Ferritin Complements Iron Tolerance of Yeast Deficient in Vacuolar Iron Sequestration
For biogenic magnetization, significant amounts of magnetic compounds need to be formed inside the organism. This may be achieved by altering iron homeostasis either physiologically or genetically. Since ferrous iron is prone to oxidation to insoluble ferric iron, citric acid, a chelator of the ferric ion, can be included to prevent precipitation with no impact on biological availability. Wild-type yeast cells can grow at as high as 5 mM ferrous (Fe2+) ascorbate or 20 mM ferric (Fe3+) citrate (, wild type). Ferric citrate was less toxic and can thus be used to deliver iron to yeast without damaging the cells or forming precipitates in the media.
Yeast cells lack ferritin and sequester iron in their vacuoles. The vacuolar iron transporter Ccc1p plays a major role in iron sequestration, loss of which abolishes iron tolerance 
. Human ferritin genes consist of ferritin heavy chain FTH, ferritin light chain FTL, and the iron chaperone PCBP1 
. As described previously, the ccc1 knockout strain (ccc1Δ
) showed intolerance at 5 mM ferrous ion while as high as 20 mM ferric citrate is required to see intolerance of ccc1Δ
(), suggesting mitigated iron toxicity of ferric citrate. We found that single copy expression in yeast of the human ferritin gene set conferred iron tolerance to ccc1Δ
both in ferrous and ferric supplements (), indicating that ferritin efficiently sequesters iron in these conditions.
Altered Iron Homeostasis and Ferritin Confer Magnetism to the Cell
The four strains (wild type containing empty plasmid, ferritin-expressor, ccc1Δ
, and ccc1Δ
ferritin-expressor) were cultured in 20 mM ferric citrate liquid medium and tested for magnetization. The cell cultures were exposed to magnets and attraction was observed. Attraction of ccc1Δ
ferritin-expressor was detectable as early as 2 min after exposure. After 10 min, attraction of all strains became observable ( and Video S1
For quantitative characterization of the magnetic properties of the yeast cells, a superconducting quantum interference device (SQUID) was used. The cells were subjected to a measurement of their magnetic moment at 300 K at various magnetic fields to analyze field-dependent magnetization. Without ferric citrate supplementation all the four strains similarly exhibited negative values proportional to the applied field (, no iron supplemented), indicating that they are diamagnetic. As is the case for most biological materials, their mass magnetic susceptibility (m3
) was comparable to that of water (−9.051×10−9) (). When supplemented with ferric citrate, all the strains exhibited positive values. At high fields (2,500 to 10,000 Oe), magnetization is proportional to field and not saturating, indicating a dominant contribution of paramagnetism. At low fields (0 to 2,000 Oe), an upward concave curve of magnetization was observed, indicating additional ferro/ferri-magnetic contribution, which typically saturates within this region. This suggests that the cells contain mostly paramagnetic (or superparamagnetic) material with a slight amount of ferro/ferri-magnetic material. Mass magnetic susceptibility of the paramagnetic constituent was given based on values at high fields (). Those of ferritin-expressor, ccc1Δ
, and ccc1Δ
ferritin-expressor were approximately 1.3, 1.8, and 2.8 times larger than that of wild type, respectively. Previous studies on magnetic susceptibility of isolated ferritin ranged from 3.7×10−8
m3/kg (originally 2.95×10−6
em in cgs unit) at room temperature, depending on the sample and measuring method 
(reviewed in 
). Thus, we observed a gain of magnetic susceptibility due to ferritin expression, while a non-ferritin contribution was also present, indicating that ferric citrate supplementation induces basal magnetization in yeast. ccc1Δ
showed increased magnetization compared to wild type, suggesting that non-vacuolar iron may have more magnetic contribution than previously thought. The synergistic effect of ferritin and ccc1Δ
can be explained by higher availability of iron to ferritin in the cytosol.
Mass magnetic susceptibility of ferric citrate–supplemented cells.
Magnetized Cells Contain Electron-Dense Deposition within Membranous Structures
Ultrathin section transmission electron microscopy showed accumulation of electron-dense deposits (). Although these varied in shape, size, and amount among cells, wild type cells typically contained round particles associated with membranous structures that are most likely the vacuoles ( wild-type), while ccc1Δ cells tended to contain aggregates within mitochondria (
ccc1Δ). As the mitochondria are where cells convert inorganic iron into heme and iron-sulfur clusters, the observed deposits could be caused by iron overload due to the defect in vacuolar iron sequestration, and may contribute to the higher magnetic susceptibility in ccc1Δ. Ferritin expression had little observable effect on the electron micrographs ( ferritin) perhaps due to the small size of the iron binding center.
Electron micrographs of magnetized cell.
Electron-Dense Deposits Contain Iron and Phosphorous
To reveal the elemental composition of the electron-dense deposits, magnetized cells were analyzed by energy-dispersive X-ray spectroscopy (EDS). Elemental maps were obtained for detectable elements, and iron, phosphorous, oxygen, and nitrogen showed characteristic distributions associated with cellular structures (). Nitrogen was distributed throughout the cell consistent with association with biogenic molecules such as proteins. In wild type cells grown in ferric citrate, phosphorous, iron, and oxygen were slightly concentrated within membranous structures presumably vacuoles and in electron-dense round particles (, magnified images). In ccc1Δ cell, iron showed increased localization to the clusters of electron-dense crystals (). Phosphorous also accumulated in the clusters. Oxygen showed a similar pattern to phosphorous with less contrast. These two types of electron-dense deposits (small round particles in wild type and clustered crystals in ccc1Δ) thus contained iron, oxygen, and phosphorous with different composition stoichiometries (). The elemental maps were further analyzed to estimate relative amounts of iron and phosphorous (); iron was higher in ccc1Δ than in wild type cells, whereas phosphrous showed only a small increase in ccc1Δ.
EDS elemental analysis for magnetized cell.
Cells Can Be Trapped by a Magnetic Column
Magnetic columns have been used for separation of biomaterials labeled with magnetic particles. To test if our yeast cells behave similarly, the cells were applied to a magnetic column. Normally grown yeast cells were not retained on the column under the conditions tested (, normal). The cells supplemented with ferric citrate were retained by the magnetized column (). Among the four strains, the order of rate of trapped cells is in agreement with their magnetic susceptibility measured by SQUID, indicating that this system can be used for comparison of cell magnetization, as well as to separate magnetic cells.
A Component of TORC1 Is Important for the Magnetization
Genetic control of magnetization would greatly expand the engineering potential of magnetic cells. To explore this possibility as well as to gain further insight into the nature of the biogenic magnetization, we sought yeast gene knockout strains that show altered magnetization. Candidate genes to be tested were selected based on their functional or phenotypic description associated with iron homeostasis or oxidative stress from Saccharomyces
Genome Database (http://www.yeastgenome.org
). Mutant strains were grown in 20 mM ferric citrate medium and their attraction towards a magnet was observed (). Strains showing reproducible altered attraction were selected and subjected to magnetic column separation to confirm and quantify their magnetization. From the initial screen of 60 strains (Table S1
was found to show consistent reduction of magnetization compared to wild type (). Tco89p is known to be a nonessential component of TORC1 
. TORC1 globally regulates cell growth in response to nutrient, stress, and redox states (reviewed in 
). To ask if and how TORC1 is involved in the magnetization, Tor1p, the other nonessential component of TORC1 and Ssd1p, which coordinates with TORC1 to maintain cell integrity 
, was tested. Both tor1Δ
showed little change in the magnetization (), indicating that challenged cell integrity is not associated with the magnetization. Magnetization of the cells positively correlated with copy number of TCO89
(), showing a dose-dependent effect of TCO89
on the magnetization. Expressing TCO89
under a galactose inducible promoter pGal1 showed induction of magnetization ().
Effect of TCO89 on magnetization.
Magnetization by TCO89 Is Independent from CCC1 But Dependent on TORC1 Activity
Loss of TCO89 in ccc1Δ decreased magnetization (), suggesting that iron sequestration into the vacuole by CCC1 does not have a predominant effect on induction of magnetization by TCO89. In contrast, TCO89 affects magnetism through TORC1 activity. We used rapamycin, an inhibitor for TORC1 at sub-lethal doses. Rapamycin treatment reduced magnetism in wild type and more prominently in multi-copy TCO89, while no reduction is observed in tco89Δ (), indicating that induction of magnetism by TCO89 is through TORC1 activity.
Effects of TORC1 activity on magnetization.
Carbon/Nitrogen Balance Affects Magnetism
As TORC1 processes nutritional signals, we tested if the nutritional environment affects magnetism. Compared to synthetic-defined medium, we observed a reduction of magnetism when cells were grown in rich medium with the same amount of iron (). The magnetism then increased as extra glucose was added to rich medium. In contrast, addition of extra nitrogen (i.e., amino acids and nucleotides) to synthetic defined medium decreased magnetism, suggesting that the relative availability of carbon and nitrogen has impact on the formation of magnetism. Effect of TCO89 became less prominent in rich medium or when extra nitrogen was added. These results indicate that higher carbon availability has a positive effect on magnetization, which is enhanced by TOC89.
TCO89 Controls Redox State
As iron homeostasis has a close relationship with redox state, we asked if TCO89 has any function associated with redox control. Cellular redox activity can be monitored by a biocompatible redox indicator methylene blue, which loses its color when reduced. Equal numbers of cells were spotted and grown on plates containing methylene blue to observe colony staining. Compared to wild-type or plasmid-complemented cells, tco89Δ exhibited little color while multi-copy TCO89 cells were blue (), indicating that TCO89 leads cellular redox to an oxidized state in a dose-dependent manner. tco89Δ also showed compromised cell growth in the presence of methylene blue presumably because a higher rate of methylene blue reduction interferes with cellular metabolism.
Nicotinamide adenine dinucleotide phosphate (NADP) is a coenzyme that serves as a redox mediator in protection against oxidative stress. Cells harboring multi-copy TCO89 showed higher levels of both NADP+ (oxidized) and NADPH (reduced), while tco89Δ had slightly lower NADP+ and higher NADPH ().
Genes for Carbon Metabolism and Mitochondrial Redox Affect Magnetism
We expanded genetic screening for candidates related to carbon metabolism and redox using the magnetic column entrapment procedure (Figure S1
). Gene knockouts affecting oxidative damage, such as GRX2, GRX3, and SOD2, did not show significant changes, while POS5, a gene for mitochondrial NADH kinase, showed reduction in magnetism. In contrast, UTR1 did not affect magnetism, which is a cytoplasmic ATP-NADH kinase. Gain of magnetism was seen with loss of YFH1, which has been reported to accumulate iron in mitochondria 
, and whose human homolog FXN is responsible for the neurodegenerative disease Friedreich's ataxia 
. Regarding carbon metabolism, gene knockouts for SNF1 and ZWF1 showed reduction in magnetism. SNF1 is required for processing carbon stress signals (reviewed in 
) and ZWF1 codes for an enzyme at the branch point of the pentose phosphate pathway