In eukaryotic cells, amino acid biosynthesis is feedback-inhibited by amino acids through inhibition of the conserved protein kinase Gcn2. This decreases phosphorylation of initiation factor eIF2α, resulting in general activation of translation but inhibition of translation of mRNA for transcription factor (TF) Gcn4 in yeast or ATF4 in mammals. These TFs are positive regulators of amino acid biosynthetic genes. As several enzymes of amino acid biosynthesis contain iron–sulfur clusters (ISCs) and iron excess is toxic, iron and amino acid homeostasis should be co-ordinated. Working with the yeast Saccharomyces cerevisiae, we found that amino acid supplementation down-regulates expression of genes for iron uptake and decreases intracellular iron content. This cross-regulation requires Aft1, the major TF activated by iron scarcity, as well as Gcn2 and phosphorylatable eIF2α but not Gcn4. A mutant with constitutive activity of Gcn2 (GCN2c) shows less repression of iron transport genes by amino acids and increased nuclear localization of Aft1 in an iron-poor medium, and increases iron content in this medium. As Aft1 is activated by depletion of mitochondrial ISCs, it is plausible that the Gcn2–eIF2α pathway inhibits the formation of these complexes. Accordingly, the GCN2c mutant has strongly reduced activity of succinate dehydrogenase, an iron–sulfur mitochondrial enzyme, and is unable to grow in media with very low iron or with galactose instead of glucose, conditions where formation of ISCs is specially needed. This mechanism adjusts the uptake of iron to the needs of amino acid biosynthesis and expands the list of Gcn4-independent activities of the Gcn2–eIF2α regulatory system.
Nutrient acquisition and metabolic pathways are regulated by feedback inhibition of their accumulated products, intracellular nutrients and final metabolites, respectively. The goal of this regulation is to guarantee adequate intracellular supply of final products while avoiding toxic overaccumulations. Regulation may operate at several levels such as transcription, mRNA stability and translation, protein stability and protein activity. In addition to this intrinsic regulation, nutrient acquisition and metabolic pathways respond to extrinsic cues from the environment and development and growth/proliferation stages of cells .
A major nutrient-sensing pathway in all eukaryotes is the one mediated by the conserved protein kinase Gcn2 . In this pathway, amino acid biosynthesis is feedback-inhibited by amino acids through inactivation of Gcn2. Accumulation of uncharged tRNAs caused by deficiencies in individual amino acids and other cellular stresses activates Gcn2 [2–4]. This kinase phosphorylates eukaryotic translation initiation factor 2 at Ser51 of its α-subunit (eIF2α) , which causes inhibition of translation initiation and reduces global protein synthesis. Phosphorylated eIF2α, however, favors the selective translation of mRNAs encoding the bZIP transcription factors (TFs) Gcn4 in Saccharomyces cerevisiae [2–4] and ATF4 in mammals . These TFs induce the expression of a large number of genes involved in amino acid biosynthesis and uptake [2,4,6,7] and autophagy [7,8]. This Gcn2-based regulatory pathway in yeast has been named the ‘General Amino Acid Control' . Some functions of Gcn2 are independent of Gcn4 and some others are not related to amino acid biosynthesis, uptake or autophagy. On the other hand, all known functions of Gcn2 depend on phosphorylatable eIF2α [1,2], and therefore we will refer to this regulatory system as the ‘Gcn2–eIF2α pathway', abbreviated as ‘GeIp'.
Several amino acids such as leucine, isoleucine, valine, lysine, methionine, cysteine, and glutamate have an iron-dependent biosynthetic step, that is, enzymes that contain iron–sulfur clusters (ISCs) as cofactors [9–11]. This implies that iron and amino acid biosynthesis should be co-ordinated because excess of cellular iron is toxic, generating reactive oxygen species (ROS), which damage lipids, proteins and DNA [10,11]. This implies that iron uptake should be cut down when cells reduce expression of iron-containing enzymes.
Whereas Gcn2 is the sole eIF2α kinase in yeast, mammals possess four homologous stress-sensing kinases able to phosphorylate eIF2α. Two of them, GCN2 and HRI, respond to amino acid starvation and iron/heme deficiency, respectively . Unlike mammals, iron deficiency does not activate a specific iron-sensing eIF2α kinase in yeast but it up-regulates many transcripts involved in amino acid biosynthesis and uptake [9,11,12], as expected if the GeIp were activated.
Iron transporters from duodenal enterocytes, plants and fungi have many conserved components such as ferric reductases, Fe2+–H+ symporters (low affinity and low specificity), multi-copper oxidases associated with Fe3+ permeases (high affinity and high specificity), and Fe3+-siderophore-H+ symporters [11,13–17]. In most animal tissues, iron uptake into cells occurs by endocytosis of iron–transferrin complexes but transmembrane transport is still needed to reach the mitochondria and the cytosol . Iron starvation induces expression of these transporters by activating iron-responsive TFs such as Aft1 in yeast [19,20], FIT [14,15] and PYE  in Arabidopsis, and hipoxia-inducible TFs such as HIF2α in animals .
The mechanism of the response to iron starvation has been analyzed in great detail in the model organism S. cerevisiae, where it is mediated by TF Aft1 and, to a lesser extent, by its paralog Aft2 [11,13,19,23]. Aft1 activates transcription of a set of genes that contains in its promoter the binding site motif YRCACCCR (Y = pyrimidines, R = purines) and that is known as the ‘iron regulon' [11–13]. The encoded proteins are involved in iron uptake, mobilization from intracellular stores and the metabolic remodeling that enables cells to adapt to iron scarcity. Cellular iron status does not affect expression of Aft1, rather it modulates its localization . Aft1 is transported into the nucleus in an iron-independent manner  but is exported to the cytosol by exportin Msn5 when intracellular iron levels are high . Thus, Aft1 accumulates in the nucleus upon iron depletion. Nuclear export of Aft1 is not essential for inhibition of Aft1 activity under iron-replete conditions, but the critical step is its dissociation from DNA . Sensing of intracellular iron by Aft1 requires glutaredoxins Grx3 and Grx4  and the cytosolic proteins Fra1 and Fra2 . Under iron-replete conditions, the Grx3/Grx4/Fra1/Fra2 complex binds to the ISC and interacts with Aft1 to induce its dissociation from target promoters [30,31]. The inhibitory signal of high-iron conditions depends on mitochondrial ISC biosynthesis by an unknown mechanism [30,31]. Sensing of iron starvation in animals and plants is not as well known as in yeast, but it seems that mitochondrial ISCs are critical regulators of cellular ion homeostasis in all eukaryotes . In vertebrates 2-oxoglutarate-dependent dioxygenases (hydroxylases) are sensors of energy metabolism, oxygen availability, and iron homeostasis .
Our work on the connection between iron and amino acid homeostasis started from the observation that supplementation with amino acids of yeast cells growing with ammonium as the sole nitrogen source represses genes of the iron regulon. This regulation is dependent on Gcn2 and phosphorylatable eIF2a but not on Gcn4 and is mediated by decreased nuclear localization of Aft1. This mechanism adjusts the uptake of iron to the needs of amino acid biosynthesis and expands the list of Gcn4-independent activities of the GeIp, and it may be operative in animals and plants.
Yeast strains and plasmids
Most of the S. cerevisiae strains used in the present work were isogenic to BY4741 (MATahis3Δ1 leu2Δ0 met15Δ0 ura3Δ0) . All null mutants (Δ symbol) were derived from BY4741 by gene disruption with kanMX4 [35,36]. The strain expressing a dominant mutant allele of Gcn2 (GCN2c-M719V-E1537G, in the following referred to as GCN2c)  was derived from a gcn2Δ mutant in the BY4741 background . A strain (RS-88) expressing the S51A mutation of eIF2α (SUI2 gene) containing pRS414 [SUI2-S51A, TRP1] and its wild-type strain (RS-86) containing pRS414 [SUI2, TRP1] were derived from H1645 (MATaura3-52 leu2-3,-112 trp1-Δ63 Δsui2, p919[SUI2, URA3]) after plasmid shuffling . For the intracellular localization of Aft1, BY4741, gcn2Δ, and GCN2c cells were transformed with the high-copy plasmid pRS426 expressing Aft1 tagged with green fluorescence protein (GFP) under control of its endogenous promoter .
Media and assays for cell growth
SD medium contained 0.7% yeast nitrogen base without amino acids (YNB; ammonium as the sole nitrogen source, Laboratorios Conda, Madrid, Spain), and 2% glucose was buffered with 50 mM MES (2-(-N-morpholino)ethanesulfonic acid) taken to a pH of 6.0 with Tris base and then supplemented with the auxotrophic requirements of the strains. To produce SGal medium, glucose was substituted by galactose. For low-iron media the indicated amounts of the iron chelator bathophenanthroline disulfonate (BPS) were added. When specified, normal YNB was substituted by YNB without iron (Formedium, Hunstanton, Norfolk, U.K.) to strengthen iron starvation. Amino acid supplementation was effected with 1 g/l of casamino acids, a mixture of amino acids obtained by acid hydrolysis of casein (see Table S1 for the composition of the employed product from Laboratorios Conda, Madrid, Spain). This is the minimum concentration providing maximum increase in yeast growth on SD medium. YPD medium contained 2% glucose, 2% peptone, and 1% yeast extract (Laboratorios Conda, Madrid, Spain). The growth temperature was 28°C.
Cell concentration was measured by absorbance at 660 nm with a UV-3100PC spectrophotometer of VWR (Radnor, Pennsylvania, U.S.A.). This corresponded to ‘attenuance', with readings mostly due to scattering of cells and it was only proportional to the cell concentration for values below 0.6. One unit was equivalent to 1.1 mg fresh weight of cells/ml. In most experiments, pre-cultures were grown overnight in SD medium to the stationary phase (absorbance 2.0–2.5). For growth assays in very low-iron medium, pre-cultures were grown overnight in SD medium without iron (prepared with YNB without iron). Mid-exponential cultures were prepared by inoculation with stationary cultures and collected with absorbance 0.4–0.5. Late-exponential cultures were collected with absorbance 1.1–1.2. For assays in SGal medium, pre-cultures were grown in SD medium. Cells were also grown in YPD medium to the stationary phase (absorbance 8–11).
Growth was monitored in a continuous way with the Bioscreen C microbiological workstation (Oy Growth Curves Ab Ltd, Helsinki, Finland). Starting absorbance readings were 0.02 and recorded values using a wideband filter (420–580 nm) were corrected for the nonlinearity at higher cell densities . Data are the mean of three biological repetitions, with four technical repetitions each.
mRNA quantification by real-time RT-PCR
Total RNA of cells grown in SD medium to the mid-exponential phase was extracted as described . RNA was further purified using a NucleoSpin RNA II kit (Macherey-Nagel, Düren, North Rhine-Westphalia, Germany), and 1 µg RNA was reverse transcribed using the Maxima first-strand cDNA synthesis Kit for RT-qPCR (Thermo Fisher Scientific, Waltham, Massachusetts, U.S.A.). PCR amplifications were performed on the first cDNA strand corresponding to 30 ng of total RNA in a final volume of 20 µl. Quantitative (real-time) PCR (qRT-PCR) was performed using an Applied Biosystems 7500 Real-Time PCR System (Thermo Fisher Scientific, Waltham, Massachusetts, U.S.A.) with the 5× PyroTaq EvaGreen qPCR Mix Plus (ROX) (Cultek, Madrid, Spain). PCR amplification specificity was confirmed with a heat-dissociation curve (from 60 to 95°C). Efficiency of the PCR was calculated and was similar for the different samples. Gene UBC6 was selected as the internal standard  and relative mRNA abundance was calculated using the comparative ΔCt method . Data are the mean of three biological repetitions, with four technical repetitions each. The sequence of primers used for cDNA amplification is given in Table S2.
Determination of intracellular iron concentration
Cells were grown in 150 ml of either normal SD medium or low-iron SD medium (with 20 µM BPS added) to the mid-exponential phase. Cells were collected and washed twice by centrifugation with an ice-cold solution of 50 mM Tris–HCl (pH 6.5), and 10 mM EDTA. Cell pellets were digested in 1 ml of a 5 : 2 mixture of nitric and perchloric acids for 1 h at 80°C . After digestion, samples were diluted to 4.0 ml with MilliQ water, filtered and then flamed in a GBC SensAA atomic absorption spectrometer (GBC Scientific Equipment, Braeside, Victoria, Australia). All samples were measured three times (three technical repetitions) using an iron hollow cathode lamp (SMI-LabHut, Gloucester, U.K.), and dataare the mean of three biological repetitions.
Assay of succinate dehydrogenase
Cells were grown during 48 h in YPD to reach the stationary phase with maximum respiratory induction. The absorbance was 10.5–11.0 for strains RS-86 and RS-88, 8.5–9.0 for BY4741 and its gcn2Δ derivative and 5.5–6.0 for the GCN2c mutant. Cells from 50 ml cultures were centrifuged 5 min at 800 g and resuspended with 1 ml extraction buffer [50 mM Tris–HCl (pH 8.0), 0.6 M sorbitol, 0.1 M KCl, 5 mM EDTA, 5 mM dithiothreitol] and 40 µl of 25-fold concentrated protease inhibitors cocktail (Roche Applied Science, Penzberg, Germany). Homogenization was effected after addition of 2 ml of glass beads (BioSpec Products, Bartlesville, OK, U.S.A.; 0.5 mm, precooled at −20°C) and shaking in a vortex at top speed during 3 min. Liquid was recovered with a pipette and centrifuged for 5 min at 800 g and 4°C to remove debris. A mitochondrial fraction was pelleted from the supernatant by centrifugation for 20 min at 15.000 g and 4°C and resuspended in 200 µl buffer [10 mM imidazole-HCl (pH 7.0), 0.6 M sorbitol, 1 mM EDTA, 1 mM dithiothreitol]. Protein was determined by the method of Bradford . The colorimetric assay of succinate dehydrogenase by reduction of 2,6-dichlorophenolindophenol was performed as described . Activity units corresponded to nmol × min−1 × mg protein−1.
Strains were grown overnight in SD medium with 20 µM BPS and 5 µl samples of live cells were mounted onto slides for microscopy. Fluorescence microscopy was performed using a Leica DM5000 B microscope (Leica Microsystems, Wetzlar, Germany) with 40× objective lens and GFP filter. Images were processed using the NIS elements F software (Nikon, Minato, Tokyo, Japan). Four independent experiments were performed and five fields per every condition were examined in each experiment. The percentage of cells showing nuclear fluorescence of Aft1-GFP was quantified by scoring more than 250 cells/condition.
Amino acid supplementation represses iron regulon genes by an Aft1-dependent mechanism
Some of the enzymes of amino acid biosynthesis require iron in the form of ISCs and in yeast these are Aco1, Glt1, Ilv3, Leu1, Lys4, and Met5 . Except for Glt1, all of them are essential for the biosynthesis of certain amino acids such as glutamate (Aco1), isoleucine and valine (Ilv3), leucine (Leu1), lysine (Lys4) and methionine and cysteine (Met5) . Amino acid biosynthesis in media with ammonium as the sole nitrogen source (SD) imposes a higher demand for cellular iron than in media supplemented with amino acids. Under the latter conditions all amino acid biosynthetic enzymes are repressed at the transcriptional level by the GeIp  and cells would require less iron. Regulation of iron uptake would be needed because excess of intracellular iron results in the generation of toxic ROS. Therefore, we investigated if amino acid supplementation changes the expression of iron regulon genes.
Expression of 10 representative iron regulon genes [11–13] was analyzed and, as indicated in Figure 1A, we found a general repression by amino acid supplementation, with inhibitions of expression ranging from 25% (ARN2) to 75% (FET3 and SIT1).
To investigate if Aft1 mediates this down-regulation by amino acids, gene expression in aft1Δ cells was also examined. As already described, expression of iron regulon genes is greatly decreased in this mutant [11–13], with FET3 and SIT1 as the most Aft1-dependent genes (Supplementary Figure S1). In the aft1Δ mutant, repression by amino acid supplementation of the six genes investigated was much less than in control cells (Figure 1B): FIT2, ARN1, ARN2, and CTH2 have no significant repression by amino acids, while FET3 and SIT1 were repressed by only 30 and 40%, respectively (Figure 1B). This partial but significant effect of Aft1 may be due to the presence of Aft2  and perhaps other TFs acting on iron regulon genes. These results indicate that amino acids trigger a down-regulation of the iron regulon by a mechanism dependent on Aft1.
The iron regulon is down-regulated in response to amino acid supplementation in an AFT1-dependent manner.
As the iron regulon genes are mostly concerned with iron uptake [11–13], it could be expected that amino acid supplementation would cause a decrease in cellular iron content. As described in Figure 1C, iron content of cells grown in medium supplemented with amino acids was 20% lower than in cells grown without amino acid supplement. The majority of this decrease was Aft1-dependent, because in the aft1Δ mutant the small decrease in iron content caused by amino acid supplementation was not statistically significant. We conclude that the Aft1-dependent repression of iron regulon genes is the cause of the decrease in cellular ion content induced by amino acid supplementation.
Gcn2 activates expression of the iron regulon by a mechanism independent of total cellular iron
The GeIp is a positive regulator of amino acid biosynthesis that is feedback-inhibited by amino acids through inactivation of Gcn2 and decreased phosphorylation of eIF2α . To investigate if Gcn2 participates in the regulation of iron regulon genes, expression of representative genes of this regulon was analyzed in a gcn2Δ null mutant and in a strain expressing a Gcn2 dominant allele (GCN2c) that leads to constitutive activation of the pathway [37,38]. As indicated in Figure 2A, gcn2Δ cells showed expression values from 36 to 70% of wild type in six of the eight Aft1-regulated genes investigated. In contrast, expression of seven of these genes increased from 1.4- to 8.5-fold in the GCN2c strain when compared with wild-type cells (Figure 2C). Differences in iron regulon expression cannot be explained by changes in the total iron content because there are no significant differences in this parameter between wild type and either gcn2Δ or GCN2c mutants grown under normal conditions (Figure 2E, left block).
Gcn2 controls the expression of the iron regulon.
When yeast strains were cultured in low-iron medium (SD supplemented with 20 µM BPS), cells of the gcn2Δ strain had less iron content than the wild type, whereas cells of the GCN2c strain contained more iron than the wild type (Figure 2E, right block), in accordance with the expression of iron transporters from the iron regulon (Figure 2B,D). Under these low-iron conditions, gcn2Δ cells showed a more uniform decrease in expression of all genes investigated, with values from 50 to 70% of wild type (Figure 2B). On the other hand, the effect of the GCN2c mutation on expression of Aft1-dependent genes is attenuated, with fold-change values from 1.1 to 3.5 (Figure 2D). This is expected if Gcn2 were an activator of Aft1 because, under low-iron conditions, this TF may be close to maximum activity [11,12]. These results suggest that differences of iron content in low-iron medium are the consequence rather than the cause of differential expression of the iron regulon in gcn2Δ and GCN2c mutants. Therefore, Gcn2 is a positive regulator of expression of the iron regulon by a mechanism independent of total cellular iron.
If the repressing effect of amino acids on the iron regulon is mediated by the GeIp, amino acid supplementation in the gcn2Δ strain should not cause repression, and as shown in Supplementary Figure S2, there is no repression by amino acids in this mutant. Interestingly, the expression of most iron regulon genes investigated is increased by amino acid supplementation in the gcn2Δ mutant. This suggests operation in the gcn2Δ mutant, but not in the wild type, of an unknown pathway activated by amino acids that positively regulates the iron regulon. The nature of this system is currently under investigation.
Phosphorylatable eIF2α, but not Gcn4, is required for expression of iron regulon genes
We have tested if expression of iron regulon genes is affected by other components of the Gcn2 pathway, such as eIF2α and Gcn4 . As indicated in Figure 3A, iron regulon genes are down-regulated from 32 to 74% in a strain (SUI2-S51A) carrying a mutation in the phosphorylation site of Sui2/eIF2α. This indicates that activation of expression of iron genes by Gcn2 operates at the translational level via phosphorylation of eIF2α. The classical downstream component of the GeIp is the TF Gcn4, which induces expression of genes involved in amino acid, purine and vitamin biosynthesis, amino acid transport, autophagy, and peroxisomal and mitochondrial proteins [2,4]. However, analysis of the gcn4Δ mutant strain revealed that lack of this TF does not lead to down-regulation of Aft1-dependent genes, as could be expected if the activation signal by Gcn2 required Gcn4. The expression of some iron regulon genes (FET3, FTR1, SIT1, CTH2) is not affected by the gcn4Δ mutation, while some other genes (FIT2, FIT3 and ARN3) are up-regulated in gcn4Δ cells (Figure 3B). This can be explained because the gcn4Δ mutant has some amino acid deficiencies  that would activate Gcn2 and therefore Aft1.
Phosphorylatable eIF2α, but not Gcn4, is required for expression of the iron regulon.
Constitutive activity of Gcn2 increases the nuclear localization of Aft1
Transcriptional activation of iron regulon genes requires Aft1 translocation from the cytosol to the nucleus, where it binds its target promoters [24–26]. We have investigated Aft1 localization in growing cells from wild type, gcn2Δ and GCN2c mutants transformed with a plasmid expressing GFP-tagged Aft1. In GCN2c cells, Aft1-GFP was detected in the nucleus of 50% of the cells, whereas only 22% of wild-type cells showed this nuclear localization (Figure 4A,B). This is in agreement with our observation of higher expression of iron genes in the GCN2c strain (Figure 2C,D). Cells from the gcn2Δ mutant and from the wild type supplemented with amino acids exhibited less nuclear localization of Aft1 than wild-type control. These results suggest that Gcn2 activation induces iron regulon genes by promoting Aft1 nuclear localization, as described during iron starvation [24–27].
Gcn2 positively regulates the nuclear localization of Aft1.
Succinate dehydrogenase activity is negatively regulated by the Gcn2 pathway
Sensing of intracellular iron by Aft1/Aft2 TFs is dependent on the biosynthesis of mitochondrial ISCs, a complex parameter that can be estimated from the activity of mitochondrial enzymes containing essential ISCs such as aconitase [30,31], a mitochondrial matrix marker , or succinate dehydrogenase, an internal membrane marker . We discarded aconitase because the GeIp, via TF Gcn4, is a positive regulator of expression of the ACO1 gene , and this complicates the interpretation of activity assays in mutants of this pathway. We have tested the hypothesis that the GeIp inhibits formation of ISCs, by measuring the activity of succinate dehydrogenase in mutants of this pathway. For these experiments, we used YPD medium because respiratory adaptation after glucose exhaustion in stationary cultures is very poor in SD medium. As indicated in Figure 5, the mutant with high constitutive kinase activity of Gcn2 (GCN2c) has only 4% of the succinate dehydrogenase activity of wild type, a very dramatic reduction. On the other hand, neither gcn2Δ nor SUI2-S51A mutants have significant differences in succinate dehydrogenase activity with their respective wild types. This can be explained because in amino acid-rich YPD medium Gcn2 activity is already inhibited. Although direct measurements of ISCs are still needed, these results suggest that the GeIp is a negative regulator of the biosynthesis of these important catalytic and regulatory molecules.
Gcn2 strongly inhibits succinate dehydrogenase activity.
Gcn2 inhibits growth in media with high requirement for Fe–S clusters
A prediction from the hypothesis that the GeIp inhibits biosynthesis of ISCs would be that, under conditions of strong iron starvation, growth would be inhibited by Gcn2 activity, especially in the GCN2c mutant with high constitutive activity of this kinase. As indicated in Figure 6A, under normal growth conditions (SD medium) this mutant has a small growth defect , with a longer lag phase and reduced growth rate as described . This has been attributed to general inhibition of translation initiation . In SD medium made with YNB without iron and supplemented with 10 µM BPS (very low-iron medium), this growth defect is exacerbated and the GCN2c mutant is unable to grow (Figure 6A). In agreement with these results, we observed that the GCN2c mutant is unable to grow in synthetic medium with galactose as the carbon source in place of glucose (Figure 6B). Galactose metabolism is mostly respiratory , and respiration requires large amounts of iron-based cofactors, including ISCs [9,53].
Gcn2 activity inhibits growth in media with high requirement of biosynthesis of Fe–S clusters.
The gcn2Δ mutant also has a growth defect in normal SD medium (longer lag phase and reduced growth rate), probably because of inhibition of amino acid biosynthesis . However, in very low-iron medium it grows similarly to the wild type (Figure 6C). Similar results were obtained with the SUI2-S51A mutant (Supplementary Figure S3).
These results support the hypothesis that the GeIp inhibits biosynthesis of mitochondrial ISCs.
The cross-talk between iron and amino acid homeostasis identified in the present work is schematized in Figure 7. Regulation occurs at the transcriptional level and excess of amino acids inhibit expression of iron transport genes mediated by Aft1. The mechanism is based on the known inhibition of the GeIp by amino acids and the proposed inhibition of mitochondrial ISCs by the GeIp. In summary, excess of amino acids would generate a signal of iron sufficiency that inhibits Aft1. As amino acids repress expression of biosynthetic enzymes containing ISCs, concomitant repression of iron transporters would contribute to prevent intracellular iron from reaching toxic levels by ROS production. This connection between iron and amino acid homeostasis operates as an anticipatory safeguard: Aft1 would be inhibited by amino acids before reduced synthesis of ISC-containing enzymes could cause sufficient iron accumulation to inhibit this TF.
Scheme of the proposed mechanism that connects the homeostasis of iron and amino acids.
The mechanism of inhibition of mitochondrial ISCs by the GeIp can only be speculated on. As phosphorylation of eIF2α is needed but Gcn4 is not involved, it expands the list of Gcn4-independent activities of the GeIp regulatory system . GeIp could activate translation of mRNAs encoding component(s) of the ISC assembly machinery. Alternatively, phosphorylated eIF2α could have a direct effect on ISC assembly. There are, however, no examples of eIF2α activities outside ribosomal protein synthesis [2,3].
In addition to amino acid supplementation, another situation where iron requirements are decreased is during growth with high glucose concentrations. Under these conditions, energy metabolism is mostly fermentative, and respiratory enzymes containing heme and ISCs are repressed . Glucose, through the Tpk2 protein kinase, represses expression of iron transporter genes , and this is also an anticipatory safeguard before excess iron is accumulated to inhibit Aft1. Although the mechanism was not investigated, it does not seem plausible that glucose, a general repressor of mitochondrial activities, would promote the formation of mitochondrial ISCs. Direct inhibition of Aft1 by glucose-activated Tpk2 seems more likely.
A recent study with Arabidopsis thaliana  has shown cross-regulation between iron and sulfur uptake systems. Sulfate transporter SULTR1;1 is up-regulated by sulfur deficiency but down-regulated by iron deficiency. Similarly, iron transporter IRT1 is up-regulated by iron deficiency but down-regulated by sulfur deficiency. This is interpreted as an adaptation to the lowered availability of the partner nutrient for synthesis of ISCs. Our results suggest that ISCs may participate in this cross-regulation.
A final remark is that cross-talks between metabolic and regulatory pathways are becoming important for the understanding of cellular physiology  and for the proper engineering of novel metabolic pathways into existing metabolic networks . Most metabolic pathways are interconnected [1,58] and the mechanisms of these cross-talks would be a field of intense research in the future. Filling this gap in knowledge will be important in biomedicine because of the growing appreciation for the role of altered metabolic regulations in disease .
M.C.-M., M.D.P. and H.B. performed experiments. S.A., M.A.N. and R.S. conceived and directed the project, and M.C.-M. and R.S. wrote the paper, with input from all the authors.
This work and the pre-doctoral contract of Marcos Caballero-Molada was funded by a grant to Fertinagro Nutrientes SL (Teruel, Spain) of CDTI (Centre for the Development of Industrial Technology; Madrid, Spain) with identification number IDI-20110197.
We thank Dr Jerry Kaplan (University of Utah, Salt Lake City, Utah, U.S.A.) for high-copy plasmid pRS426 expressing Aft1 tagged with GFP under control of its endogenous promoter.
The Authors declare that there are no competing interest's associated with the manuscript.