region of mouse Nramp1 mRNA (unpublished observations).
One mechanism that could account for iron-regulated mRNA stability is the observation that iron regulates protein kinase C (PKC) mRNA transcription (1). Our group (10) and others (38) have reported that the PKC activity of macrophages fromBcgr mice is greater than that of macrophages from Bcgs mice. The Nramp1 protein contains three putative PKC phosphorylation sites (6, 46). In preliminary studies we have found that inhibition of PKC inhibits the induction of stable mRNA by macrophages fromBcgr mice. Thus, PKC activity, iron, mRNA stability, and the function of Nramp1 are linked. The role of iron in regulating PKC and the effects of PKC inhibitors onNramp1-mediated iron transport and antimicrobial activity are currently being explored by us.
The recent identification of the murine gene controlling microcytic anemia as Nramp2 and the demonstration thatNramp2 is a metal ion transporter, which transports Fe2+, Zn2+, Mn2+, Co2+, Cd2+, Cu2+, Ni2+, and Pb2+, provides further support that the members of this gene family function as metal ion transporters (17, 18, 24,26). We propose that the role of Nramp1 is to transport iron from the cytoplasm into the maturing phagolysosome. This results in a lowering of biologically active cytoplasmic iron, which is sufficient to stabilize mRNA. The increased stability of mRNAs alone can account for the pleiotropic effects that have been attributed toNramp1. The influx of iron into the phagosome, under conditions in which cellular iron may be limited, serves as a catalyst for the Fenton/Haber-Weiss reaction, which results in the limitation of mycobacterial growth.
ACKNOWLEDGMENTS
This work is supported by grants HL59795, AI42901, and MH54966 from the National Institutes of Health.
Notes
Editor: S. H. E. Kaufmann
FOOTNOTES
- Received 8 October 1998.
- Accepted 1 December 1998.
- Copyright © 1999 American Society for Microbiology
