c-Myc Represses and Miz-1 Activates the Murine Natural Resistance-associated Protein 1 Promoter*

Iron is essential for growth, and impaired iron homoeostasis through a non-conserved mutation within murineNramp1, also termed Slc11a1, contributes to susceptibility to infection. Nramp1 depletes the macrophage cytosol of iron, with effects on iron-regulated gene expression and iron-dependent processes. Wu and colleagues (Wu, K.-J., Polack, A., and Dalla-Favera, R. (1999) Science 283, 676–679) showed converse control of iron regulatory protein expression (IRP2) and H-ferritin by c-Myc, suggesting a role for c-Myc in enhancing cytoplasmic iron levels for growth. We investigated if c-Myc also regulates Nramp1 expression. We show an inverse correlation with cell growth, and in co-transfection experiments c-Myc represses the Nramp1 promoter. Within theNramp1 promoter we identified six non-canonical E boxes, which are not important for c-Myc repression. By deletion analysis the repressor site maps to one or more initiator elements flanking the transcriptional initiation site. Co-transfections with the c-Myc interacting zinc finger protein (Miz-1) show that Miz-1 can overcome c-Myc repression of Nramp1, and, from a deletion construct lacking E box sites, Miz-1 activates the Nramp1promoter. These studies reinforce the link between c-Myc and iron regulation and provide further evidence that c-Myc negatively regulates genes that decrease the iron content of the cytosol. The results provide further support for a divalent cation antiporter function forNramp1.

was isolated as the positional gene candidate for Ity/Lsh/Bcg (4). In subsequent studies the full-length sequence of the encoded polypeptide was identified (5). Mouse strains resistant to infection encode Gly at codon 169 within Nramp1 whereas susceptible mice encode Asp (6). The G169D polymorphism is sufficient to explain the outcomes of model infections within inbred mouse strains at pre-T cell stages of infection, as confirmed by gene targeting and transgenesis experiments (7,8). Allele D169 is phenotypically null (7). When Nramp1 was cloned, the biochemical basis for its control over the proliferation of intracellular pathogens was not obvious, but the sequence suggested a transporter function (4). Subsequent studies showed the encoded polytopic integral membrane Nramp1 protein was expressed in a perinuclear location, on intracellular late endosomal/lysosomal membranes (9 -11). Nramp1 underwent rapid recruitment to the periphery of a pathogencontaining vesicle (10,11) and displayed a more peripheral location in response to treatment with interferon-␥ (11). These observations led to the suggestion that growth control of microbial pathogens could be achieved by the transport of some toxin into the lumen of the phagosome or by the sequestration of some essential nutrient. The identity of a candidate transport substrate was revealed from studies on a highly sequencerelated gene, Nramp2 (DMT1/DCT1/SLC11A2). Nramp2 was initially isolated as an orphan gene (12) but was re-isolated by functional cloning through a divalent cation/iron uptake assay (13). In addition, identical mutations, G185R, within Nramp2 in the mk mouse and the Belgrade rat are associated with impaired intestinal and erythroid cell iron uptake (14). Based on the striking sequence similarity between the two polypeptides (15), Nramp1 was also predicted to transport divalent cations or iron within murine macrophages, and infection susceptibility occurs through impaired divalent cation transport.
The many pleiotropic effects attributed to Nramp1 (16) should be explained by differential cation transport. Because Nramp1 protein is expressed within internal membranes, the differential partitioning of divalent cations between the cytosol and the lumen of the internal vesicle should contribute to the pleiotropic effects described (16). Inducible nitric-oxide synthase (iNos) is expressed at quantitatively higher levels in functional-Nramp1 macrophages (17)(18)(19)(20). Ferrous iron provides the link between iNos expression and Nramp1, because Dlaska and Weiss (21) revealed that iron loading inhibits iNos expression via the NF-IL6 transcription factor. Enhanced iNos expression in Nramp1 functional strains could be explained by depletion of iron availability within the cell cytosol that is sensed by transcription factors such as NF-IL6. However, there is controversy surrounding the direction of Nramp1-mediated cation transport. Some experimental evidence supports cation/ ferrous iron influx, into the vesicle lumen, as measured by direct transport of cations (22), and analysis of Nramp1 transport activity in Xenopus oocytes (23). Other data support transport out of the intracellular vesicle lumen (24,25). Divalent cations were proposed to accumulate or transit the cytosol, but these were not assayed directly (24). Based on the two transport hypotheses; Nramp1-mediated antimicrobial mechanisms have been suggested as redox-mediated killing (22,23) or iron starvation (24,25). Additionally, Nramp1 not only depletes iron from the cytoplasm but also is proposed to operate in a pathway ultimately leading to iron secretion from the macrophage (26,27). New approaches to delineate Nramp1 function are therefore required to resolve the controversy of transport and antimicrobial mechanisms.
We reported that transfectants of the Nramp1 D169 RAW264.7 cell, expressing the functional Nramp1 G169 allele, constitutively exhibit lower redox-active cytoplasmic iron levels relative to control transfectants or parental cells (28). In these cells iron-regulated processes are affected, including IRP2 activity and the iron-regulated protein kinase C isoform, protein kinase C␤1 (28). The changes are supportive with Nramp1 limiting cytoplasmic iron availability and transport into the vesicle lumen. Cytoplasmic iron depletion may be important for maintaining macrophage function during inflammation when high levels of reactive oxygen and nitrogen intermediates are generated and to minimize the effects of oxidative stress on cell viability (29). 2 Cytosolic iron depletion could explain the elevated expression of iNos in Nramp1 functional cells (17)(18)(19)(20).
Links between the c-Myc family of basic helix-loop-helix transcription factors and iron homoeostasis have recently been reported (30). A role for c-Myc was described in elevating cytoplasmic iron for cell proliferation, with stimulatory and inhibitory effects on the expression of genes that either increase or decrease, respectively, the extent of the LIP. Further data in support of a link between c-Myc and iron have been provided by the expression cloning, in a ferrous transport-defective yeast strain, of a novel complementing maize myc transcription factor (31). Kyriakou and colleagues (32) have described iron-dependent growth of peripheral blood mononuclear cells mediated via c-Myc. We therefore sought to investigate a role for c-Myc in the control of Nramp1. Our results indicate that c-Myc represses Nramp1 expression by acting at the Inr that spans the transcriptional initiation site (33). Furthermore, Miz-1 (34) positively regulates Nramp1 expression, and in co-transfection experiments Miz-1 overcomes the inhibitory effects of c-Myc. The data provide an independent method that lends further support to the hypothesis that Nramp1 reduces the iron pool within the cytoplasm and therefore functions as a proton divalent cation antiporter (23).

EXPERIMENTAL PROCEDURES
Preparation of Nramp1-expressing RAW264.7 Clones-Nramp1transfectant RAW264.7 cell clones have been prepared to study Nramp1 function, as described previously (28), with the full-length Nramp1 cDNA in the sense, lines R32 and R37, or antisense, line R21, orientations driven from the human ␤-actin promoter.
Anti-Nramp1 Antibodies-Antibodies immunoreactive to Nramp1 were prepared from recombinant GST-Nramp1 fusion proteins to the N-terminal domain as described previously (28). Western blotting was performed as before (28).
Analysis of Cell-proliferative Responses-Cell proliferation was determined on lines R21, R32, R37, parental cells, or primary bone marrow cells, as indicated, by staining 96-micro-well plates at time points with 0.5% crystal violet in 20% methanol and by determining the extent of either [ 3 H]thymidine or BrdUrd incorporation (Amersham Biosciences). Each sample was performed in replicates of at least three. Post crystal violet staining, bound dye was resolubilized in 100% methanol, and absorbances at 540 nm were determined. Samples were diluted as necessary. 1-2 ϫ 10 3 cells were plated out at day 1 for growth experiments. Statistical analysis was performed using Student's t test.
CBA Bone Marrow-derived Macrophages and N11 Microglial Cells-Bone marrow-derived macrophages were flushed out of the femurs of young adult males and matured in the presence of 20% L929 media as a source of macrophage colony stimulating factor. Cells were harvested at intervals for Nramp1 expression determination and for cell growth quantitation. N11 microglial cells that express mature 90-to 100-kDa Nramp1 protein were used for analysis of Nramp1 regulation following serum removal and replacement and were maintained as described previously (9).
Genomic Cloning and DNA Sequence Analysis-Genomic phage clones were isolated from a C57/BL6/CBAF1J FixII library (Stratagene) by screening with a PCR-derived probe spanning the 5Ј-end of the published murine Nramp1 gene sequence (33). The probe extended from Ϫ265 bp of the major transcriptional initiation site to a BamHI site 43-48 bp 3Ј of the last base of exon 2. From the screen two clones were isolated, 1.4 and 2.3, that were similar by restriction analysis. A SalI restriction fragment of ϳ9 kb was subcloned into pBluescript and was analyzed by sequencing (Oswel, Southampton). The sequence of the genomic segment used in these studies has been deposited in the EMBL data base (accession number AJ458183). The sequence is also available from the mouse genome data base (www.ensembl.org/ Mus_musculus/). Nramp1 5Ј-flanking sequence was analyzed for transcription factor binding sites (35). 3 A restriction fragment used for promoter studies was from an XbaI site at Ϫ1555 bp, relative to the major transcriptional initiation site (33), to a synthetic BamHI site introduced immediately downstream of exon 1. This modification also converts the natural Nramp1 ATG translational initiation codon to TTG. The 1655-bp fragment was cloned via XbaI and BamHI into the CAT reporter plasmid pBLCAT3, called pHB4, and used in transactivation experiments (see below). Other constructs in pBLCAT3 were prepared from more 3Ј HindIII, Ϫ868 bp (pHB6), or SphI Ϫ71 bp (pHB8) restriction sites, numbered relative to the transcriptional initiation site. The HindIII 5Ј truncation removes four of the six identified E-box sites, and the SphI truncation removes all the E-box sites. Other constructs were prepared from synthetic oligonucleotides (Oswel, Southampton); pHB 20, pHB21, and pHB22 have similar 3Ј termini at ϩ34 bp and 5Ј termini at Ϫ71 bp, Ϫ34 bp, and ϩ7 bp, respectively. An enhancer construct based on pCAT-enhancer (Promega) was also constructed for studies in Raw264.7 cells, based on pHB20/1, and termed pHB20E/21E. All promoter construct inserts were sequenced by Oswel.
c-Myc and Miz-1 Transactivation Studies-Transfection studies were performed in COS-1 cells and Raw 264.7 cells using LA (Invitrogen) or by electroporation. Briefly, 1-3 ϫ 10 6 cells/ml (COS-1), 1 ϫ 10 7 cells/ml Raw264.7) were transfected by electroporation (450 V, 500 microfarads (COS-1 cells); 350 V, 975 microfarads (Raw264.7 cells)) at room temperature in 0.5 ml of complete media using 10 g of Nramp1 promoter CAT constructions, 1 g for LA experiments, and the indicated amount of pEF-c-Myc (provided by Yongfeng Shang, Harvard Medical School) or pEF empty plasmid. LA transfections were performed according to the manufacturer's instructions. pCMV MIZ-1 expression plasmid was provided by Frank Hä nel, Hans Knoll-Institut for Naturstoff-Forschung, Heidelberg. Plasmid DNA used in transfections were prepared using Machery-Nagel Maxi Prep kits, and the total quantity of DNA in any transfection was normalized to 20, 2, or 3 g for LA transfections, with non-recombinant expression plasmid, or as indicated. Electroporated cells were immediately placed back in culture and harvested 48 h later. Cells were washed and lysed by multiple cycles of freeze thaw. Protein concentrations of soluble fractions were determined (Pierce), and 50 g of soluble proteins (20 g of LA), or as indicated, used for chloramphenicol acetyltransferase (CAT) activity determination by standard methods. CAT activities were quantitated using a PhosphorImager (Amersham Biosciences). CAT activities are described as percentage conversion of substrate to product/substrate plus product for 50 or 20 g of soluble protein except where indicated. The human ␤-actin promoter CAT construct was used as a transfection control between experiments, and a range of protein amounts were assayed (not shown). Experiments on the repression of Nramp1 by c-Myc have been repeated on more than 20 independent occasions with various vectors and plasmid DNA preparations.
The transfection efficiency for Raw264.7 cells was considerably lower than COS-1 cells, and for analysis of reporter gene expression we used an enhancer vector (Promega) to increase reporter gene expression levels. Direct analysis of reporter gene expression in Raw264.7 cells has also been undertaken using a semi-quantitative approach of digital microscopy. pEGFP-N3 (CLONTECH) was modified to remove the endogenous CMV promoter by digestion with AseI and BglII, 5Ј of the promoter and within the polylinker, respectively. Ends were repaired, DNA molecules were purified/ligated, and clones were recovered (p⌬CMV-EGFP-N3). Nramp1 promoter fragments, as described above for pHB4, were cloned into this plasmid, called pHB15, and transfected, as above, and cells were plated onto coverslips in single wells of eightwell plates (Nunc). After 48 h, cells were analyzed for fluorescence. Ten randomly selected fields at ϫ40 magnification (Zeiss digital microscope) were collected and analyzed for fluorescence above background using Metamorph version 4.6 imaging software (Universal Imaging Corp.). Results are presented as the mean fluorescence intensity over background for each field.

RESULTS
Decreased Proliferation of Nramp1-expressing RAW264.7 Cell Clones-Nramp1 G169 transfectants of the BALB/c-derived RAW264.7 (Nramp1 D169 ) macrophages show decreased prolif-eration compared with control cells not expressing the functional Nramp1 allele. A representative experiment is shown (Fig. 1a) conducted for Nramp1-expressing lines R32 and R37 and control, antisense line R21 and parental cells. The former express the mature 90-to 100-kDa Nramp1 polypeptide, whereas the latter two clonal cell lines do not (Fig. 1d). The enhanced cell proliferation for Nramp1 D169 cells is maintained over all cell densities analyzed. DNA synthesis data are normalized for cell number using a colorimetric assay (Fig. 1b). Greatest differences, ϳ3-fold, are detected when using 10,000 -20,000 cells per well. Addition of low molecular weight iron to cultures abolished the significant differences obtained between lines expressing and not expressing Nramp1 protein (Fig. 1c), whereas differences without iron were significant (p Ͻ 0.05), suggesting Nramp1 expression limits iron availability for cell proliferation.
Proliferative Response of CBA Bone Marrow Cells Precedes Nramp1 Induction-On successive days in culture, samples were taken from bone marrow cells, CBA strain (Nramp1 G169 ), for Nramp1 Western blotting (Fig. 2, a and b). Cell numbers were assessed in parallel cultures by a colorimetric assay, and DNA synthesis was quantitated by BrdUrd incorporation (Fig.  2c). Lanes R21 and R37 (Fig. 2a) are positive control extracts from Raw264.7 control and Nramp1 G169 -expressing transfec- tants (see Fig. 1c). Lanes 1-5 are from CBA bone marrow cells cultured from 1 to 5 days in the presence of macrophage growth factor (Fig. 2a). Confirmation that protein is loaded onto all tracks was provided by Amido Black staining an Immobilon membrane after immunodetection (Fig. 2b). Immunoreactive Nramp1 protein was not detectable at days 1 and 2, although full-length c-Myc immunoreactivity was present (not shown). The mature Nramp1 90-to 100-kDa protein was induced from day 3 and persisted until day 5, and the 45-kDa aglycosyl species could be detected on day 5 (Fig. 2a). The experiment shown is a representative of four such experiments, all of which demonstrate Nramp1 induction at day 3. Over the 5-day period in culture, the number of cells, assayed using the colorimetric assay, increased 6-fold (Fig. 2c). DNA synthesis peaked at day 2, and a temporal separation was apparent between DNA synthesis and the onset of the appearance of Nramp1 immunoreactivity (Fig. 2a). After the peak in DNA synthesis, BrdUrd incorporation levels remained higher than at day 1. Further evidence for a link between Nramp1 expression and cell growth was provided in microglial N11 cells, these cells endogenously expressed mature 90-to 100-kDa Nramp1. Following 48 h of serum starvation levels of the Nramp1 protein were low, and addition of serum (10%) to serum-starved cultures revealed a further decrease of the 90-to 100-kDa Nramp1 before an increase at 24 and 32 h (Fig. 3a). Amido Black staining of the membrane revealed equivalent protein loading (Fig. 3b). In contrast, Lamp1 expression is increased after serum addition at 8 h (not shown). To study the basis for this regulation the flanking region of the Nramp1 gene was isolated and analyzed to assess if there is a mechanistic link between cell proliferation and/or c-Myc and Nramp1 expression.
Isolation and Analysis of Nramp1 5Ј-Flanking Sequence-A restriction map of an isolated clone is shown together with locations of exons identified by comparison of cDNA (accession number X75355) and genomic sequences (Fig. 4a). Also aligned with the linear map of the genomic clone are the constructs  (33). Fragments of the Nramp1 promoter used in this study correspond to XbaI, HindIII, and SphI restriction endonuclease cleavage sites, corresponding to constructs pHB4, pHB6, and pHB8 and constructs prepared from synthetic oligonucleotides, pHB20, pHB21, and pHB22. The 3Ј-ends of constructs pHB4/6/8 correspond to a synthetic BamHI site located in place of the ATG translational start codon at ϩ99 bp and constructs pHB20/21/22 to ϩ34 bp. In expanded format is shown the location of candidate elements within the construct pHB4 as indicated. b, sequence of murine 5Ј flanking sequence, including exon one (underlined) to the 3Ј splice site. Sequence extends up to the XbaI site of construct pHB4. Restriction sites used in the preparation of the constructs are indicated in boldface, as are the six non-canonical E-boxes (MycMax), the two initiator elements (Inr), and the Sp1 site in pHB20. A GT repeat element is shown in italics and underlined. The first two amino acids of Nramp1 are shown in single-letter code. The sequence of the promoter fragment used in this study has been deposited with the EMBL data base (accession number AJ458183). prepared for this study. These are from 5Ј restriction sites XbaI, HindIII, and SphI to a synthetic 3Ј BamHI site, which removes the translational initiation site, and are labeled pHB4, pHB6, and pHB8, respectively, and three constructs were prepared from synthetic oligonucleotides, pHB20, pHB21, and pHB22. A preliminary analysis of the murine Nramp1 regulatory region has been reported (33), and comparative analyses with human and chicken undertaken (37). These revealed a series of conserved transcription factor binding sites for macrophage expression (PU.1) and induction by inflammatory mediators (␥-IRE, ISRE, IBP-1 and NF-IL6). However, the previously described murine sequence only extended to Ϫ265 bp (33). In the sequence presented here (Fig. 4b), we report multiple non-canonical E-boxes or Myc-Max binding sites of consensus CANNTG, shown schematically (Fig. 4a). Sites are numbered 1 to 6 from 5Ј to 3Ј at locations relative to the major transcriptional initiation site (33): #1, Ϫ1444 bp (CACATG); #2, Ϫ1212 bp (CACATG); #3, Ϫ1199 bp (CACATG); #4, Ϫ951 bp (CACTTG); #5, Ϫ588 bp (CATGTG); #6, Ϫ130 bp (CAACTG). All these sites share similarity with the high affinity E-box site (CACGTG), and non-canonical E-box hexamers or heptamers (CATGTG, CATGCG, CACGCG, CACGAG, and CAACGTG) can bind c-Myc-Max heterodimers in vitro but with lower affinity (38). Nramp1 site #5 is identical to one of these lower affinity E-box elements and binds to c-Myc rather than the upstream factor. Putative c-Myc-Max binding site #5 is conserved in human (CAAGTG) incorporating a single T to A change at position 3 of the hexameric binding site (39). E-box sites #5 and #6 are retained in the deletion construct pHB6 but not in construct pHB8-22.
Also conserved within the 5Ј flanking sequence is a simple tandem repeat element Ϫ376 to Ϫ307 bp (Fig. 3, a and b). In human a similarly located element is polymorphic and specific alleles are linked reciprocally with susceptibility, or resistance to, rheumatoid arthritis, Crohn's disease, or infectious diseases and is proposed to form Z-DNA (39 -41). The human NRAMP1 promoter allele 3 is of the form T(GT 5 )AC(GT 5 )AC(GT 9 )G and aligns (not shown) with the mouse promoter repeat of form (T/G) 27 C(G/T) 3 . Preparation of this region of DNA from other mouse strains by PCR revealed polymorphism. The DNA sequence (Fig. 4b) of the isolated clone is derived from the C57BL/6 strain rather than CBA on the identification of a polymorphism described in these strains (6) further 3Ј. Alleles from both Nramp1 G169 strains studied (CBA, C3H) have the same form TGCG(TG) 22 C(GT) 3 , whereas those from Nramp1 D169 strains (Balb/c, C57BL/6, DBA1) are identical to the sequence shown in the genomic clone (TG) 27 C(GT) 3 . The Nramp1 promoter also contains two copies of an initiator element (Inr) of consensus YYAN(T/A)YY, as described previously (33). Shown (Table I) is an alignment of the Inr with other Inrs, adapted from Ref. 42. Nramp1 Inr site 1 shows a perfect match to the consensus, whereas site 2 displays a pyrimidine to purine substitution at position 7.
Analysis of Nramp1 Promoter Function in COS-1 Cells: Repression by c-Myc-To examine if the murine Nramp1 promoter is regulated by c-Myc we used a transient transfection approach in COS-1. Constructs pHB4, pHB6, containing 6 and 2 c-Myc-Max binding sites, respectively, and pHB8, pHB20, and pHB21 that lack c-Myc-Max binding sites all demonstrated loss in promoter activity following co-transfection with the c-Myc expression plasmid (2 g). Also shown is the -fold repression by c-Myc 3.1 Ϯ 1.3 (Fig. 5a). These results indicate that the c-Myc-mediated transcriptional repression is not due to the E-box sites. pHB22 was inactive with and without co-transfected c-Myc, as expected. Construct pHB21 contains only the two Inr elements, described above, and a single Sp1 site on the antisense strand suggesting the c-Myc inhibition is mediated via the Inr (42). In these experiments we also observed a 4-to 5-fold increase in promoter activity on deletion from Ϫ868 bp (pHB6) to Ϫ71 bp (pHB8) (p ϭ 0.0004 and p ϭ 0.001 for pooling activities pHB4/6 and pHB8 -21, respectively), indicating the region contains some suppressor element. There was no further increase in promoter activity on deletion from Ϫ71 to Ϫ34 bp (pHB21).
A dose-dependent inhibition of Nramp1 promoter activity in   (Fig. 5b), and the inhibition by c-Myc was statistically significant (p ϭ 0.001, comparing 0 and 2 g of added c-Myc). In some experiments we observed a biphasic response to c-Myc, which happened on a regular basis, and higher c-Myc levels were again inhibitory. The basis for this biphasic response is not known but has been described previously on c-Myc titration experiments with the AdML and cyclin D promoters (34). In our hands it does not appear to occur at a reproducible level of co-transfected c-Myc, and in this experiment occurred at 0.2 g of added c-Myc, whereas 0.5 g of c-Myc was again inhibitory.

COS-1 cells is shown
Analysis of Nramp1 Promoter Activity in Raw264.7 Cells-Nramp1 is a macrophage-specific gene in mouse and therefore attempts were made to analyze the promoter activity in cell lines of this lineage. However, as a consequence of low transfection efficiency it was not feasible to undertake this study using either CAT or luciferase reporters in our hands with pHB4. We therefore evaluated the use of quantitative digital microscopy for eGFP reporter gene expression in single cells. To evaluate the stability of the eGFP fluorophore for the study proposed here, Raw264.7 cells were transiently transfected with eGFP plasmid and analyzed 48 h later. A representative field was identified, captured, and then constantly exposed to the UV light for a period of up to 10 min. Images were captured at time intervals during this period of persistent UV exposure. All images were analyzed for fluorescence intensity using the Metamorph software package. Results indicate (Fig. 6a) little change in the fluorescence intensity over a period of 5 min. However, from 5 to 10 min a decline of 1.5 log units of fluorescence intensity was determined. This experiment indicates that, if images are collected within the 5-min window, it is possible to use this method for quantitative analysis. In another experiment Raw264.7 cells were transfected from 1 to 20 g of eGFP-N3 wild-type plasmid (Fig.  6b). Ten fields for each plasmid dose were evaluated, and data show a dose-dependent increase in fluorescence intensity with increasing plasmid dose. This method was employed to evaluate Nramp1 promoter activity in Raw2264.7 cells following transient transfection (Fig. 6c). As in COS-1 cells, increasing c-Myc plasmid in the transfection caused a dosedependent inhibition of Nramp1 promoter activity (compare data with that of Fig. 5b).
Using a construct containing an SV40 enhancer element, it was possible to measure Nramp1-driven reporter gene activity in transiently transfected Raw264.7 cells using the CAT reporter gene. In this experiment a significant dose-dependent inhibition of the reporter gene activity from pHB20E was observed that reproduced the inhibition curve obtained by digital microscopy in Raw264.7 cells and of c-Myc inhibition in COS-1 cells (Fig. 6d, compare with Figs. 6c and 5b). These data indicate that the effects we describe in COS-1 cells for c-Mycmediated inhibition of Nramp1 are also relevant to macrophage lineage cells. The differences between 0 g of cotransfected c-Myc and 2 g are significant (p ϭ 0.0001). The differences in the dose-response curves between Figs. 6, c and d, were due to the transfection method employed (Fig. 6c,  electroporation, Fig. 6d, LA) and the amount of Nramp1 promoter construct used. In both cases the ratio of promoter to co-transfected c-Myc is the same, 5 g in Fig. 6c corresponds to 0.5 g in Fig. 6d. Using LA in Raw264.7 cells similar effects to those described in COS-1 cells were observed on deleting Ϫ868 bp to Ϫ71 bp (data not shown) as described (Fig. 5a).
Modulation of Nramp1 Promoter Activity by Sequence Variation within the GT Repeat-Previous work has identified func-FIG. 6. Analysis of Nramp1 promoter function in Raw264.7 macrophages. a, pEGF-N3 (CLONTECH) was transfected into Raw264.7 cells. A field was selected and an image was captured. Subsequent images were captured after continual exposure to the excitation wavelength. The graph shows the effect of sustained illumination on output of fluorescence intensity (FI) with time. b, dose-dependent increase in FI with amount of pEGFP-N3 plasmid. Raw264.7 cells were transfected with indicated amount of pEGFP-N3, and FI was determined. c, dose-dependent modulation of Nramp1 promoter by co-transfected c-Myc. FI recorded is plotted against the amount of c-Myc used in the transfection 0 -10 g, all DNA in transfections was normalized to 20 g of DNA. Student's t test showed significance compared with pHB15 (0 g of c-Myc) p Յ 0.008 for 2, 5, and 10 g of c-Myc. d, dose-dependent reduction in CAT reporter gene activity with increasing c-Myc. Raw264.7 cells were transfected using LA with pHB20E (1 g) and the indicated amount of c-Myc, total DNA in all transfections was 3 g. Student's t test compared with pHB20E alone; *, p ϭ 0.035; **, p ϭ 0.029. tional sequence polymorphisms within the human NRAMP1 promoter that correlate with infectious and autoimmune diseases susceptibility and resistance (39). The polymorphism is associated with a simple tandem, microsatellite-like repeat and is suggested to function as an enhancer element (39). As we show sequence variation within this region in mouse in G169 and D169 allelic mouse strains, we evaluated if promoter activity is influenced by this sequence variation. Surprisingly, we found that the promoter within a G169 allele strain, CBA, displayed less functional activity (2-to 3-fold, p ϭ 0.000075) than that found in a D169 strain, C57Bl/6, pHB4. However, both promoters were still susceptible to an equivalent (2-fold) inhibition by c-Myc co-transfection (not shown). Activities of both promoters were not significantly different with 2 g of co-transfected c-Myc.
Miz-1 Relieves the Inhibition of c-Myc and Transactivates the Nramp1 Promoter-One or more Inrs are located within the smallest construct tested that reveals c-Myc inhibition. Recent studies have shown that Miz-1 can bind to Inrs and function as a positive regulator of gene expression (34). c-Myc antagonizes the effects of Miz-1. Using a co-transfection approach in COS-1 cells (Fig. 7a) 1 g of c-Myc produced a 50% reduction in activity (**, p ϭ 0.0015), as before (Fig. 5b). Co-transfection of 1 g of Miz-1 suppressed the repression of c-Myc on Nramp1 promoter activity for construct pHB4, such that the promoter activity was not significantly different from the control (promoter alone) but significantly greater than c-Myc treated (*, p ϭ 0.036). Surprisingly, Miz-1 could not significantly en-hance the activity of Nramp1 (pHB4) on c-Myc untreated cells in this experiment. We performed a dose-response analysis of Miz-1 with promoter constructs pHB4/20/22 (Fig. 7b). As before in untreated cultures (Fig. 5a) pHB20 was more active than pHB4. Miz-1 co-transfection caused a 5-and 6-fold induction of pHB4 and pHB20, respectively, that achieved significance for pHB4 only at the highest does of Miz-1, and for all except the lowest dose of Miz-1 for pHB20 over untreated cultures. Therefore, pHB20 appears to be more sensitive to transactivation to low doses of Miz-1. In contrast no increase in pHB22 was observed.

DISCUSSION
In this report we present functional data on the murine Nramp1 promoter and show inhibition of Nramp1 expression during cell growth. We show that ectopic Nramp1 expression can reduce cell proliferation. Together these data are supportive for a link between Nramp1-dependent iron homoeostasis and transcription factors associated with cell growth. A recent paper by Wu and colleagues (30) showed c-Myc regulates the expression of genes that modulate the extent of iron within the labile iron pool (30), specifically that c-Myc enhances the LIP for cell proliferation. These data led us to investigate the murine Nramp1 promoter for regulation by c-Myc. In the extended murine Nramp1 promoter sequence presented, we identified six c-Myc-Max binding, E-box sites (35). 3 However, although we observed DNA protein complexes binding to site 5 (data not shown), these do not appear important for the inhibitory effects of c-Myc on Nramp1 expression, and a c-Myc site 5 mutant did not show any altered transcriptional responses (data not shown). However, deletion of c-Myc-Max site 6, and its flanking region, does correlate with an increase in promoter activity and increased responsiveness to Miz-1 transactivation, particularly at low Miz-1 doses. Peukert and colleagues (34) showed induction of the AdML and cyclin D promoters by Miz-1, and the cyclin D construct containing a number of E box sites was induced to a lesser extent, suggesting that the presence of E box sites within the promoter can interfere with the ability of Miz-1 to transactivate. This hypothesis is consistent with the increase in Nramp1 promoter activity we observed on deleting from Ϫ868 to Ϫ71 bp, including E box site #6, but we cannot rule out roles for other factor binding sites. However, the smallest construct we tested was still sensitive to c-Myc inhibition, indicating that E box site #6 is not of primary importance for c-Myc repression. This deletion construct contained two Inrs, a target for c-Myc repression. The Inr has been described as a core promoter element that can replace the TATA box in TATAless promoters (43). Gene repression for generating the transformed phenotype has been shown to be of equal importance as gene-inductive events (44), and some, but not all, c-Myc gene repression is mediated through the Inr.
We have not undertaken a more detailed analysis of the Ϫ868 and Ϫ71 bp region and the associated negative regulatory element, but it does harbor the polymorphic microsatellite element described here, and by others in the human NRAMP1 promoter (39). In studies on the human promoter, the polymorphic repeat was suggested to function as an enhancer element and alleles driving high expression protect against infectious but, conversely, promote susceptibility to autoimmune diseases such as rheumatoid arthritis (39,40). Recent studies have also indicated that promoter polymorphism contributes to susceptibility to Crohn's disease, including the identification of a novel allele (41). Although the results of this study do not contradict the work in human, we favor the model of a negative-acting element being located within the region of the promoter where the GT element is located. This is based on the enhancement of promoter activity we see when this region is deleted. We propose that the repressor, between Ϫ868 and Ϫ71 bp in the C57Bl/6 strain is attenuated. The proposed attenuation contributes to stronger promoter activity. We have no evidence for or against the repressor being attributed directly to the microsatellite itself, but suggest that the altered spatial (2.04 nm) and angular separation (216 o ) of factor binding sites on either side of the repeat element may be sufficient to change potential protein-protein interactions and consequently the activity of the putative repressor. Our current experiments are directed at identifying the nature of the potential repressor element by further deletion mutagenesis. Both promoter alleles are subject to the same level of repression by c-Myc, consistent with c-Myc operating downstream of the polymorphic site at the initiator, and promoter activities of the two allelic variants of c-Mycrepressed Nramp1 are not significantly different.
In human, NRAMP1 alleles with low promoter activity associate with protection against autoimmune disease, including rheumatoid arthritis. However, a major driving force for keeping the high activity promoter in the gene pool is to provide protection against infectious disease (16). The major genetic determinant that contributes to infectious disease susceptibility in mouse is the G169D polymorphism (6), and this polymorphism has not been described in human. We do not propose that the mouse promoter polymorphism identified will dominate the G169D mutation, because the D169 allele is null. However, potential promoter sequence variation within G169 strains could lead to differential iron homoeostasis and contrib-ute to a phenotype that may be a better model of human infectious, or autoimmune disease susceptibility, than the G169D polymorphism. The search is now on to identify the higher activity promoter variant on the background of the G169 allele or to construct the more active promoter allele in combination with the G169 allele in transgenic mice. Such studies will allow a phenotype of differential Nramp1 expression to be examined in a model organism.
As indicated in the introduction there is controversy regarding the direction of cation transport mediated by Nramp1 (22)(23)(24)(25). This is probably due to the experimental approaches used by different groups. However, the results of this present study provide another independent description for the direction of cation flux mediated by Nramp1. Our current data agree with our previous results (28) that Nramp1 depletes the cytosol of divalent cations, thus Nramp1 must operate as a proton divalent cation antiporter (23), in contrast to Nramp2 that encodes a proton divalent cation symporter (13). This finding is remarkable given the high degree of sequence identity within the hydrophobic cores of the two molecules (15). Our evidence is based on the observation that c-Myc antagonizes Nramp1 and on the data of Wu et al. (30), indicating that c-Myc enhances the cation pool within the cytoplasm for cell proliferation, including enzymes such as ribonucleotide reductase. If Nramp1 were to enhance levels of iron within the cytosol and operate as a symporter, as suggested (24,25), then we would anticipate that c-Myc would positively regulate Nramp1 expression and ectopic Nramp1 expression would increase rather than reduce cellular growth rates, as we described. These current data are in contrast to those of Jabado (24), who, using a fluorescent reporter linked to particulate yeast, zymosan, showed Nramp1dependent extrusion of vesicular divalent cations, and of the hypothesis reached by Gomes and Appelberg (25) that Nramp1 must deprive cations from intracellular pathogens. Preliminary experiments we have undertaken revealed that zymosan A question mark is drawn against the arrow from Miz-1 to labile iron pool (LIP) because no direct effect of Miz-1 on the LIP has been demonstrated. Nramp1 is negatively regulated by c-Myc, and Miz-1 relieves the inhibition of c-Myc and activates Nramp1 (this study), c-Myc increases the size of the LIP by modulating expression of IRP2 and H-ferritin (30), and Nramp1 negatively modulates DNA synthesis (this study) and decreases the labile iron pool (28). c-Myc has positive effects on growth in an iron-dependent manner and modulates iron transport in eukaryotic cells (30). Nramp1 also increases IRP2 (28) and suppresses ferritin expression (27).
itself has a major effect on the distribution of divalent cations within a macrophage and support our and others collective hypothesis of an iron secretory role for Nramp1 (26,27). An iron secretory role for Nramp1 could rationalize data on Nramp1 function, both influx into the vesicular lumen and deprivation from pathogens of essential iron for intracellular growth.
Our model (Fig. 8) is supportive of an interaction between c-Myc (cellular growth) and Nramp1 and iron. Expression of Nramp1 from a heterologous expression plasmid disrupts this normal regulatory control allowing effects of ectopic Nramp1 expression on growth to be established. We do not believe this arises from Nramp1 overexpression, because levels of protein in the cell lines and primary macrophages are comparable and dilute in parallel (not shown). These current data and those of others indicate that c-Myc must be able to sense levels of iron within a cell by some ill-defined mechanism. In experiments described by others on Miz-1, links with TGF␤ signaling pathways and co-activator proteins have been identified (45,46). This is of interest given that recent studies by Roig and colleagues (47) showed vitamin D is a potent agonist for human NRAMP1 induction and the well established links between vitamin D, c-Myc suppression, and TGF␤ (48,49). In current experiments we are actively pursuing the role of Miz-1 function for Nramp1 regulation based on the model of p15 ink4b (45,46). It will also be of interest to determine if Miz-1 has effects on iron levels other than through Nramp1 by controlling genes such as ferritin or Ireg1, which both will lead to reductions in the level of iron within the cytosol. In future experiments we plan to investigate if Miz-1 is a major regulatory factor for Nramp1 within the macrophage and to delineate how Miz-1 activity is regulated in this cell. Preliminary experiments have revealed a number of protein DNA complexes binding to a DNA probe corresponding to the two Inr sites, and the specificity for Miz-1 binding was shown by competition studies.
Previous studies in vivo have shown restricted expression of Nramp1, to specific macrophage populations undergoing erythrophagocytosis, and macrophages within an experimental brain lesion (26), but so far no mechanistic basis for this differential control within macrophage sub-populations has been put forward. There are data suggesting Miz-1 translocation to the nucleus involves microtubule depolymerization (50), and oxidant stress has been reported to promote microtubule disruption (51). A mechanism of microtubule disruption, leading to Miz-1 nuclear translocation, could lead to Nramp1 induction. That Nramp1 may be induced by microtubule dynamics could have a physiological basis, because a role for microtubules in phagosome movement has also been described (51), and some data suggest an interaction between Nramp1 and microtubules (36). Thus, the ingestion of particulates by macrophages could be one signal that leads to the induction of Nramp1 in specific macrophage populations. Future studies will be aimed at examining the effect Miz-1 on Nramp1 regulation by a range of stimuli.