HIF-1 Regulates Iron Homeostasis in Caenorhabditis elegans by Activation and Inhibition of Genes Involved in Iron Uptake and Storage

Caenorhabditis elegans ftn-1 and ftn-2, which encode the iron-storage protein ferritin, are transcriptionally inhibited during iron deficiency in intestine. Intestinal specific transcription is dependent on binding of ELT-2 to GATA binding sites in an iron-dependent enhancer (IDE) located in ftn-1 and ftn-2 promoters, but the mechanism for iron regulation is unknown. Here, we identify HIF-1 (hypoxia-inducible factor -1) as a negative regulator of ferritin transcription. HIF-1 binds to hypoxia-response elements (HREs) in the IDE in vitro and in vivo. Depletion of hif-1 by RNA interference blocks transcriptional inhibition of ftn-1 and ftn-2 reporters, and ftn-1 and ftn-2 mRNAs are not regulated in a hif-1 null strain during iron deficiency. An IDE is also present in smf-3 encoding a protein homologous to mammalian divalent metal transporter-1. Unlike the ftn-1 IDE, the smf-3 IDE is required for HIF-1–dependent transcriptional activation of smf-3 during iron deficiency. We show that hif-1 null worms grown under iron limiting conditions are developmentally delayed and that depletion of FTN-1 and FTN-2 rescues this phenotype. These data show that HIF-1 regulates intestinal iron homeostasis during iron deficiency by activating and inhibiting genes involved in iron uptake and storage.


Introduction
Iron is essential due to its presence in proteins involved in DNA synthesis, mitochondrial respiration and oxygen transport. Regulation of cellular iron content is crucial: excess cellular iron catalyzes the generation of reactive oxygen species that damage DNA and proteins, while cellular iron deficiency causes cell cycle arrest and cell death. Dysregulation of iron homeostasis caused by iron deficiency or iron excess leads to hematological, neurodegenerative and metabolic diseases in humans. Iron must therefore be maintained within a narrow range to avoid the adverse consequences of iron depletion or excess.
Maintaining iron content within this physiological range requires precise mechanisms for regulating its uptake, storage and export (for reviews, see [1,2]). In mammals, dietary non-heme Fe 3+ is reduced by membrane bound ferric reductases (e.g. duodenal cytochrome B or DCYTB) before transport across the enterocyte apical membrane by divalent metal transporter-1 (DMT1, also known as NRAMP2, SLC11a2 and DCT1) [3]. Cytosolic iron is either transported across the basolateral membrane into the circulation by ferroportin or sequestered in ferritin in a form unable to catalyze free radical formation [4,5]. Iron export by ferroportin is dependent on oxidation to Fe 3+ by membrane and soluble multicopper oxidases where it is incorporated into transferrin for delivery to tissues. When body iron stores are high, cytosolic iron is not exported into blood, and is instead sequestered into ferritin [5,6]. Iron in ferritin is lost by sloughing of enterocytes into the intestinal lumen.
Here, we show that HIF-1 activates smf-3 transcription and inhibits ftn-1 and ftn-2 transcription during iron deficiency. Transcriptional activation of smf-3 and repression of ftn-1 and ftn-2 is dependent on IDEs in their promoters that are similar but not identical. These studies show that HIF-1 is a key regulator of intestinal iron uptake and storage during iron deficiency in C. elegans.

HIF-1 inhibits iron-dependent transcription of ftn-1 and ftn-2
Previous studies showed that ftn-1 and ftn-2 transcription is activated by iron and inhibited by iron chelators [22][23][24]. Transcriptional regulation is mediated by the IDE located in the promoters of ftn-1 and ftn-2 genes ( Figure 1A). The IDE contains two WGATAR sequences that are binding sites for the intestinal specific ELT2 GATA transcription factor. Mutation of either of the WGATAR sequences abolished expression of an ftn-1::GFP-his reporter showing that ELT-2 is required for ftn-1 transcription under iron sufficient conditions [24]. The IDE also contains three canonical HREs (TACGTG) in the reverse orientation that have been identified in hif-1 target genes [15,31], suggesting a role for HIF-1 in iron-dependent ftn-1 and ftn-2 regulation.
To test this model, hif-1 RNAi was used to deplete HIF-1 in worms carrying an ftn-1::GFP-his or an ftn-2::GFP-his reporter. GFP expression was quantified using the COPAS Biosort after growth in NGM or NGM supplemented with the membrane permeable Fe 2+ chelator, 2,29-dipyridyl (NGM-BP) [32]. These reporters contain 1.9 kb of ftn-1 or ftn-2 promoter sequences, including the IDE, fused to the initiator ATG of nuclear-localized GFP-histone [24]. BP reduces expression of ftn-1::GFP-his and ftn-2::GFP-his in worms fed control (empty vector L4400) RNAi by 60% and 80%, respectively, compared to worms grown on NGM ( Figure 1B and 1C). By contrast, the BP-induced reduction in GFP expression is blocked by hif-1 RNAi. Furthermore, hif-1 RNAi increases GFP expression in worms grown on NGM, indicating that HIF-1 is expressed under normal growth conditions, and is capable of inhibiting ftn-1 and ftn-2 transcription.

HIF-1 binds to the IDE in vitro and in vivo
Electrophoretic mobility shift assays were used to determine whether HIF-1 binds to the HREs in the IDE. Radiolabeled wildtype IDE or an IDE containing mutations in the three HRE sites (HRE3m) was incubated with reticulocyte lysatesynthesized HIF-1 and AHA-1 ( Figure 3A). Complex formation

Author Summary
Due to its presence in proteins involved in hemoglobin synthesis, DNA synthesis, and mitochondrial respiration, eukaryotic cells require iron for survival. Excess iron can lead to oxidative damage, while iron deficiency reduces cell growth and causes cell death. Dysregulation of iron homeostasis in humans caused by iron deficiency or excess leads to anemia, diabetes, and neurodegenerative disorders. All organisms have thus developed mechanisms to sense, acquire, and store iron. We use Caenorhabditis elegans as a model organism to study mechanisms of iron regulation. Our previous studies show that the ironstorage protein ferritin (FTN-1, FTN-2) is transcriptionally inhibited in intestine during iron deficiency, but the mechanisms regulating iron regulation are not known. Here, we find that hypoxia-inducible factor 1 (HIF-1) transcriptionally inhibits ftn-1 and ftn-2 during iron deficiency. We also show that HIF-1 activates the iron uptake gene smf-3. Transcriptional activation and inhibition by HIF-1 is dependent on an iron enhancer in the promoters of these genes. HIF-1 is a known transcriptional activator, but its role in transcriptional inhibition is not well understood. Our data show that HIF-1 regulates iron homeostasis by activating and inhibiting iron uptake and storage genes, and they provide insight into HIF-1 transcriptional inhibition.
is only observed when HIF-1 and AHA-1 are present together in the reaction with wildtype IDE ( Figure 3A). Addition of HIF-1 antibody to the reaction led to the formation of a slower migrating HIF-1/AHA-IDE complex ( Figure 3A, lane 5). Formation of HIF-1/AHA-1-IDE complexes is competed away by unlabeled wildtype IDE but not by unlabeled HRE3m IDE, showing that HIF-1 specifically binds to the HREs ( Figure 3B).
To determine whether the IDE is a direct HIF-1 target in vivo, ChIP was performed on chromatin isolated from vhl-1(ok161) and hif-1(ia04) mutant animals. The binding of HIF-1 to the IDE or to the ftn-1 coding region used as a negative control was determined by ChIP using HIF-1 antibody. IDE DNA was enriched 4-fold in vhl-1(ok161) immunoprecipitates as compared to hif-1(ia04) immunoprecipitates ( Figure 3C). These studies indicate that HIF-1 binds to HREs in vitro and occupies the ftn-1 IDE in vivo. The DMT1 ortholog SMF-3 is regulated by HIF-1 during iron deficiency The activation of DMT1 by HIF-2a in iron-deficient mice increases intestinal iron uptake [7,8]. C. elegans express three DMT1 homologs, SMF-1, SMF-2 and SMF-3. SMF-1 and SMF-3 are expressed in the apical membrane in intestinal cells and are involved in Mn 2+ uptake [28,30], whereas SMF-2 is mainly expressed in pharyngeal epithelial cells [28][29][30]. To determine whether HIF-1 regulates smf-1, smf-2 or smf-3 expression during iron deficiency, their mRNA levels were quantified in N2 wildtype and hif-1(ia04) mutant animals cultured in NGM or NGM-BP. BP increases smf-3 mRNA levels 2-fold as compared to untreated N2 wildtype animals, but has no effect on smf-1 or smf-2 mRNA levels ( Figure 4A and data not shown). In hif-1(ia04) mutant animals, smf-3 mRNA levels are reduced by 50% as compared to N2 wildtype animals, and are not increased by BP ( Figure 4A).
Inspection of the 59 upstream regulatory region of smf-3 reveals a 118-nt element harboring three tandem GATA binding sites flanked by two HREs ( Figure 4B). To determine whether this element functions as an iron enhancer, transgenic strains carrying a transcriptional reporter containing 1500 nt of smf-3 promoter sequences fused to GFP-his were generated. GFP expression was quantified in these strains after growth on NGM, NGM-ferric ammonium citrate (NGM-FAC) or NGM-BP. FAC reduces GFP expression, whereas BP increases GFP expression as compared to untreated worms ( Figure 4C). To show that the BP-induced increase in GFP expression is due to iron chelation, GFP expression was quantified after culture of worms on NGM-BP in the presence of an equimolar amount of FAC or MnCl 2 . BP plus FAC, but not BP plus MnCl 2 , reduces smf-3(-1500)::GFP-his expression ( Figure S1). hif-1 RNAi completely blocks BP-induced increased GFP expression ( Figure 4D). Similarly, expression of a smf-3(-1500)::GFP-his reporter containing 250 nt of upstream sequences containing the IDE is increased by BP in control RNAi fed worms and is reduced by BP in hif-1 RNAi fed worms ( Figure  S2). Taken together, these data indicate that smf-3 is transcriptionally activated by HIF-1 during iron deficiency.

Iron content is reduced in smf-3(ok1035) mutant animals
The localization of SMF-3 to the apical membrane of intestinal cells [28,30] and its regulation by HIF-1 suggest a role for SMF-3 in intestinal iron uptake. If SMF-3 has a role in iron uptake, iron content might be expected to be reduced in smf-3(ok1035) mutant animals. We found that ftn-1 mRNA levels, which are positively correlated with cellular iron levels ( Figure 2), are reduced in smf-3(ok1035) mutants as compared to N2 wildtype animals, but not in smf-1(ok1748) or smf-2(gk133) mutant animals ( Figure 5A). Quantification of metal content by inductively-coupled plasma spec- troscopy (ICP) shows that total iron content in smf-3(ok1035) mutant animals is 45% of N2 wildtype animals consistent with reduced ftn-1 mRNA levels in these animals ( Figure 5B). The total iron content in smf-1(ok1748) and smf-2(gk133) mutant animals is not significantly different as compared to N2 wildtype animals. The total Mn content in smf-3(ok1035) mutant animals is 60% of N2 wildtype controls ( Figure 5B) in agreement with SMF-3 as a regulator of Mn uptake [28,30]. The Mn content in the smf-1(ok1748) and smf-2(gk133) strains tended to be lower as compared to N2 wildtype controls, but did not reach significance. These results are consistent with a role for SMF-3 in intestinal iron transport.
Reduced viability of hif-1(ia04) mutant animals by iron deficiency is rescued by ftn-1 and ftn-2 RNAi hif-1(ia04) mutant animals have no overt phenotype in normoxia, but display reduced viability in oxygen concentrations less than 1% [13,34]. Similarly, we find that hif-1(ia04) mutant animals develop normally under normoxic conditions, but are developmentally delayed when cultured under iron deficient normoxic conditions (NGM plus 20 mM BP) as compared to N2 wildtype animals ( Figure 6A). As total iron content is reduced in smf-3(ok1035) mutant animals and smf-3 expression is reduced in hif-1(ia04) animals, these data suggest that total iron content might also be reduced in hif-1(ia04) animals. ICP analyses show that the  total iron content in hif-1(ia04) mutant animals is 60% of N2 wildtype animals ( Figure 6B). The total Mn content is also reduced in hif-1(ia04) mutant animals.
We next questioned whether the developmental delay observed in hif-1(ia04) mutant animals cultured in BP can be rescued by reducing FTN-1 and FTN-2 expression, which would lead to an increase in the cellular labile iron pool. This pool contains chelatable redox-active iron that constitutes less than 5% of total cellular iron [32,35]. The modulation of ferritin levels is one mechanism for regulating the labile iron pool in mammalian cells: ferritin overexpression reduces the iron pool, while ferritin depletion increases this pool [36][37][38]. We depleted ftn-1 and ftn-2 by RNAi in N2 wildtype and in hif-1(ia04) mutant animals cultured on NGM or NGM plus 20 mM BP, and the number of worms reaching L4 stage was measured. ftn-1/ftn-2 RNAi increases the number of hif-1(ia04) mutant animals reaching L4 stage from 28% in untreated animals to 78% in ftn-1/ftn-2 RNAifed animals ( Figure 6C). These data indicate that the developmental delay observed in hif-1(ia04) mutant animals during iron deficiency can be partially rescued by reducing FTN-1 and FTN-2 levels.

Discussion
Previous studies showed that ftn-1 and ftn-2 transcription in intestine is activated by iron and inhibited by iron deficiency [22][23][24]. Transcriptional regulation is dependent on an IDE containing HIF-1 and GATA binding sites. The GATA factor ELT2 regulates basal intestinal ftn-1 and ftn-2 transcription during iron sufficiency, but how transcription is repressed by iron deficiency was not understood. Here, we identify HIF-1 as a negative regulator of ftn-1 and ftn-2 transcription during iron deficiency. We also show that smf-3 is regulated by HIF-1 during iron deficiency, but unlike ftn-1 and ftn-2, HIF-1 activates smf-3 transcription. The activation of smf-3 and inhibition of ftn-1 and ftn-2 by HIF-1 provides a mechanism to increase iron uptake and decrease iron storage during iron deficiency.
In mice, increased iron uptake by DMT1 during iron deficiency is mediated by HIF-2a [7,8]. In contrast to the transcriptional regulation of ftn-1 and ftn-2 in C. elegans, vertebrate ferritin H-and -L subunit mRNAs are translationally repressed during iron deficiency by the binding of IRP1 and IRP2 to an iron-responsive element (IRE) in ferritin mRNAs [26]. When cellular iron levels increase, IRP1 is converted into a cytosolic aconitase concomitant with loss of RNA-binding activity and IRP2 is degraded, which leads to increased ferritin translation. Although C. elegans express an IRP1 homolog (ACO-1), it lacks RNA-binding activity and functions solely as a cytosolic aconitase [22]. The regulation of iron homeostasis by ftn-1and ftn-2 is essential as depletion of FTN-1 and FTN-2 rescues the growth delay observed in hif-1 null worms grown under iron limiting conditions. Recent studies show that constitutive expression of ferritin-H and -L subunits in mice with intestinal specific deletion of IRP1 and IRP2 have reduced survival rates [39]. Depletion of the intestinal ferritin-H gene in mice leads to reduced iron sequestration in intestine and increased body iron stores [40]. Taken together, these studies show that the precise regulation of intestinal ferritin expression is essential for appropriate control of iron absorption.  The transcriptional inhibition of ftn-1and ftn-2 genes by HIF-1 during iron deficiency was unexpected. HIF-1 is a potent transcriptional activator of hundreds of HRE target genes [15,41,42]. HIF has also been reported to function as a repressor, but the mechanism of transcriptional repression is not fully understood [43][44][45][46][47][48]. For some HIF-negatively regulated genes, transcriptional repression may be an indirect effect due to HIF activation of a transcriptional repressor [41,42]. HIF-1 has also been shown to upregulate microRNA mir-120, which represses gene expression [49]. For other genes, direct binding of HIF to HREs in the promoters of negatively-regulated genes has been demonstrated using ChIP analysis [45][46][47]. These studies led to models whereby HIF-1 negatively regulates transcription by the recruitment of corepressors and histone modifying complexes or by competing with transcriptional activators for HRE binding.
One question is how HIF-1 mediates transcriptional activation and inhibition through the IDE. The ftn-1, ftn-2 and smf-3 IDEs differ in number, spacing and orientation of the GATA and HIF-1 binding sites. It is likely that architecture of the IDE dictates physical interactions of ELT-2, HIF-1, coactivators and other transcription factors to activate or repress transcription. Further studies are needed to determine how the structure of the IDE affects HIF-1 transcriptional responses.
HREs can be flanked by transcription factor binding sites, and it is the cooperation of HIF with these transcription factors that enhance transcription or direct cell specific expression [11]. The GATA binding sites flanking the HREs in ftn-1, ftn-2 and smf-3 are essential for intestinal expression because mutations of these sites or depletion of ELT-2 reduces intestinal expression of ftn-1 and smf-3 GFP transcriptional reporters [24] (preliminary data, SJR and EAL). Computational studies have shown that the majority of intestinal specific genes contain GATA binding sequences, leading to the notion that ELT2 is a global regulator of intestinal gene expression [50,51]. Several studies have shown that other transcription factors cooperate with ELT-2 to modulate its function in response to nutritional or physiological signals [52][53][54][55][56].
Based on our findings, a model for HIF-1 regulation of intestinal iron homeostasis is proposed (Figure 7). During iron sufficiency, ELT-2 binds to GATA binding sites in the IDE to activate intestinal expression of ftn-1 and ftn-2. Both GATA binding sites are required for ftn-1 expression as deletion of either site abolished expression of an ftn-1 GFP transcriptional reporter [24]. In addition, mutation of all three HREs in the ftn-1 IDE abolished GFP reporter expression, suggesting that a transcriptional activator may bind to the HREs enhancing ELT-2 function [24]. This activator may be a member of the basic helix-loop-helix (bHLH/PAS) family or a bHLH transcriptional activator that binds to noncanonical E-box elements [57][58][59]. smf-3 transcription is reduced during iron sufficiency due to decreased HIF-1a levels. Our data also show that a small amount of HIF-1 is expressed during normal growth conditions that can interact with HREs in ftn-1 and smf-3. During iron deficiency, HIF-1 is stabilized and transported to the nucleus where it dimerizes with AHA-1. We propose that HIF-1/AHA-1 competes with a transcriptional activator for binding to the ftn-1 and ftn-2 HREs, inhibiting ftn-1 and ftn-2 transcription. HIF-1/AHA-1 binds to the smf-3 HREs, recruits coactivators and cooperates with ELT-2 to activate transcription.
SMF-1, SMF-2 and SMF-3 have been characterized with regard to their role in Mn 2+ homeostasis and sensitivity [28,30]. SMF-1 and SMF-3 are localized to the apical intestinal epithelium consistent with a role in metal uptake [28,30], whereas SMF-2 is primarily expressed in pharyngeal epithelium [28,29]. SMF-1 and SMF-2 are also expressed in dopaminergic neurons where they mediate sensitivity of neurons to Mn and neurotoxins [29]. The high homology of SMF-3 with mammalian DMT1, its localization in the apical membrane of intestinal cells [28,30], its regulation by iron and a reduction in total iron content in smf-3(ok1035) mutant animals indicate a role for SMF-3 is intestinal iron transport in C. elegans. The regulation of iron homeostasis by HIF-1 provides a mechanism to ensure that C. elegans maintain sufficient iron stores for growth and survival when iron is limiting.

C. elegans strains and culture
Strains were cultivated on nematode growth medium (NGM) agar plates seeded with Escherichia coli OP50 at 20-22uC. For iron chelation experiments, synchronized larvae were grown for 24 h on NGM plates then transferred to NGM plates supplemented with 100 uM 2,29-dipyridyl (BP) for 18 h unless indicated. For iron experiments, larvae were grown for 18 h on NGM plates supplemented with 6.6 mg/ml ferric ammonium citrate (FAC). The pH of FAC-supplemented NGM agar was adjusted to pH 7.0.

Rescue
Mixed-stage populations of N2 wildtype and hif-1(ia04) animals containing gravid adults were transferred to RNAi plates seeded with bacteria expressing ftn-1 and ftn-2 dsRNA or empty vector control (L4400) for 24 h at 20-22uC. Synchronized larval stage (L1) worms were obtained by treating RNAi fed gravid adults with alkaline hypochlorite and allowing eggs to hatch overnight in sterile S-basal media. L1 larvae were spotted on ftn-1 and ftn-2 or L4400 RNAi plates supplemented with or without 20 mM BP and incubated at 20-22uC for 48 h prior to scoring for larval stage using a Leica MZ9.5 stereomicroscope.
C. elegans sorting ftn-1::GFP-his, ftn-2::GFP-his, smf-3(-1500)::GFP-his and smf-3 (-250)::GFP-his reporter lines were synchronized by treating gravid Figure 7. Model for HIF-1 iron-dependent activation and inhibition of intestinal iron uptake and storage. Iron sufficiency: ELT-2 binds to GATA binding sites located in the ftn-1 and ftn-2 IDEs. We propose that ELT-2 cooperates with an unidentified transcriptional activator (ACT) that binds to the HREs and enhances transcription. smf-3 is transcribed at basal levels during iron sufficiency. HIF-1 is expressed during normal growth conditions, but at low levels. Iron deficiency: HIF-1 is stabilized and dimerizes with AHA-1. HIF-1/AHA-1 (denoted by HIF-1) displaces the unidentified transcriptional activator binding to the ftn-1 and ftn-2 HREs and inhibits transcription. HIF-1/AHA-1 binds to the smf-3 HREs, recruits coactivators and cooperates with ELT-2 to enhance smf-3 transcription. Whether ELT-2 is bound to the ftn-1 GATA sites during iron deficiency and to the smf-3 GATA sites during iron sufficiency has not been determined. doi:10.1371/journal.pgen.1002394.g007 adults with alkaline hypochlorite followed by hatching eggs in Sbasal medium (0.1 M NaCl, 0.05 M potassium phosphate, pH 6.0, 5 ug/ml cholesterol) overnight. Synchronized L1 worms were grown on control or hif-1 RNAi plates for 32 h. Worms were then rinsed from the plates and washed in M9 buffer (22 mM KH 2 PO 4 , 42 mM Na 2 HPO 4 , 86 mM NaCl, 1 mM MgSO 4 ) and spotted onto fresh 10 cm control or hif-1 RNAi plates supplemented with 100 mM BP and grown for 18 h. Worms were rinsed from plates and washed with MT buffer (M9 buffer containing 0.1% Triton X-100) by repeated rounds of centrifugation until free of debris. Worms were analyzed using the COPAS Biosort (Union Biometrica, Somerville, MA) as described [24]. Prior to data acquisition, gating parameters were established by visualizing a sorted population by microscopy. The same gating parameters were used for all experimental conditions during GFP fluorescence acquisition and subsequent analysis. Data were analyzed using FCS Express Version 3.0 lite (De Novo Software, Ontario, Canada). Procedures for sorting non-RNAi-treated worms was the same as above except that synchronized L1 stage worms were grown on NGM plates seeded with OP50 for 32 h prior to being transferred to fresh NGM plates and NGM plates supplemented with either 100 uM BP, 100 uM FAC, 100 uM BP plus100 uM MnCl 2 or 100 uM BP plus 100 uM FAC. Worms were grown for an additional 18 h prior to data acquisition as performed above.

Microscopy
Images of GFP expression were captured using an Axio Imager (Carl Zeiss MicroImaging, Inc, Thornwood NY) outfitted with the Zeiss filter set 38HE (BP 470/40HE, dichroic FT 495 HE, BP 525/50 HE) and an AxioCam HRm camera (Carl Zeiss MicroImaging, Inc, Thornwood NY) using AxioVision software. Following acquisition, images were rotated, cropped and sized using Adobe Photoshop.

RT-PCR
Synchronized L1 larvae were grown on NGM plates seeded with OP50 for 48 h prior to harvest. For iron chelation experiments, synchronized worms were transferred to new NGM plates or NGM plates supplemented with 100 uM BP after 32 h and harvested at 48 h. At harvest, worms were washed from plates and cleaned by repeated rounds of centrifugation and resuspension in ddH2O. Pelleted worms were resuspended in TRIZOL (Invitrogen) and total RNA was purified according to manufacturer's protocol. cDNA was synthesized using 1 ug of purified total RNA using SuperscriptIII Supermix for qRT-PCR (Invitrogen). Semiquantitative RT-PCR (sqRT-PCR) was performed using Recombinant Taq Polymerase (Invitrogen). PCR products were run on ethidium bromide stained 1.2% agarose gels. Images were captured on a FluorChem IS-8900 (Alpha Innotech) and analyzed using ImageQuaNT (Molecular Dynamics) and ImageJ. Primer sequences are shown in Table S1.

Chromatin immunoprecipitation (ChIP)
ChIP assays were performed as previously described [62]. Mixed-stage vhl-1 (ok161) and hif-1 (ia04) worms were grown on two 150 mm NGM plates and harvested. Worm pellets were homogenized in crosslinking buffer (1% formaldehyde in PBS) and incubated for 20 min at room temperature. The reactions were quenched with glycine for 20 min and snap frozen. The frozen pellets were resuspended in HLB buffer (50 mM HEPES-KOH, pH 7.5, 150 mM NaCl, 1 mM EDTA, 0.1% sodium deoxycholate, 1% Triton-X100, 0.1% SDS, 1 mM PMSF) containing protease inhibitor cocktail (Calbiochem), sonicated and the supernatants were precleared using 30 ul salmon sperm DNA/ protein A agarose beads. After centrifugation, 10% of each supernatant was kept as input, while the remaining supernatants were incubated with CeHIF antibody (4 ul) overnight at 4uC. DNA-protein antibody complexes were incubated with Protein A agarose (30 ul slurry) for 2 h at 4uC, and the beads were then resuspended in proteinase K buffer (20 mg/ml proteinase K) and incubated at 45uC for 2 h. The samples were further incubated at 65uC overnight to reverse crosslinks. DNA was purified using a QIAquick PCR purification kit (Qiagen) according to manufacturer's protocol. Quantitative (q) PCR was then performed using Sybr Green mix (Invitrogen) run on an ABI 7000 Sequence Detection System and additionally visualized by 2% ethidium bromide stained agarose gels. Primers flanking the ftn-1 IDE and control primers within the ftn-1 coding region are shown in Table  S1.

Metal content by inductively-coupled plasma-optical emission spectroscopy (ICP)
Synchronized L1 worms were obtained by treating gravid adults from each strain with alkaline hypochlorite followed by hatching eggs in S-basal medium overnight. L1 worms were grown on OP-50 seeded NGM plates. L4 worms were washed extensively with M9 buffer and incubated in M9 buffer for 30 min at room temperature to allow for purging of the gut followed by two rinses with ddH2O. Empty tubes were run in parallel to serve as controls. Samples were pelleted and frozen at 280Cu. Samples and controls were brought up in 200 ul of metal free 40% nitric acid (Optima) and heated to 95uC for 2 min in a heating block. Solubilized samples were diluted to a final nitric acid concentration of 10% with ddH2O and measured on an Optima 3000 XL ICP-OES (Perkin Elmer). Serial dilutions of commercially available mixed metal standards were used to calibrate the instrument. Results were normalized to the simultaneously acquired signal for sulfur for each sample [63]. The data are presented as a percentage of N2 wildtype animals. At least three independent biological replicates were performed.

Statistical analysis
Data are presented as the means 6 SEM. Two-tailed unpaired Student's t test were used for statistical analysis. Data are considered statistically significant at p,0.05. Figure S1 ftn-1::GFP-his and smf-3(-1500)::GFP-his expression are regulated by iron, but not by MnCl 2 . MnCl 2 has been shown to regulate transcriptional expression of smf-3 [28,29]. To determine whether MnCl 2 regulates expression of smf-3(-1500)::GFP-his, L1 stage worms were cultured on NGM, NGM-100 uM BP, NGM-100 uM BP plus 100 uM FAC and NGM-100 uM BP plus 100 mM MnCl 2 for 18 hr. Worms were harvested and GFP expression was quantified by the COPAS Biosort. Dot plots of sorted worms are shown where each dot represents one worm. Mean GFP fluorescence (6 SEM) is expressed as a percentage of worms grown on NGM (n = 1000 worms/treatment), *p,0.0001. (A) BP increases smf-3(-1500)::GFP-his expression, which is blocked by addition of equimolar amount of FAC plus BP. By contrast, GFP expression is not reduced when MnCl 2 is present with BP. (B) BP reduces ftn-1::GFP-his expression and is reversed when FAC is present with BP. Addition of MnCl 2 with BP did not increase GFP expression. (EPS) Figure S2 smf-3(-250)::GFP-his expression is regulated by HIF-1. GFP-his reporters were constructed containing 250 nt of smf-3 promoter sequences 59 to the initiator ATG ( Figure 6A). This reporter contains the entire smf-3 IDE. To determine whether smf-3(-250)::GFP-his was regulated by HIF-1, L1 stage worms carrying smf-3(-250)::GFP-his were spotted on hif-1 or control (empty vector L4400) RNAi plates supplemented with or without 100 uM BP. GFP fluorescence was quantified by the COPAS Biosort after 18 h. Dot plots of sorted worms are shown where each dot represents one worm. Mean GFP fluorescence (6 SEM) is reported as a percentage of control RNAi fed worms on NGM (n = 650 worms/treatment), *p,0.0001. BP increases GFP expression in control RNAi fed worms after growth on NGM-BP as compared to NGM-control fed RNAi. By contrast, BPinduced increased GFP expression is inhibited in worms fed hif-1 RNAi. (EPS)