In general, immunity to mycobacterial species depends on the activation of macrophage microbicidal activities through the IL-12/IL-23–IFN-γ axis. Genetic deficiencies in IL-12Rβ1, IFN-γR1, IL-12p40, STAT-1, IFN-γR2 or NEMO were traced back to a rare congenital syndrome, Mendelian susceptibility to mycobacterial diseases involving weakly virulent species [31] or to disseminated TB (IL-12Rβ1) [32]. Ingestion of Mtb by pulmonary macrophages and dendritic cells induces proinflammatory cytokines secretion, including interleukin (IL)-12, IL-1β, and Tumor necrosis factor (TNFα). IL-12 stimulates a T-cell helper 1 (T_H1) response that in turn promotes M1 macrophage microbicidal activities; in addition TNFα broadly modulates macrophage activities such as cytokine and chemokine secretions, microbial killing, programmed necrosis or apoptosis. Despite this antimicrobial potential anti-inflammatory cytokines including IL-10 and TGF-β can also be produced by Mtb-infected macrophages, translating an alternative (or M2) state of activation. The resulting downregulation of proinflammatory cytokines and T-cell proliferation and activation leads to a balanced response, which contains bacteria unless external factors perturb that equilibrium and reactivate infectious disease [30,33].

Strong evidence supports a critical role for genetic factors in susceptibility or resistance to TB but the mechanisms involved remain elusive. Host resistance to TB is a complex trait due to an intricate balance of host-Mtb interactions involving multiple genes and compounding factors (e.g., phenotype definition, study design, human and microbial population genetic heterogeneity and linkage disequilibria, socio-economical determinants, environmental factors such as nutrition, co-infections, epigenetics, ...). TB resistance has been associated with several gene polymorphisms but with low consistency between replicate studies, while most clinical association results are not yet validated functionally by molecular studies [34,35,36].

NRAMP1 may be a factor of critical importance for host defense to the development of TB because several NRAMP1/SLC11A1 genetic polymorphisms were consistently associated with TB resistance/susceptibility in many studies, including the disseminated form of pediatric TB [37,38,39]. SLC11A1 consistent association with host resistance to developing TB however has not established whether this link regards progression from latent infection to active TB disease or susceptibility to primo-infection [34,37,38,40].

Genetic variants under evolutionary selection often have accumulated to relatively high frequencies, which might represent an equilibrium between benefits procured for TB resistance and high risk of other diseases, such as adverse consequences of microbicidal activities [36]. Significant associations but moderate in strength were established with pulmonary TB in different populations, most consistently in Asian populations, also in African and western populations and regarding strains from distinct lineages of Mtb [37,38].

Polymorphisms at the 3' end of SLC11A1 (D543N, 1729 + 55del4) show stronger associations for western populations, and the frequency of the non coding deletion allele 1729 + 55del4 [41] correlates positively with ancient urbanization [42]. This suggested that longer time period of contact between humans and Mtb favoured selection of resistance alleles. As increased human density could favor aerosol transmission and the evolution of more virulent Mtb strains, such pathogen pressure could drive the increase in frequency of SLC11A1 resistant alleles by natural selection [35,42,43]. Although the functional basis for selecting such allele remains unknown it is worth noting that D543N, 1729 + 55del4 polymorphisms lie in a region of SLC11A1 that is predicted to contain regulatory elements (Section 5.2.1).

In few instances SLC11A1 genetic polymorphism was correlated functionally in macrophages. Hence, the polymorphism -274C/T is a silent nucleotide substitution in codon 66 (Phe) of SLC11A1 exon 3, which was identified as a major risk factor to acquire pediatric TB [39]. To study the possible impact of the SLC11A1 -274C allele on protein function, an assay measuring intracellular recruitment to the phagosome of the macrophage late endosomal/lysosomal marker mannose 6-phosphate receptor (M6PR) was employed [44]. Mouse macrophages expressing a functional Nramp1 control the intracellular fate of diverse pathogens such M. bovis BCG, Salmonella enterica Serovar Typhimurium and Leishmania donovani albeit in different ways, depending on the pathogen intracellular strategy [10].

Soon after invading macrophages Salmonella establishes a replicative niche by subverting the normal process of phagolysosomal maturation and excluding for instance, the M6PR from the enclosing vacuole membrane. Assaying M6PR recruitment to the phagosome in human macrophages provided readout of SLC11A1 activity suggesting a link between SLC11A1 -274C allele and decreased innate resistance to pediatric TB [45]. One possibility may be that this synonymous polymorphism impacts translational efficiency [46]; also, prediction of a regulatory element located 150 bp upstream of the polymorphism -274C/T lying in a domain of open chromatin, epigenetically marked in CD14⁺ monocytes (MNs) and monocyte-derived macrophages (MDMs), might suggest interference with gene regulation (Section 5.2.3).

Hence, several SLC11A1 polymorphisms associated with TB represent potential regulatory variants rather than non-synonymous SNPs altering protein activity. Among other effects, the impact of genetic variation on chromatin structure and the transcriptional machinery is of significance. The combinatorial nature of gene promoters and chromatin modifiers implies that non coding polymorphisms may be active only in a target cell (e.g., macrophage) or a given immune context (pro- or anti-inflammatory). In both human histiocytic U-937 cells and the monocytic cell line THP-1 SLC11A1 variant -237C stimulates promoter activation, while hypoxia-inducible factor 1 alpha (HIF-1α) regulates NRAMP1 expression levels by interacting with a microsatellite repeat under evolutionary selection pressure in the promoter region [47,48,49].

The SLC11A1 promoter region contains several polymorphisms influencing gene expression. The role of a (GT)_n microsatellite repeat consisting of 9 identified allelic variants was characterized in details regarding the two most frequent alleles. These alleles differ by one GT repeat, a variation that affects SLC11A1 promoter activity: allele 3 contains 9 repeats; it is the most frequent allele and drives higher gene expression levels in transient transfection assays using U-937 and THP-1 cells, while allele 2 with 10 repeats has higher Z-DNA-forming propensity but drives lower expression levels in transient transfection assays [47,48,49,50].

Z-DNA formation can facilitate transcriptional initiation and activation. This left-handed conformation of the double helix may be formed at TG repeats within promoters upon remodeling of a mononucleosome by the human SWI/SNF complex [51]. Interestingly in THP-1 cells, SLC11A1 allele 3 but not allele 2 bound the transcription factors ATF-3 and c-Jun [50]. SLC11A1 allele 3 transcriptional activation was decreased by co-transfecting ATF-3, an effect that was amplified in cells activated with bacterial lipopolysaccharide (LPS) suggesting that ATF-3 dimer may act as repressor [50] (Section 4.4).

Alleles other than allele 3 were significantly associated with an increased risk of TB, independent of disease type, with a meta-analysis mean odd ratio of 1.31 comparing cases to controls (95% confidence interval: 1.59–1.08) [37]. Impaired SLC11A1 expression may contribute to limit macrophage pro-inflammatory potential, and consequently, reduce formation of nitric oxide (NO), tumour necrosis factor-alpha and interleukin-6 and increasing interleukin-10 production [52]. It has previously been shown in mice that lack of Nramp1 limits macrophage lipocalin expression [53], production of NO [54,55] and intracellular signaling [56,57]; it also affects tissue Fe distribution and the expression of liver genes involved in Fe metabolism [58]. Conversely, absence of Nramp1 was associated with enhanced production and signaling by the anti-inflammatory cytokine IL-10 [59] and SLC11A1 genetic polymorphism was correlated with variation in LPS-induced IL-10 secretion [60]. Because Fe metabolism is central to macrophage function, and given the impact of cytoplasmic Fe on ROS-based signaling, it is possible that NRAMP1 cell-autonomous activity at the phagosomal membrane affects macrophage inflammatory potential, hence exerting pleiotropic effects on immune response [61,62,63,64,65]. It remains to be established with statistical significance whether the relatively high frequencies of SLC11A1 promoter (GT)_n repeat alleles 2 and 3 represent an equilibrium between the survival advantage of TB resistance and increased risk of auto-immune and inflammatory diseases such as type 1 diabetes [36,66].

The hypoxia-inducible factor alpha (HIF-1α) represents a strong candidate transcriptional activator binding at SLC11A1 promoter (GT)_n repeat. HIFs are O₂ sensitive transcription factors mediating transcriptional adaptation to hypoxic environments [67]. When stabilized under low O₂, HIFs can translocate to the nucleus and dimerize with their obligate partner Aryl hydrocarbon receptor nuclear translocator (ARNT) a.k.a. HIF-1β , then recruit coactivators such as histone acetyltransferase (HAT) activities (e.g., CBP/p300). HIF heterodimers activate genes involved in adaptation to hypoxic stress through recognition of and binding to hypoxia-response elements in their promoter. More broadly, HIFs regulation senses oxygen availability, redox status, nutrient availability, as well as inflammatory signals. Hence HIFs integrate metabolic and innate immune responses to infection and inflammation; HIF1α activity predominates in M1 macrophages whereas HIF-2α is preponderant in M2 macrophages [67].

HIF-1α is key for M1 microbicidal macrophage functions since HIF-1α basal activity can be stimulated in non hypoxic conditions by a variety of compounds such as growth factors, cytokines, nitric oxide, LPS and a range of infectious microorganisms [68,69]. NF-ΚB activity is required for HIF-1α gene expression and for HIF-1α protein accumulation in response to hypoxia; also, HIF-1α is activated during in vitro differentiation of MNs into macrophages. Microbial pathogens such as Gram-positive (Group A Streptococcus, Staphylococcus aureus) and Gram-negative (Pseudomonas aeruginosa and Salmonella Typhimurium) bacteria, as well as protozoan parasites (Toxoplasma gondii, Leishmania amazonesis) and viruses induce HIFs in normoxia [68,69].

SLC11A1 proximal promoter (GT)_n repeat variation regulates allele expression. This microsatellite contains two predicted hypoxia responsive elements and demonstrates Z-DNA-forming ability both in vitro and in vivo. Transient co-transfections in Baby hamster kidney cells revealed that under normoxic conditions HIF-1α induced up to 4-fold increase in luciferase expression while HIF-2α could not transactivate NRAMP1 promoter, and SLC11A1 expression levels after transfecting Chinese hamster ovary cells depended on the presence of HIF-1α. In murine macrophages co-transfected with HIF-1α and SLC11A1 proximal promoter transcriptional activity was upregulated in response to hypoxia mimetics and HIF stabilizers (CoCl₂, iron chelator dipyridyl, nitric oxide). The microsatellite allele 3 binds directly the transcriptional regulator HIF-1α/ARNT heterodimers in vitro, and in vivo when THP-1 macrophages are activated by pathogen or proinflammatory signals [49].

Current genetic data thus suggest that SLC11A1 promoter cis-acting regulatory variation as well as synonymous coding polymorphism can possibly contribute to heritable differences in gene/protein activity that may account for increased resistance/susceptibility to TB. In this regard, appraisal of all the regions involved in the transcriptional regulation of SLC11A1 will facilitate localizing functional SNPs that may influence with gene expression.

The myeloid lineage is part of the haematopoietic system which comprises blood cells and is organized hierarchically. Under the current paradigm of hematopoietic differentiation blood cell types arise from hematopoietic stem cells (HSC) which can self-renew or evolve through multiple progenitor and intermediate maturation states into 12 terminally differentiated cell types [76,77]. Phenotypically, distinct lymphoid and myeloid branch points from the HSC occur early during blood cell development, albeit probably not simply dichotomously [78]. Progeny of the common lymphoid precursor (CLP) include B cells, T cells, NK cells and plasmacytoid dendritic cells (DCs) whereas the megakaryocyte-erythroid progenitors (MEPs) and granulocyte-macrophage and myeloid DC (mDC) progenitors (GMP-DC) are generated from common myeloid precursors (CMPs, Figure 2A) [76,79].

mRNA profiling of intermediate and terminally differentiated cell types revealed that genes are tightly coexpressed in modules which are restricted to specific lineages or common to multiple hematopoietic lineages and interconnected. Five dominant phenotypes: HSPCs, differentiated erythroid cells, granulocytes/MNs, B cells and T cells are distinguished by a set of differentially expressed genes specific to each lineage as compared to the others [76]. Analyses in HSPCs showed that the promoter of genes expressed at high level in mature granulocytes and MNs may be bound early in hematopoiesis by factors that specify and maintain differentiation, such as PU.1 and the CAAT enhancer binding protein (C/EBP) β [76].

SLC11A1 high levels of expression in blood neutrophils and MNs prompted testing the promyelocytic cell line HL-60 as a model to study SLC11A1 regulation during either monocytic or granulocytic differentiation pathways using various pharmacological agents [70]. SLC11A1 transcript accumulation was detected co-induced with selected marker genes after 4–6 days differentiation into macrophage- (Phorbol miristate acetate, PMA, MCSF-R), monocyte- ((1α,25)OH₂ VitD3, VitD, CD14) or granulocyte- (DMSO, DMF, IL8Rb) like cells, suggesting that SLC11A1 expression occurred late in the myelo-monocytic differentiation program. Accordingly, SLC11A1/NRAMP1 was detected in tertiary granules of neutrophils [72]. However, gene expression was not induced in response to granulocytic differentiation triggered with all-trans retinoic acid (ATRA), classically used for leukemia differentiation therapy [81]. The data indicate that SLC11A1 expression is induced along either mono- or granulocytic pathways, and protein detection in the membrane of phagosomes of both types of phagocytes supports that SLC11A1 constitutes a functional marker of professional phagocytes [71,72]. The results also imply that selective pathways induce SLC11A1 expression, presumably by activating particular (combination of) transcription factors.

Vitamin D active metabolite ((1α,25)OH₂ VitD3, VitD) binds the vitamin D receptor (VDR) and the VitD-VDR complex assembles with the retinoic X receptor (RXR, which binds 9-cis retinoic acid, 9-cis RA) [82] or with itself to form VDR-VDR homodimers [83,84]. Dimeric VDR complexes move to the nucleus where they bind to accessible VitD response elements. The VDR is ubiquitously expressed and there is growing interest to study relationships of serum levels of VitD serum precursor, 25OH VitD3, to chronic metabolic, cardiovascular, neoplastic and immunologic diseases and for considering VitD supplementation in prevention and treatment of numerous disorders [85].

Retinoids do not induce SLC11A1 expression despite inducing granulocytic maturation (ATRA) [70] or representing the specific ligand of VDR preferred partner for heterodimerization (RXR, 9-cis RA) [71]. Since VDR functions also as homodimer and NRAMP1 expression was more efficiently up-regulated specifically in the presence of VitD genomic vs. non-genomic agonists [71], VDR-dependent nuclear events were presumed. Gene regulation by VitD involves local chromatin remodeling events that occur in a time frame that varies with target genes [86,87,88]. SLC11A1 expression was found slow and moderate compared to the monocyte marker CD14, implying perhaps an indirect process mediated by a VitD-induced factor that would bind to and regulate SLC11A1 promoter.

The regulation of SLC11A1 by VitD may have physiological implications since this secosteroid hormone stimulates, through VDR binding, myelopoiesis and the maturation of MNs towards macrophages with an M2—or anti-inflammatory—phenotype [89,90]. VitD also potently inhibits the maturation of dendritic cells into immunogenic antigen presenting cells [91,92,93,94,95,96]. At the same time VitD plays a key role notably through macrophages in tissue repair and peptide antimicrobial response in mammals [93,94,97,98]. Both VitD and VDR contribute to host innate resistance to infections, especially with Mtb [29].

Local productions of VitD by epithelial and immune cells as well as adipocytes exert autocrine or paracrine immunomodulating effects [85,96,99]. Hence injury or infection trigger via macrophage and keratinocyte TLR2/1 the synthesis of IL-15; IL-15 stimulates cytochrome P450, family 27, subfamily B, polypeptide 1 (CYP27B1) activity that enables transformation of 25OH VitD3 into the active form of the hormone, VitD, which in turn boosts autophagic and antimicrobial responses as well as pathogen detection [100,101]. Such autocrine production of VitD by macrophages notably allows to mount potent antimicrobial effectors that effectively counter-act the strategies of the devastating intracellular pathogens Mtb and HIV [102,103,104].

This paracrine antimicrobial role of VitD is required for IFN-γ-mediated microbicidal activity of human macrophages. T_H1 cell-secreted IFN-γ induces VitD antimicrobial pathway (cathelicidin and β4 defensin peptides, CYP27B1 and VDR), which is similar to the response triggered by TLR2/1 ligands (e.g., Mtb-derived 19-kD triacylated lipopeptide). Both rely on monocyte secretion of IL-15 albeit through different pathways (STAT1 or MyD88). Monocytes stimulation by IFN-γ depends on the presence of sufficient VitD serum precursor as IFN-γ stimulates CYP27B1 activity and treatment with IFN-γ is inhibited by a VDR antagonist. In addition, IFN-γ-induced autophagy and phagosome maturation are VDR-dependent. A concentration superior to 45 nM of 25OH VitD3 is strictly required to observe IFN-γ—and IL-15-dependent monocyte antimicrobial responses (autophagy, antimicrobial peptides). VitD appears thus instrumental for acquired immunity to activate macrophage antimicrobial responses that overcome intracellular pathogens evasion strategy [105].

In HL-60 promyelocytic cells, inducing differentiation with VitD was required to obtain upregulation of the accumulation of the endogenous SLC11A1 mRNA in response to IFN-γ. However, once isolated from its native chromatin environment SLC11A1 proximal promoter became directly responsive to IFN-γ treatment for two days, even though the cytokine induces little monocytic differentiation [71]. Thus SLC11A1 may be part of the antimicrobial arsenal that is primed by VitD (or genomic agonists) and VDR, whose activity increases SLC11A1 proximal promoter accessibility and responsiveness to transcriptional stimulation by antimicrobial factors such as IFN-γ. SLC11A1 sequence 647 bp upstream of the ATG comprises a basal promoter element, starting 263 bp 5' of the ATG, which alone drives maximal transcriptional activity in non myeloid background (T lymphocytes and epithelial cells) and independent of VDR agonist. More distal upstream elements are required for maximal promoter expression level in promyelocytic HL-60 cells and to obtain further VDR-dependent transcriptional upregulation [71].

During differentiation various epigenetic marks decorate the genome to demarcate domains of activity that are affected in a hierarchical manner. For instance, HSC pluripotency genes get progressively turned off with the onset of differentiation, by recruiting histone deacetylases and affecting histone methylation pattern to induce formation of heterochromatin [77]. Conversely, pioneer transcription factors such as the haematopoietic lineage-determining factors (e.g., PU.1 and C/EBPs) can interact with large sets of cis-regulatory elements by “opening” inaccessible chromatin domains, as their binding initiate nucleosome remodeling and prime targets genes for expression in cell-type-specific manner [106]. SLC11A1 lacks conventional TATA or CAAT box and other signals generally important for transcription activation [107]. Progressive deletion of SLC11A1 candidate promoter region 647 bp upstream of the ATG and DNAse I footprinting analysis allowed to test the hypothesis that transcription factors gain access and bind to this region during myelo-monocytic differentiation.

SLC11A1 major transcription start site (TSS) was mapped by 5' RACE PCR and S1 mapping in monocytic THP-1 cells and HL-60 cells differentiated into MNs, respectively [108,109]. Two cis-acting sites required for myelo-monocytic expression and the binding factors they recruit were identified. Double stranded DNA probes corresponding to footprints protected in vitro from DNAse I digestion after incubation with nuclear extracts from differentiated HL-60 cells were characterized by electromobility shift and supershift assays. The candidate cis elements were tested in vitro by site-directed mutagenesis and the effect of linker-mutations was verified by reporter assays after transfection or co-transfection of promoter constructs with selected transcription factors. Interaction of the candidate factor with the promoter area containing the site defined was verified by chromatin immuno-precipitation (ChIP) assay. The results showed that SLC11A1 major TSS is adjacent to a 5' C/EBP binding site, which is required for transcriptional activation during monocytic and granulocytic differentiation, and within an area bound by C/EBPα in promyelocytic cells and by C/EBPβ beginning 24 h after induction of differentiation with VitD. Site-directed mutagenesis of this C/EBP binding site also abrogated transcriptional activity of the promoter in non myeloid background implying it is required for recruiting the basal transcription complex. A binding site for the Specificity protein 1 (Sp1), which transactivates SLC11A1 promoter in vivo, was delineated in the more upstream region of the proximal promoter that is required for myeloid expression [109].

Several members of the C/EBP family of bZIP transcription factors have important roles in myeloid development, and in macrophage in particular, as C/EBPα , C/EBPβ and C/EBPδ represent three of the 18 regulators most frequently represented in the 14 modules enriched for macrophage-related gene signatures [25]. The induction of C/EBPβ expression by another bZIP family transcription factor, the cAMP responsive element-binding protein (CREB), results in specific upregulation of M2-associated genes in response to LPS; C/EBPβ is also required for muscle repair after injury, indicating that the CREB-C/EBPβ axis is crucial for M2 macrophage polarization [110]. C/EBPβ contributes in general to regulate myeloid proliferation and differentiation, the expression of inflammatory genes as well as host defense against microbial infections [111].

C/EBPβ (a.k.a. NF-IL6) is a key regulator of monocytic cells development [111]. It serves as “pioneer factor” providing the sequence specificity to endow general chromatin remodelling complexes, including the “switch/sucrose nonfermentable” (SWI/SNF) nucleosome remodelling complex, with transcription enhancer-specific roles [112]. More generally C/EBPs form a functional interface with the transcription factor PU.1 member of the E26 transformation specific (ETS) family, which is expressed at high levels to modulate the regulation of cell-specific macrophage functions [110]. Hence, PU.1 and C/EBP α/β convert fibroblasts into macrophage-like cells [113] and determine the ratio of myelo-monocytic to erythrocytic cells committing from common myeloid progenitors (CMP, Figure 2A) [76]. C/EBPα is indispensable for early granulocyte development, as the balance between PU.1 and C/EBPα controls the bifurcation between the mono- and granulocytic pathways, and C/EBPα upregulates the transcription of several granulocyte-specific factors [78]. All the other C/EBP isoforms also contribute to the transcriptional regulation of granulocytic genes later in development after the promyelocyte stage [78]. Given their prominent and complementary roles in myelo-monocytic differentiations it seems fitting that SLC11A1 expression in professional phagocytes is controlled by C/EBPs.

Regarding macrophages, SLC11A1/NRAMP1 gene expression was detected not only in pro-inflammatory (M1) but also anti-inflammatory (M2) macrophages [114,115]. Moreover, Nramp1 activity was also found to contribute to erythrocyte iron recycling after inducing hemolytic anemia [73] and the protein localizes to the membrane of phagosomes containing apoptotic erythrocytes [74]. C/EBPβ expression is stimulated by and mediates the monocytic differentiation program induced by VitD, known to skew macrophage phenotype towards M2 activity [89,90]. The VDR, which mediates VitD genomic actions [116] is another key determinant of human myelo-monocytopoiesis [76]. At homeostasis, macrophage M2 polarization is maintained notably through recycling of apoptotic cells by phagocytosis. Apoptotic cells liberate various mediators including oxysterols and fatty acids. These lipid mediators are agonists of the LXRs and PPARs, nuclear receptors related to VDR that coordinate efficient engulfment and anti-inflammatory metabolic recycling of apoptotic cells [117,118,119]. These data suggest that C/EBPβ control of NRAMP1 expression may relate to steady state activities of tissue macrophages and maintenance of tissue integrity.

The contribution of the Specificity protein 1 (Sp1) to regulate SLC11A1 transcription may support this view because Sp1 has been widely associated with basal levels of constitutive expression, including at various myeloid promoters. Despite being widely expressed the transcription factor Sp1 can regulate the expression of tissue-specific genes through combinatorial effects with a limited number of key transcription factors [120]. Hence, Sp1 is essential for myeloid-specific promoter activity of C/EBPδ , the complement and pattern recognition receptors CD11b and CD14, respectively, lactoferrin and the Myeloid Elf-1 like Factor (MEF) which codes for an ETS protein that activates the promoters of myeloid effectors and stimulators such as the granulocyte macrophage colony-stimulating factor, interleukin-3, lysozyme, human beta defensin-2 and perforin [121,122,123,124,125,126]. As previously observed in other myeloid promoters [120,124,126] several Sp1 binding sites were found in SLC11A1 proximal promoter upstream region which is required to confer myeloid specific expression and VDR-dependent upregulation. Mutating one binding site for Sp1 had less impact than abrogating SLC11A1 TSS C/EBPα/β site [71,109]. Hence, it is speculated that Sp1 contributes to SLC11A1 myeloid-specific expression through combinatorial interactions including transcription factors key for myelomocytic differentiation such as C/EBPβ.

Another cis element key for SLC11A1 expression along macrophage-like differentiation induced with PMA was recently characterized. PMA induces SLC11A1 transcriptional activation in HL-60 cells and the mRNA is stabilized by the interaction of a 3' AU-rich element (ARE) with a ubiquitously expressed RNA-binding protein, Hu antigen R (HuR). PMA-induced migration of HuR from the nucleus to the cytoplasm paralleled increased binding of HuR to SLC11A1 ARE and the accumulation of transcript. Overexpression of HuR in HL-60 cells stabilized SLC11A1 mRNA levels induced after 3 days PMA treatment, whereas siRNA HuR knockdown limited both SLC11A1 transcript and protein accumulation [127]. The RNA recognition motifs of HuR interact with mRNAs involved in cell cycle, cell death and differentiation, immunity, and inflammation. HuR can either stabilize or promote destabilization of its targets, through interactions involving suppressive RNA-binding proteins (RBPs), microRNAs, and associated factors. Hence HuR-null macrophages show exacerbations in the biosynthesis of inflammatory cytokines and chemokines; HuR is essential for balancing proinflammatory response and homeostatic regulation to control the extent of the inflammatory response [128]. Given the pro-inflammatory potential of SLC11A1/NRAMP1 activity it should be interesting to further analyze how macrophage activation phenotype may modulate HuR-dependent SLC11A1 protein expression level.

Progress in understanding molecular mechanisms of SLC11A1 transcriptional regulation during macrophage-like differentiation of HL-60 cells pointed at a PMA-responsive element, which was located by nested deletions of a reporter construct including the upstream myeloid-specific region. The PMA responsive element is adjacent in 5' to the (GT)_n repeat sequence that is converted into Z-DNA conformation and which contains two binding sites for the HIF-1α/ARNT heterodimer [49,129] (Section 3.3). Macrophage-like differentiation induces recruitment to the PMA-responsive element of the Activator protein-1 (AP-1)-like factor ATF-3/JunB, together with β-actin and BRG1, which were revealed by DNA affinity pulldown assays using double stranded oligonucleotide probes and antibodies against ATF-3, JunB, β-actin and the actin dependent regulator of chromatin BRG1, as well as ChIP assays. ATF-3 siRNA knockdown prior to DNA affinity pulldown assays targeting BRG1 and β-actin indicated that the presence of ATF-3 was necessary to recruit BRG1 and β-actin to the DNA probe, and immunoprecipitation assays confirmed that the three factors form a complex [129].

In the days following PMA-induced differentiation of HL-60 cells, β-actin translocates from the cytoplasm to the nucleus in a process that requires p38 MAPK activity [129]. ChIP-on-chip assays showed the genome-wide map of nuclear β-actin binding to promoters of genes involved in diverse functions (cell growth and differentiation, ion transport, adaptive and immune responses and signaling). β-actin binding was correlated with recruitment of RNA polymerase II (RNA Pol II) at six promoters including SLC11A1 [130]. siRNA knockdown of β-actin, ATF-3 or BRG1 reduced SLC11A1 transcription while co-transfection assay showed additive effects of these factors to activate SLC11A1 transcription. The formation of Z-DNA was detected by transfection of a SLC11A1 promoter construct in differentiated HL-60 cells, which were then cross-linked with formaldehyde and permeabilized before adding a fusion protein that binds Z-DNA structure and catalyzes double-stranded DNA cleavages in the vicinity. Detection of the cleavage sites by LM-PCR showed digestion products in cells differentiated with PMA that was reduced by siRNA BRG1 knockdown, demonstrating that this ATPase subunit of the SWI/SNF chromatin remodeling complex plays a critical role in Z-DNA formation and SWI/SNF-mediated transcriptional regulation [116].

The Activating transcription factor 3 (ATF3) is a member of the ATF/cyclic AMP responsive element binding family (CREB) family of transcription factors. ATF3 is expressed in a number of splice variants and interacts with multiple partners to form dimeric complexes, acting either as repressor or activator of various genes. For instance in Toll-like receptor (TLR)-activated macrophages, ATF3 recruits the histone deacetylase 1 (HDAC1) to the promoters of genes encoding cytokines or involved in metabolic and inflammatory responses, exerting negative regulation by epigenetic modification [131]. ATF3 is generally recognized as a repressor that is transcriptionally up-regulated by TLR signaling (TLR2/6, TLR3, TLR4, TLR9) and which regulates negatively the transcription of pro-inflammatory cytokines, such as interleukin (IL)-6 and IL-12p40, notably by recruiting HDAC1 and antagonizing transcription of target genes. Accordingly, atf3 deficiency induces overt susceptibility to endotoxic shock induced death [132]. However, ATF3 has the ability to interact with a number of transcription factors and other bZIP-containing proteins including AP-1 (c-Jun, JunB, JunD), C/EBP or MAF families of proteins, and depending on the promoter context, these heterodimers can act either as repressors or activators so that the role of ATF3 as transcriptional repressor or activator cannot be generalized [132]. In addition, microsatellite allelic variation can affect the binding of ATF-3 and c-Jun, as shown for SLC11A1 proximal promoter [50].

The recruitment of β-actin to SLC11A1 promoter stimulates transcriptional activation [130]. The basic property of β-actin is to polymerise into helical filament which can be used to produce force inside the cell. Actin in a nucleus (nuclear actin) is a component of many important chromatin remodeling complexes and it has been connected to most steps of the transcription process by all three RNA polymerases from the regulation of transcription initiation to the processing of pre-mRNAs [133]. It has been proposed that β-actin and actin-related proteins produce the maximum ATPase activity of SWI/SNF and favor stable association between chromatin and the remodeling complex [134]. Cell differentiation is normally a non-reversible process that silences transcriptional programs that would otherwise reactivate genes which were active in stem cells. But abundant actin in the Xenopus oocyte nucleus enables to reactivate heterologous embryonic genes, indicating that nuclear actin allows transcriptional activation by de-repressing silenced genes and this regulation is an evolutionary conserved feature [135]. Accordingly, ATF-3/JunB binding would allow the recruitment of β-actin and BRG1 to SLC11A1 promoter, enhancing the ATPase activity of the SWI/SNF chromatin remodeling complex to create a Z-DNA structure that may facilitate interaction with HIF-1α/ARNT heterodimers and the activation of SLC11A1 transcription in the days following induction of differentiation with PMA.

SLC11A1 transcriptional activation during VitD and PMA differentiation followed similar time frames. Since β-actin translocates from the cytoplasm to the nucleus after two days differentiation induced with PMA and, because C/EBPs can also recruit the SWI/SNF chromatin remodeling complex, it would be interesting to determine if VitD-induced SLC11A1 transcription depends also on β-actin stimulation. Alternatively the mode of transcriptional activation could vary depending on the developmental program since C/EBPα was shown to bind SLC11A1 TSS in undifferentiated promyelocytic cells. In contrast, NRAMP1 Z-DNA interaction with ATF-3 and HIF-1α/ARNT were evidenced in cells more advanced along the monocytic differentiation pathway, and in response to typical M1 macrophage activating stimuli such as LPS and IFN-γ [136]. Yet both HIF-1α and C/EBPs contribute to regulate cell energy metabolism, which is affected by macrophage polarization, and recent evidence suggested that HIF-1α and C/EBPα can interact and induce reciprocal functional changes [137]. In monocytic U937 cells, C/EBPα directly up-regulates the transcriptional expression of galectin-1, also regulated by HIF-1α, which interacts with and enhances the transcriptional activity of C/EBPα [138].

Induction of SLC11A1 transcription early in the myelo-monocytic development program may require as well recruitment of the co-activator Mediator multiprotein complex, which acts as molecular bridge between promoter-bound activators and the core transcriptional machinery. Both C⁄EBPβ and Sp1 can bind elements of this complex which may serve to integrate signaling toward the transcriptional machinery during cell differentiation. Another possibility would consist in recruiting coactivator complexes with HAT activities such as p300; histone acetylation may either provide binding surfaces for other activator proteins or facilitate chromatin decondensation to increase accessibility to the transcription machinery. However it might be equally important to consider as well the possible contribution of distal elements which may provide critical signals to recruit the basal transcription machinery [139,140,141,142].

SLC11A1/NRAMP1 localizes at 2q35 in a densely populated chromosomal region between the gene encoding the unknown Orf C2ORF62, upstream, and CTDSP1, coding for the CTD (carboxy-terminal domain, RNA Pol II, polypeptide A) small phosphatase 1, adjacent to SLC11A1 3' end (Figure 3A). The three genes have the same orientation coded by plus strand and are apparently free of miRNA regulatory sites (TargetScan, not shown). Few CpG islands are present in regions free of repeated sequence. These are 0.5–2 kilobase (kb) DNA fragments rich in CpG dinucleotides, usually constitutively protected from methylation by specific transcription factors such as Sp1, and which adopt an accessible conformation for potential regulatory factors [136]. One is found at the 3' end of C2ORF62 and two delimitate CTDSP1; three smaller elements (<300 bp) are present within SLC11A1, between exons 7–8 and 10–11, while a few CpG dinucleotides were found mainly clustered in the basal proximal promoter region around the TSS [109].

A single predicted transcription start site (SwitchGear Genomics) locates 41 bp downstream the major TSS mapped experimentally [108,109]. SLC11A1 TSS established by independent approaches maps 148 bp 5' of the translational initiation codon. The proximal promoter region of SLC11A1 and those of flanking genes display sequence conservation among vertebrates, together with exons and 3' UTR (SLC11A1 and CTDSP1); some areas in SLC11A1 putative distal promoter also show sequence conservation suggesting that some cis regulatory elements might have been preserved across species.

SLC11A1 locus displays several areas that correspond to hypersensitive sites to DNAse I digestion in intact nuclei among 148 cell types, which were identified by direct sequencing of the ends of DNAse I “double-hit” fragments (UW DNAse I HS, Figure 3B) [153]. The resulting sequencing footprints correspond to individual DNAse I cutting events that indicate open chromatin regions and correlate with active transcription. Some of the DNAse I footprints were detected in every cell type tested whereas others were found only in myelo-monocytic cells: CD14⁺ MNs, HL-60 and NB4 promyelocytes; some are also present in the megakaryocytic lineage (K-562) and not in hepatocyte, lymphocyte or fibroblast cells for instance (HepG2, GM12878 and NHLF, respectively, Figure 3D). Several sequence segments sensitive to DNAse I digestion (Figure 3B,D) do overlap with small chromosome fragments delineated independently by chromatin immunoprecipitation assay coupled to massively parallel sequence-based detection (ChIP-Seq) which targeted specific transcription factors interacting with DNA either directly or indirectly, in selected cellular background (SYDH TFBS, Figure 3E).

Substantial overlap between the sequence segments obtained by either enzymatic digestion of open chromatin or mechanical fragmentation based on specific interaction with nuclear proteins hence suggests candidate regulatory factors trans-interacting with potential cis acting elements [154]. For instance, three DNAse I footprints around SLC11A1 gene and found in all 148 cell types tested (Figure 3B) overlap with DNA fragments recovered by ChIP-Seq that bound ubiquitous factors, e.g., the locus insulator CCCTC-binding factor (CTCF) and RNA Pol II (Figure 3E). Insulator elements exert a dual function by preventing the spread of heterochromatin (barrier function) and transcriptional enhancers from activating unrelated promoters (enhancer blocking) [139].

Among the 50 DNAse I footprints examined over the selected 76,543 bp DNA fragment that spans from C2ORF62 to VIL (Figure 3B), 14 signals were absent from myelo-monocytic nuclear extracts, including nine found in C2ORF62 or VIL; seven footprints were found only in myelo-monocytic cells, and for seven others half of positive extracts were from the myelo-monocytic lineage.

Hence myeloid specificity of SLC11A1 transcription may be appreciated by comparing the Chromatin state segmentation results (based on ChIP-Seq data, Broad ChromHMM, for the insulator CTCF, histone marks of transcriptional activation, either methylation or acetylation, H3K4me1-3, H4K20me1, H3K36me3, H3K9ac, H3K27ac and histone methylation mark of inactive chromatin H3K27me3) between fibroblastic (NHLF), hepatocytic (HepG2), B lymphocytic (GM12878) and megakaryotic cells (data not shown).

The gene flanking SLC11A1 in 3', CTDSP1, is the most active and seems expressed in all cell types considered with by decreasing order: B lymphocytes, fibroblasts, megakaryocytes and hepatocytes. Lymphocytes and fibroblasts show strong enhancer activity and highly active transcription. In contrast, SLC11A1 is inactive in these cell types, either actively repressed by Polycomb proteins in fibroblasts or in the form of genetically inactive and tightly coiled heterochromatin in B lymphocytes (Broad ChromHMM). Though HepG2 chromatin shows weak/poised enhancer activity in the 3' part of SLC11A1, the 5' promoter is rather inactive. The myeloid K-562 nuclear context suggests weak level of gene transcription in the absence of enhancer or promoter activity, and not confirmed in transcriptomic analysis (ENCODE Caltech RNA-Seq, Figure 5A). Thus, as expected, SLC11A1 chromatin context lacks histone marks of transcriptionally active gene in non myelo-monocytic lineages.

A more detailed appraisal of local histone modification landscape and that includes nuclei of CD14⁺ MNs provides crucial information to delineate areas of myelo-monocytic specific transcriptional activity and compare these with DNAse I footprinted regions (Figure 4B,D,E). Regarding CTDSP1, CD14⁺ MNs chromatin displays abundant marks of high transcriptional activity (H3K4me1-3, H3K36me3, H3K9ac, H3K27ac and absence of H3K27me3) that are delimited by two strong CTCF binding sites. Among other cell types, the presence of dual methylation marks of transcriptional activation/inactivation in HepG2 cells (H3K79me2/H3K27me3, respectively) may explain lower levels of CTDSP1 transcript (Figure 5A). Concerning the putative Orf upstream of SLC11A1, C2ORF62, only HepG2 chromatin displays signs of promoter activity (H2A.Z, H3K4me1-3, H3K9ac) but no evidence of gene transcription such as VIL for instance (Figure 5A; and H3K79me2 and H3K36me3 marks in the body of VIL gene, Figure 4F). HepG2 data shows also some histone modifications at SLC11A1, but different from those observed in CD14⁺ MNs demonstrating strong promoter activity in several regions as well as gene transcription.

SLC11A1 gene demonstrates in CD14⁺ MNs histone modifications in four regions of different sizes that indicate strong enhancer and promoter activities as well as gene transcription. A large 5' region starting in the middle of C2ORF62 and spanning ~10 kb shows boundaries that seemingly correspond to DNAse I footprints which were found enriched in myelo-monocytic extracts and could indicate enhancer activity (DNAse I hypersensitive regions extending from Footprint #6 to 4_B, Figure 4A–D). Another large region starts at SLC11A1 TSS and extends to a common footprint found only in MNs and HL-60, and that yields stronger signal in mature cells consistent with gene expression (DNAse I hypersensitive regions #1,8,9, Figure 4A,D). The third region is smaller; it encompasses NRAMP1 3' UTR and exhibits marks of putative enhancer function (DNAse I hypersensitive regions #10 and 3, Figure 4A,D). The histone marks H4K20me1, H3K79me2 and H3K36me3 that typify transcribed genes link the second and third regions, implying the full coding region of SLC11A1 is marked epigenetically. Lastly, a smaller area upstream of SLC11A1 TSS is predominantly acetylated (DNAse I hypersensitive region #7, Figure 4A,D). All these regions show similar composition of histone marks, in which H3K27ac and H3K4me1,2 predominate thus suggesting the corresponding elements exert cell-type specific activity [154]. Indeed, in CD14⁺ MNs and MDMs, these epigenetic marks co-localize with sites of interaction with the transcription factors C/EBPβ and PU.1, both involved in human macrophage-specific gene regulation [80].

The 5' boundary of SLC11A1 predicted upstream enhancer element is indicated by moderately intense cell-type specific histone marks H3K4me1 and H3K27ac, which correspond to the position of a DNAse I footprint detected in promyelocytic bi-potential precursors as well as protein-DNA interactions identified in megakaryotic cells and involving the factors PU.1, EGR-1 and USF1 (DNAse I Footprint #6, Figure 4A–D). The 3' boundary of this large predicted enhancer region also shows preference for cell-type specific histone marks (H3K4me1 and H3K27ac) and matches areas that were sensitive to DNAse I selectively in myelo-monocytic cells, including predicted sites for C/EBPβ and other unknown factor (DNAse I Footprints #2,4_B, Figure 4A–D). At least four other DNAse I sensitive sequence elements were found in this region strongly marked with H3K4me1-2 and H3K9ac and H3K27ac (Figure 4A–D), which thus appears to comprise several candidate cis elements enhancing myelo-monocytic expression of SLC11A1. Several areas within this large 5' region were associated with transcription factors known to interact with macrophage specific enhancers (C/EBPβ, PU.1, EGR-2, Figure 6) [80].

The marked region that overlaps the first half of SLC11A1 Orf is delimited in 5' by the C/EBPβ site/TSS (DNAse I Footprint #1, Figure 4A–D), which corresponds to punctual deposition of H2A.Z variant, and in 3' by a commonly recognized element (showing few marks in lymphocytes and fibroblasts, not shown) that is hypersensitive to DNAse I digestion in MNs as well as HL-60 cells (DNAse I Footprint #9, Figure 4A–D). Again, both ends of the region show rather tissue-specific composition of histone marks (H3K4me1 and H3K27ac), and the first half displays also extensive H3K9ac mark. Regarding histone methylation marks, H3K79me2 starts at SLC11A1 TSS and covers the whole region, whereas H4K20me1 and H3K36me3 begin at the second exon or in the middle of this region and decorate the rest of the gene. This region is also notably covered with H3K4me3 mark, hereby indicating high level transcriptional activity (Figure 4A–D).

The downstream area located close to the end of SLC11A1 Orf spans two regions hyper-sensitive to DNAse I digestion (Footprints #3 and 10), which were detected in promyelocytic bi-potential precursors (Figure 4A,B). Corresponding protein-DNA interactions identified in K-562 megakaryotic cells involve the ETS factor ELF-1. This small area is predominantly marked by H3K4me1 (Figure 6) [80], as well as H3K4me2 and H3K27ac, consistent with a putative tissue-specific enhancer role, while H3K36me3 and H4K20me1 marks indicates that transcription proceeds to the 3' end of SLC11A1 (Figure 4A–D).

The fourth small region matches a footprint that was detected in MNs, whose position corresponds to a potential C/EBPβ site detected by ChIP-Seq in HepG2 cells upstream of the TSS (DNAse I Footprint #7, Figure 4A,B). It displays also typical histone marks of a lineage-specific transcriptional enhancer (H3K27ac, H3K4me1-2, H3K9ac, Figure 4A–D). Furthermore, in both CD14⁺ MNs and MDMs this area also displays H3K4me1 and H3K27ac marks and binds the transcription factors C/EBPβ and PU.1 (Figure 6) [80]. Hence, locus wide survey reveals excellent agreement between DNAse I footprints and both the distribution and composition of nucleosome histone marks. These data indicate that detailed analysis of candidate interacting transcription factors, including those identified by ENCODE and other ChIP-Seq analyses, will reveal the mechanistic basis of the developmental control of SLC11A1 gene expression in myelo-monocytic cells.

In comparison, in megakaryocytic cells the upstream region is partially marked, slightly at the 5' boundary and more strongly in its middle; also the third, downstream region is well marked at the 3' end (both regions bear strong H3K4me2 and H3 ac marks suggestive of enhancer activity, Figure 4E). But interestingly, the composition of histone marks detected in K-562 cells predicts rather common vs. cell-type specific regulatory elements [154].

In HepG2 cells, the ends of the upstream region are marked lightly as well as the downstream 3' region (H3K4me1,2, Figure 4F) whereas the intermediate region, covering the beginning of the SLC11A1 Orf, shows more marked boundaries (H3K4me1-3, and H3K27me3 indicative of silencing activity; Figure 4F). Hence, examining myeloid and non-myeloid cell types shows both in quantitative and qualitative aspects that the distribution of epigenetic marks indicating SLC11A1 transcriptional activity is tissue-specific.

To study further SLC11A1 predicted regulatory regions additional sets of ChIP-Seq data were included to examine the developmental modulation of the state of histone modifications at SLC11A1 locus (UW Histone and SYDH Histone) as well as the distribution of transcription factors potentially interacting with SLC11A1 sequence determinants.