Iron Deficiency and Recovery in Yeast: A Quantitative Proteomics Approach

Abstract

Iron is an essential element for life, as it is critical for oxygen transport, cellular respiration, DNA synthesis, and metabolism. Disruptions in iron metabolism have been associated with several complex diseases like diabetes, cancer, infection susceptibility, neurodegeneration, and others; however, the molecular mechanisms linking iron metabolism with these diseases are not fully understood. A commonly used model to study iron deficiency (ID) is yeast, Saccharomyces cerevisiae. Here, we used quantitative (phospho)proteomics to explore the early (4 and 6 h) and late (12 h) response to ID. We showed that metabolic pathways like the Krebs cycle, amino acid, and ergosterol biosynthesis were affected by ID. In addition, during the late response, several proteins related to the ubiquitin-proteasome system and autophagy were upregulated. We also explored the proteomic changes during a recovery period after 12 h of ID. Several proteins recovered their steady-state levels, but some others, such as cytochromes, did not recover during the time tested. Additionally, we showed that autophagy is active during ID, and some of the degraded proteins during ID can be rescued using KO strains for several key autophagy genes. Our results highlight the complex proteome changes occurring during ID and recovery. This study constitutes a valuable data set for researchers interested in iron biology, offering a temporal proteomic data set for ID, as well as a compendium the proteomic changes associated with episodes of iron recovery.

Graphical Abstract

INTRODUCTION

Iron is an essential element for all living organisms because of its ability to donate and accept electrons. It is a critical element for many biological processes, such as oxygen transport, cellular respiration, DNA synthesis, and metabolism. Cellular iron levels must be highly regulated because free iron can generate reactive oxygen species, thereby resulting in cell damage.¹

A healthy human body contains about 4 g of iron distributed as follows: hemoglobin 65–80%, ferritin 20–30%, myoglobin 5–15%, and transferrin 0.1–0.5%. Iron can also be bound to numerous enzymes where it functions as a critical redox cofactor.² The systemic levels of iron are controlled by hepcidin, which functions by regulating ferroportin (main iron exporter) at the cell membrane. The binding of hepcidin to ferroportin promotes its degradation, resulting in the inhibition of iron release. The hepcidin/ferroportin axis is the main regulator of intestinal iron absorption as well as macrophage iron release, which supports the recycling of iron after phagocytosis of damaged red blood cells.¹

Iron deficiency (ID) can result from both physiological and pathological factors. The physiological factors include periods of increased demand during rapid growth as well as inadequate supply. Pathological causes include gastrointestinal stress (infection, cancer, gastritis, etc.), obesity, and celiac disease.^3,4 ID is the most common worldwide nutritional deficiency, affecting approximately 25% of the world’s population. ID is also associated with increased risk of maternal and neonatal mortality, impaired development in children, reduced physical performance and work productivity in adults, as well as cognitive decline in the elderly. One of the Global Nutrition Targets 2025 is aimed to achieve a 50% reduction of anemia in women of reproductive age.⁵

Tissue-specific ID is often associated with severe systemic changes. In a mouse model of skeletal muscle-specific iron deficiency, a metabolic catastrophe occurs due to liver dysfunction.⁶ Furthermore, in the brain, ID induces neurodegeneration,⁷ suggesting that the role of iron in supporting normal physiology is organ-specific. When ID occurs during critical periods of life, the consequences may not be mitigated even with iron supplementation therapy.^8,9 For example, fetal iron deficiency produces long-term changes in synaptosomal substrates that remain altered even in adulthood.¹⁰ As expected, ID is associated with a number of serious pathological states, such as susceptibility to infection, heart failure, chronic kidney disease, cancer, Alzheimer’s and Parkinson’s diseases, and diabetes.¹¹ Importantly, the relationship between iron and those diseases is not fully understood.

One commonly used model to study ID and its biological consequences is the yeast Saccharomyces cerevisiae. In response to iron deficiency, S. cerevisiae upregulates the expression of approximately 30 genes (“the iron regulon”), mainly by the activation of the transcription factors AFT1/2. These upregulated genes are associated with iron acquisition, intracellular iron transport and metabolic remodeling.¹² ID-dependent activation of AFT1/2 in yeast results in the degradation of several mRNAs involved in mitochondrial respiration and Krebs cycle, as well as ergosterol, amino acid, and biotin biosynthesis. This mRNA degradation is also regulated by the coordinated action of two members of the iron regulon, Cth1/2.^13,14 This profound metabolic remodeling enables cells to survive under iron deficient conditions. In addition, little is known regarding the proteomic adaptations to fluctuating iron levels—periods of ID followed by iron replenishment (iron recovery). Here, we explored the proteomic adaptations to acute (4 and 6 h) and chronic (12 h) ID episodes, as well as the proteomic consequences of episodes of ID followed by a recovery, using quantitative proteomics. Finally, we showed that several proteins degraded during ID can be rescued using KO strains for several autophagy related genes.

METHODS

Reagents

TMT reagents, BCA protein concentration kit, trypsin and High-Select Fe- NTA phosphopeptide enrichment kit were from ThermoFisher Scientific (Rockford, IL). StageTip Empore C-18 material was from 3 M (Saint Paul, MN). SepPak cartridges (100 mg) were from Waters (Mildford, MA). Lys-C protease was from Wako (Boston, MA). Water and organic solvents were from J.T. Baker (Center Valley, PA). cOmplete protease and PhosStop phosphatase inhibitors were from Millipore-Sigma (Saint Louis, MO). Yeast synthetic complete media was from Sunrise Science (San Diego, CA), bathopenanthrolinedisulfonic acid (BPS) was from Sigma (St. Louis, MO). Unless otherwise noted, all the other chemicals were from ThermoFisher Scientific (Waltham, MA).

Yeast Growth and Sample Processing

The wild-type S. cerevisiae strain for this study was the haploid MATalpha BY4742. Starter cultures were inoculated with a patch of yeast cells and incubated overnight (30 °C, 230 rpm). The next day, synthetic complete medium plus 2% glucose complemented with 100 μM BPS (a specific iron chelator) was inoculated with a starting OD_600nm = 0.2, incubated at 30 °C and 230 rpm (cultures were performed in triplicated for each time point). After 4 and 6 h, a sample equivalent to 20 OD was obtained, cells were pelleted by centrifugation and washed twice with PBS and stored at −80 °C. At 6 h, to avoid the combined effect of iron deficiency plus nutrient limitation, the cell cultures were diluted to 0.75 OD_600nm using fresh medium and the cells were cultured for another 6 h. For the recovery experiment, cells were cultured for 12 h in synthetic complete media plus 2% glucose and 100 μM BPS (at 6 h the media was changed as above), cells were pelleted by centrifugation, washed with synthetic complete media and were used to inoculate synthetic complete media plus 2% glucose (OD_600nm = 0.75), and sampling was performed as above at 1 and 4 h.

To explore the role of some ATG genes on protein degradation during ID. A WT strain (BY4742), as well as the KO strains for ATG8, ATG32, ATG39, or ATG40 (haploid MATalpha collection) were grown for 12 hours on synthetic complete media complemented with 100 μM BPS for 12 h (at 6 h the media was changed as above).

The sample preparation was performed as reported previously.^15,16 Briefly, yeast pellets were lysed by bead-beating in lysis buffer (8 M urea, 200 mM EPPS, pH 8.5) supplemented with protease and phosphatase inhibitors. Protein concentration was determined by BCA following manufacturer instructions, then proteins were reduced using 5 mM tris (2-carboxyethyl) phosphine (TCEP), at room temperature for 15 min and alkylated using 10 mM iodoacetamide (room temperature, 30 min in the dark), and excess of iodoacetamide was quenched using 10 mM dithiothreitol (room temperature, 30 min in the dark). The samples (100 μg) were precipitated using methanol–chloroform method and resuspended in 100 μL of 200 mM EPPS (pH 8.5), then digested overnight using LysC protease at a 100:1 protein-to-protease ratio. The next day trypsin was added at a 100:1 protein-to-protease ratio and incubated for 6 h at 37 °C.

After digestion, each sample (100 μg) was labeled using TMT11-plex reagents (in the case of the ATGs KO experiment TMTpro16 reagents were used). To ensure equal amount of total peptides in each fraction a label efficiency check was performed, then the reaction was quenched using 0.3% (v/v) hydroxylamine, combined into a single tube and desalted using a SepPak. Phosphorylated peptides were enriched using a High Select Fe-NTA column, according to the manufacturer’s instructions. The unmodified proteome was fractionated through basic pH reversed-phase chromatography into 96 fractions, consolidated into 24 and 12 fractions were analyzed by the mass spectrometer.

Mass Spectrometry Analysis

For the TMT11-plex experiments, mass spectrometry data were collected using an Orbitrap Fusion or Lumos mass spectrometer (ThermoFisher Scientific, San Jose, CA) coupled to a Proxeon EASY-nLC 1200 liquid chromatography (LC) pump (ThermoFisher Scientific, San Jose, CA). Peptides were separated on a 100 μm inner diameter microcapillary column packed with ≈40 cm of Accucore resin (2.6 μm, 150 Å, ThermoFisher Scientific, San Jose, CA). For each analysis, we loaded ≈2 μg onto the column and separation was achieved using a 2.5 h gradient of 7 to 27% acetonitrile in 0.125% formic acid at a flow rate of 550 nL/min. Each analysis used an SPS-MS3-based TMT method,^17,18 which has been shown to reduce ion interference compared to MS2-based quantification.¹⁹ The scan sequence began with an MS1 spectrum (Orbitrap; resolution 120 000; mass range 400–1400 m/z; automatic gain control (AGC) target 5 × 10⁵; maximum injection time 100 ms). Precursors for MS2/MS3 analysis were selected using a Top10 method. MS2 analysis consisted of collision-induced dissociation (quadrupole ion trap; normalized collision energy (NCE) 35; maximum injection time 150 ms). Following acquisition of each MS2 spectrum, we collected an MS3 spectrum using our previously described method in which multiple MS2 fragment ions were captured in the MS3 precursor population using isolation waveforms with multiple frequency notches.¹⁶ MS3 precursors were fragmented by higher-energy collisional dissociation (HCD) and analyzed using the Orbitrap (NCE 65; maximum injection time 150 ms; resolution was 50 000 at 200 Th).

For the TMTpro16 experiment, all data were collected on an Orbitrap Eclipse mass spectrometer coupled to a Proxeon NanoLC-1000 UHPLC. The peptides were separated using a 100 μm capillary column packed with ≈35 cm of Accucore 150 resin (2.6 μm, 150 Å; ThermoFisher Scientific). The data were collected using FAIMS/hrMS2.²⁰ Each fraction was eluted using a 90 min method over a gradient from 6% to 30% ACN. Peptides were ionized with a spray voltage of 3000 V. The Thermo FAIMS Pro device was operated with default parameters. No additional gas was used for desolvation. The DV circuitry was autotuned, which independently tunes each of the sine waves and phase shifts one of the waveforms by π/2 to assemble a bisinusoidal waveform with a high amplitude of −5000 V at a 3 MHz frequency. The fractions were analyzed in the mass spectrometer, using a method incorporating three CVs (CV = −40, −60, and −80 V).

Mass Spectrometry Data Analysis

Mass spectra were processed using a Sequest-based pipeline.²¹ Spectra were converted to mzXML using a modified version of ReAdW.exe. Database searching included all entries from the S. cerevisiae database (SGD project, March 2014). This database was concatenated with one composed of all protein sequences in the reversed order. Searches were performed using a 50 ppm precursor ion tolerance for total protein-level profiling. The product ion tolerance was set to 0.9 Da. These wide mass tolerance windows were chosen to maximize sensitivity in conjunction with Sequest searches and linear discriminant analysis.^21,22 TMT tags on lysine residues and peptide N termini (+229.163 Da for TMT11, +304.2071 for TMTpro16) and carbamidomethylation of cysteine residues (+57.021 Da) were set as static modifications, while oxidation of methionine residues (+15.995 Da) was set as a variable modification. Peptide-spectrum matches (PSMs) were adjusted to a 1% false discovery rate (FDR).²³ PSM filtering was performed using a linear discriminant analysis, as described previously,²¹ while considering the following parameters: XCorr, DCn, missed cleavages, peptide length, charge state, and precursor mass accuracy. For TMT-based reporter ion quantitation, we extracted the summed signal-to-noise (S/N) ratio for each TMT channel and found the closest matching centroid to the expected mass of the TMT reporter ion. PSMs were identified, quantified, and collapsed to a 1% peptide false discovery rate (FDR) and then collapsed further to a final protein-level FDR of 1%. Moreover, protein assembly was guided by principles of parsimony to produce the smallest set of proteins necessary to account for all observed peptides. Proteins were quantified by summing reporter ion counts across all matching PSMs, as described previously.²¹ PSMs with poor quality, MS3 spectra with more than 10 TMT reporter ion channels missing, or no MS3 spectra were excluded from quantification.²⁴ For the FAIMS/hrMS2 data, peptides with SN > 160, isolation specificity >0.7 and up to 14 empty channels, were included in the quantification. Each reporter ion channel was summed across all quantified proteins and normalized to assume equal protein loading across all samples. Then, each abundance measurement was normalized by dividing each sample by the average of the control samples (time 0 h), obtaining the fold changes relative to control samples. P-values were adjusted with the Benjamini–Hochberg method (FDR < 0.01 for protein and FDR < 0.05 for phosphorylation). All statistical analyses, hierarchical clustering and data visualization were performed using GraphPad v8, Perseus²⁵ or Microsoft Excel. For the autophagy experiment, we could not validate the deletion of ATG32 strain, thus these TMT channels were not considered for further analysis.

GO Term Analysis

The server PantherGO classification system was used²⁶ to identify the relevant pathways affected by iron deficiency and further iron recovery and the whole yeast genome data set was employed as background. Multiple hypothesis testing correction was performed by the Bonferroni method.

Data Access

The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE²⁷ partner repository with the data set identifier PXD023484.

RESULTS AND DISCUSSION

ID is the most common worldwide nutritional deficiency. ID has been associated with a number of pathological conditions, however the mechanistic details thereof are not fully understood. The goal of this study is to explore the proteomic and phosphoproteomic changes associated with ID and the subsequent recovery using S. cerevisiae as a model system. To that end, we studied the proteome dynamics of ID by culturing yeast in iron depleted media (BPS treated). Cells were harvested at 4 and 6 h, which should model an early response to ID and 12 h, which we treated as a late response. We used our reported SL-TMT protocol¹⁵ which enabled us to accurately quantify more than 3600 proteins across 11 samples in a single experiment, thereby reducing technical variation (Figure 1A). After 12 h, 1890 proteins had significantly changed (FDR < 0.01) in comparison with the iron-sufficient cells (control cells). Those proteins were associated with biological processes such as nucleocytoplasmic transport, cellular amino acid biosynthetic processes and ribosome biogenesis. The most affected cellular components were the mitochondria, endoplasmic reticulum, nuclear pore, large ribosomal subunit, nucleolus and cell wall. When we compared 12 h vs 4 h of ID, 1321 proteins had a significant change (FDR < 0.01), of which 821 were decreased at 12 h and 500 were increased. The upregulated proteins were involved in the regulation of protein stability, cellular response to unfolded proteins, actin polymerization, and cellular response to oxidative stress, as well as ubiquitin-dependent catabolic processes. Conversely, the downregulated proteins were associated with rRNA processing, cytoplasmic translation and ribonucleoprotein complex export from the nucleus. Finally, principal components analysis showed that iron repleted cells were clearly separated from the iron deficient cells (Figure 1B).

Figure 1.

Overview of the iron deficiency (ID) experiments in yeast. (A) ID workflow. Overnight cultures were transferred to low iron media (synthetic complete media with 2% glucose complemented with 100 μM BPS) for 4, 6, or 12 h, and the samples were processed according to the SL-TMT protocol. (B) Principal components analysis. (C) Protein abundance changes of some members of the iron regulon. (D) Examples of protein abundance changes.

Next, we investigated the abundance of several proteins that are involved in pathways known to be affected by iron. These include members of the iron regulon, proteins involved in iron sulfur (FeS) cluster biogenesis, and a number of FeS cluster-containing proteins. In the case of the iron regulon (Figure 1C and Table S1), all the identified proteins were upregulated during ID; however, their kinetics were slightly different. For example, proteins like Fre1/6, Arn1/2/3, and Ccc2 reached their maximum values at 4 h and then decreased while proteins like Fet5, Hmx1, Cth2, and Atx1 peaked at 4 h and maintained the same level throughout the duration of the experiment, suggesting that protein expression during early and late ID is dynamic. Next, we turned our focus to proteins known to be involved in the FeS cluster biogenesis pathway (Figure 1D and Table S1). The mitochondrial iron transporters Mrs3/4 were highly upregulated during the time course, likely reflecting an attempt to sequester the limited iron supply to mitochondria. Interestingly, the expression of the Nfs1, the essential cysteine desulfurase required for FeS cluster biogenesis, and its binding partner, Isd11 were less affected. Furthermore, Grx5 and Isu1 were decreased, while Atm1, Nbp35, and Tah18 were increased. Having observed evidence of defects in FeS cluster biogenesis, we explored the protein levels of several FeS cluster proteins and, consistent with previous reports,²⁸ most of the proteins we identified were decreased, including Sdh2 and Aco1/2 (Krebs cycle), Ilv3 (branched chain amino acids synthesis), Leu1 (leucine biosynthesis), Met5 (cysteine and methionine biosynthesis), Glt1 (glutamate biosynthesis), and others. Interestingly, this was not the case for all FeS cluster proteins. Indeed, Grx7, Rad3, Pol1/2/3 seemed to be unaffected by ID at earlier time points. This finding suggests, that during episodes of ID, not all the FeS cluster proteins will be degraded with the same kinetics or at the same time. This likely reflects a greater dependence on certain proteins for yeast cell survival.

It is important to note that our data set recapitulated most of the previously reported physiological changes known to occur in response to ID, therefore validating our experimental approach. For example, a 14-fold accumulation of 2-isopropylmalate has been reported in iron-deficient cells.²⁸ This metabolite is the substrate of Leu1, which in our data set has a negative 0.5-fold change at 4 h and negative ~5-fold change at 12 h (Table S1). Another example is ergosterol biosynthesis. Erg3/6/25 are known to increase during ID²⁸ and in our data set those proteins have fold changes ranging from 1.5 to 6 and most of the enzymes involved in the pathway were affected by ID (Supporting Figure S1). This information highlighted that the cellular proteome will be remodeled in a time-dependent manner during an episode of iron deficiency, thereby having a broad repertoire of changes during an early or late response.

Considering that the ID response changes with time and that some changes induced by ID are not recovered after a period of supplementation,^8-10 we hypothesized that after a chronic ID episode (12 h), some proteins will recover to their normal steady-state levels while others will remain altered. As such, we performed an additional SL-TMT experiment in which S. cerevisiae cultures were grown on iron depleted media for 12 h. These cells were then washed and transferred to iron sufficient media for 1 and 4 h (Figure 2A, Table S2).

Figure 2.

Overview of iron deficiency (ID) and recovery TMT experiment. (A) Workflow overview. Overnight cultures were transferred to low iron media for 12 h (media change at 6 h). The cells were then washed and transferred to regular iron media (synthetic complete media plus 2% glucose) for 1 and 4 h. (B) Principal components analysis and (C) hierarchical clustering showing the tight association between biological replicates and the clear separation between iron sufficient, iron deficient, and recovered cells. (D) Relative abundance of some members of the iron regulon.

In this experiment, we quantified over 4000 proteins. As expected, after 4 h of recovery, the proteome was partially reverted to normal steady state levels (Figure 2B and 2C). We explored the protein abundance of several members of the iron regulon and found that most of the detected proteins were upregulated during ID, and returned to or trended toward their normal steady state levels (Figure 2D), suggesting that the intracellular iron levels are being replenished. In further support of this observation, several FeS cluster proteins recovered to their steady state levels, including Lys4, Aco2, Lip5, and Bio2 (Figure 3). These data highlight that after 4 h of recovery, several proteins whose abundances were regulated by iron availability returned to their steady state levels. As mentioned above, iron and cellular energy production are closely linked. Thus, we explored the protein levels of several metabolic pathways. The Krebs cycle (Supporting Figure S2) is a critical pathway for cellular energy production and biosynthesis and relies on two FeS-dependent enzymes—aconitase and succinate dehydrogenase (SDH). The yeast genome encodes both a cytosolic (Aco2) and mitochondrial (Aco1) aconitase, both of which convert citrate to isocitrate. ACO1 gene disruptions lead to defects in cellular respiration, as well as a glutamate auxotrophy.²⁹ We observed that the Aco1/2 protein levels decreased approximately 20-fold in ID, but increased significantly after just 1 h of recovery and reached its normal steady-state level at 4 h of recovery. As expected, Glt1, a FeS cluster protein that catalyzes the synthesis of glutamate from l-glutamine and 2-oxoglutarate (alpha-ketoglutarate; a Krebs cycle intermediate), had a negative 12-fold change during ID. Similar to Aco1/2, Glt1 increased 8-fold after just 1 h of recovery, reaching its normal steady-state level after 4 h. The other FeS-dependent enzyme in the cycle is SDH, which is composed of four subunits (Sdh1–4) and also functions as complex II of the electron transport chain. SDH catalyzes the oxidation of succinate to fumarate and transfers the resulting electrons to ubiquinone.³⁰ SDH levels decrease ~15-fold in ID (Figure 3) and unlike Aco1/2 failed to recover during the course of the experiment (Figure 3). Both enzymes, Aco1/2 and SDH, are regulated by Cth1/2^13,31,32 under ID conditions, but Aco1/2 recovered its steady-state levels while SDH did not. This effect suggests that additional repression mechanisms were still affecting the synthesis or assembly of the SDH. For example, Cth2 has been shown to physically interact with Dhh1 during ID to promote the degradation of SDH4.³³

Figure 3.

Examples of changes in abundance for proteins with iron sulfur clusters as well as some involved in the Krebs cycle during iron deficiency and recovery.

Most of the enzymes involved in the pathway were also negatively regulated during ID, with the exception of isocitrate dehydrogenase (IDH). IDH is a NAD+-specific octameric enzyme composed of four subunits of Idh1 and four of Idh2. IDH catalyzes the oxidative decarboxylation of isocitrate to produce 2-oxoglutarate, which is a source of NADPH for the cell³⁴ and is the rate-limiting step of the pathway. In our study, IDH was the only enzyme of the cycle that increased in abundance (2-fold), potentially to maintain flux and energy production or redox balance through NADPH production.

In addition to the Krebs cycle, ID also impacted a number of other important metabolic pathways. For example, the glycolytic enzymes enolase (Eno1), glyceraldehyde-3-phosphate dehydrogenase (Tdh1–3), and fructose-bisphosphate aldolase (Fba1) were altered (Supporting Figure S3). In addition to glycolysis, several biosynthetic pathways including those responsible for the synthesis of ergosterol, aromatic and branched-chain amino acids, and cysteine and methionine were affected in most steps (Supporting Figures S1, S4-S6).

The link between iron and some metabolic processes is bidirectional. For example, amino acid supplementation affects the expression of some members of the iron-regulon due to the activation of AFT1 and eIF2α phosphorylation by Gcn2, but is independent of Gcn4.^35,36 On the other hand, several critical steps in amino acid biosynthesis are performed by FeS cluster proteins (i.e., Leu1), thereby highlighting the complex metabolic interactions during ID. Another interesting observation is associated with the differential protein expression during these metabolic transitions. Most of the steps involved in leucine and isoleucine biosynthesis were negatively regulated during ID and rapidly recovered to their steady state levels upon iron repletion (Supporting Figure S5).

Next, we explored our data to find proteins that were positively (group 1) or negatively (group 2) regulated by ID and returned to their steady-state levels after 4 h of recovery, as well as proteins that were affected by ID and failed to recover (group 3) during the time course of the experiment (Figure 4). To establish these groups, we first determined all proteins that had a log₂ fold change of ±1 upon depletion of iron. Next, this list was divided into two groups based on the response of each protein to iron refeeding (positive or negative change). The first two groups included proteins that returned to normal steady state levels within 4 h of exposure to iron. The third group was comprised of proteins that failed to recover normal steady state abundance during this period.

Figure 4.

Abundance profiles of proteins sensitive to iron deficiency. (A) Proteins with increased abundance during iron deficiency, recovering their steady-state levels after 4 h of culture in iron sufficient media. (B) Proteins with decreased abundance during iron deficiency, recovering their steady-state levels after 4 h of culture in iron sufficient media. (C) Proteins with decreased levels during iron deficiency which did not recover after 4 h. Several representative examples are shown below the protein profile plots.

Using these inclusion criteria, we found 123 proteins in group 1 (upregulated proteins) and 129 proteins in group 2 (downregulated proteins) during ID that returned to their basal levels after 4 h of recovery. Conversely, 103 proteins were assigned to group 3 (downregulated proteins during ID and remained at the same level during the recovery period). For the group 1, ~26% have not been characterized previously (proteins of unknown function), raising the possibility that several homeostatic mechanisms during ID and recovery could be regulated by uncharacterized proteins. In this group 1, we also found proteins associated with protein degradation pathways, like Otu1 and Ubp5/7/15/13 and Atg7 (Figure 4A). Proteins in group 2 were associated with DNA-binding transcription factor activity and RNA polymerase II proximal promoter sequence-specific DNA binding (Figure 4B). The group 3 was associated with energy production and cytochrome activity. Those findings demonstrated that not all the iron-requiring proteins recover to steady-state levels at the same rate in the recovery phase.

One of the main biochemical features of a nutritional deficiency is the reduction of general mRNA translation. For example, in the case of ID in yeast, the proteins Cth1/2 are responsible for repressing several metabolic genes involved in mitochondrial respiration or iron-related processes.^13,31,32 The translation repression observed during ID has been reported previously to be also partially due to the phosphorylation and activation of eIF2α by Gcn2 kinase.³⁶ In our data set, Gcn2 increased 2-fold during ID and recovered to its normal levels around 4 h of culturing cells on iron sufficient media. Gcn4 has been reported previously to be induced in response to eIF2α activation.³⁷ In our data set, Gcn4 had a negative 40-fold change; however, some phenotypes observed during ID are dependent on Gcn2-eIF2α but independent of Gcn4 activity.³⁵

Considering these vast proteomic changes, we explored the phosphorylation dynamics during ID and recovery (Figure 5, Table S3). As expected, ID episodes followed by recovery periods promoted dramatic phosphorylation changes. In total we observed a significant change (FDR < 0.05) in 1848 individual phosphorylation sites, of which 808 were upregulated and 1040 were downregulated (Table S3). Several sites that were up- or down-regulated during ID, reached their steady-state levels after 4 h of recovery (Figure 5A). Considering those changes, we explored the protein abundance of several kinases (Figure 5B), as expected, several kinases were upregulated during ID, and most of them recovered their basal levels after 4 h of recovery. For example, Ark1 and Rad53 were upregulated, and some of their substrates (Figure 5C) had increased phosphorylation levels. This was the case for Rad53 and its substrates: Nup2 (a nucleoporin involved in nucleocytoplasmic transport) and of Swi6 (a transcription factor involved in G1/S transition).

Figure 5.

Phosphorylation profiles during iron deficiency and recovery. (A) Phosphorylation sites that increased or decreased in abundance during iron deficiency and recovered their steady-state levels after 4 h of culture in iron sufficient media. (B) Heatmap showing protein expression changes for selected kinases during ID. (C) Examples of known phosphorylation sites on substrates of Ark1 and Rad53 kinases.

In total, we observed decreased abundances for more than 400 proteins during ID (some recovered their steady state levels, while others did not). This observation raised the possibility that some protein degradation pathways were active during ID. To explore this concept, we mined our data sets for proteasome and autophagy-related proteins³⁸ and found a consistent increase in most of the detected proteins (Supporting Figure S7). In different model organisms, iron deficiency induces autophagy, specifically mitophagy.^39-41 The consistent increase in most of the autophagy-related proteins suggest that this also occurs in yeast (Supporting Figure S7A). With regard to the proteasome, we found that some proteasome subunits and proteins belonging to the ubiquitin-proteasome system increased in abundance during ID and returned to steady-state levels at 4 h of recovery (Supporting Figure S7B). In addition, some proteasome subunits or accessory proteins were phosphorylated, like Rpn1/8/11 Pre1/8, etc. (Supporting Figure S7C). In yeast,^42,43 as well as in mammalian cells,^44-46 proteasome phosphorylation has been shown to self-regulate its activity. In mammalian cells, for example, PKA is a well-known activator and Ask1 is a negative regulator.⁴⁶ Some phosphorylation sites on the proteasome have been characterized, for example Rpn1-S361 phosphorylation which has been associated with decreased proteasome activity.⁴⁵

Those observations suggested that several protein degradation pathways were active during ID. To test this idea, we performed an SL-TMT experiment (Figure 6 and Table S4). A wild-type strain, as well as strains that were null for ATG8 (a protein essential for autophagosome biogenesis),⁴⁷ ATG39 (an autophagy receptor for perinuclear ER)⁴⁸ or ATG40 (autophagy receptor for cytoplasmic ER)⁴⁸ were cultured for 12 h in synthetic complete media complemented with 100 μM BPS. For comparison, we also included the wild type strain at time 0 h (basal level proteome). In this experiment, we quantified 4633 proteins. From those, 663 proteins had fold changes smaller than 0.7 in the BPS-treated group. From those proteins with decreased abundance, 182 were rescued or partially rescued by one or several KO strains (Figure 6A). The rescued proteins were associated with glutamine, glycogen or ATP metabolism, as well as they were related to the mitochondrial membrane or vacuolar lumen (Figure 6A). We noted that the rescued proteins followed three patterns (Figure 6B), specifically high, medium or low recovery. In the first category, we found proteins like Tpo2, Sps100, Gpx1, and Ygp1. These are fully recovered proteins having a higher abundance in the Atg8/39/40 strains that the WT at time 0 h. The medium recovery refers to a subset of proteins that were increased in abundance compared to the BPS control, but not to WT levels at time 0 h control, such as Tfs1, Sfk1, Aim17, or Pet9. Finally, the low recovery subset have a small increase compared to the BPS control, and these included Cue4 and Tom20. These results suggest that autophagy is not the only pathway responsible for the protein degradation observed during ID. This finding is also evidenced by a number of proteins that remained at very low levels in all the KO strains (Figure 6C). Interestingly, some of those proteins are FeS cluster-containing proteins like Aco1/2, Bio2, and Ilv3 which participate in vital cellular processes like the Krebs cycle, biotin and branched-chain amino acid synthesis. Also, in this same group, we identified transporters like Mep2 (ammonium permease), Thi7 (thiamine transporter), or Zrt1/3 (zinc transporters).

Figure 6.

Iron deficiency (ID) and recovery in yeast with gene deletions in the autophagy pathway. (A) WT and KO strains for ATG8, ATG39, and ATG40 were cultured for 12 h in synthetic complete media complemented with 100 μM BPS. For comparison the WT strain at time 0 h was included as control (basal levels). A) Plot of the 182 proteins with significantly decreased abundance in WT cells with ID. Many of these proteins are protected from degradation by deletions in the autophagy pathway—GO category enrichment is also shown. (B) Examples of proteins in three protection-pattern categories including “high”, “medium”, and “low” recoveries. (C) Plot of proteins and several examples that decreased in abundance and were found to be independent of autophagy.

Here we provide valuable data sets that can be mined by the community to explore novel aspects of ID. This resource, in combination with other previously reported data sets, improves our understanding of ID at a systems scale. For example, the yeast transcriptome cultured on low iron conditions has been described previously,²⁸ allowing us to correlate protein and mRNA responses. As expected, the overall correlation was low (Supporting Figure S8) with r = 0.18 (p-value <0.0001). However, proteins known to be under iron control (iron regulon members or FeS cluster-containing proteins) showed a higher correlation when comparing protein and mRNA levels. For example, Fre2/5, Ccc2, and Arn1 (iron regulon) showed both increased mRNA and protein. On the other hand, proteins like Sdh1/2, Glt1, Aco1/2 (FeS clusters), and Cyc1 (cytochrome) were downregulated at both the mRNA and protein levels. We also noted a subset of proteins for which their mRNA levels showed no change or even increased during ID, but the protein abundance was consistently found at lower levels during ID. This discrepancy could be explained by the activation of protein degradation machinery. For example, several decreasing proteins were rescued by one or more ATG family KO strains used in this study. However, this was not consistent for all the proteins in this category, suggesting that other mechanisms, such as degradation via the ubiquitin-proteasome system or the unfolded protein response, are responsible. Future studies are needed to fully understand protein turnover during ID and recovery.

CONCLUSIONS

The present study constitutes a deep and valuable data set which explores the proteomic changes during acute or late ID episodes. These data show how cells remodel their proteome during the recovery phase and highlight the complex and dynamic mechanisms that cells adapt to respond to iron availability. For example, with these data we can explore the connections between iron and amino acid metabolism and how dysfunction in one pathway produces changes in other processes.

Supplementary Material

FigS1-S8

Supporting Figure S1: Ergosterol biosynthesis during iron deficiency and recovery

Supporting Figure S2: Krebs cycle during iron deficiency and recovery

Supporting Figure S3: Glycolysis during iron deficiency and recovery

Supporting Figure S4: Aromatic amino acid biosynthesis during iron deficiency and recovery

Supporting Figure S5: Branched-chain amino acid biosynthesis during iron deficiency and recovery

Supporting Figure S6: Cysteine and methionine biosynthesis during iron deficiency and recovery

Supporting Figure S7: Autophagy and ubiq-uitin-proteasome system during iron deficiency and recovery

Supporting Figure S8: Protein mRNA correlation during ID (PDF)

TableS1

Table S1: Iron deficiency time course: Protein abundances for yeast cultures treated with BPS for different exposure times (XLSX)

TableS2

Table S2: Iron deficiency and recovery: Protein abundances for yeast cultures treated with BPS and then transferred to iron sufficient media (XLSX)

TableS3

Table S3: Iron deficiency and recovery: Phosphopeptide abundances for yeast cultures treated with BPS and then transferred to iron sufficient media (XLSX)

TableS4

Table S4: Proteome of ATG8/39/40 strains cultured for 12 h in low iron media (XLSX)