Iron-dependent cleavage of ribosomal RNA during oxidative stress in the yeast Saccharomyces cerevisiae

Stress-induced strand breaks in rRNA have been observed in many organisms, but the mechanisms by which they originate are not well-understood. Here we show that a chemical rather than an enzymatic mechanism initiates rRNA cleavages during oxidative stress in yeast (Saccharomyces cerevisiae). We used cells lacking the mitochondrial glutaredoxin Grx5 to demonstrate that oxidant-induced cleavage formation in 25S rRNA correlates with intracellular iron levels. Sequestering free iron by chemical or genetic means decreased the extent of rRNA degradation and relieved the hypersensitivity of grx5Δ cells to the oxidants. Importantly, subjecting purified ribosomes to an in vitro iron/ascorbate reaction precisely recapitulated the 25S rRNA cleavage pattern observed in cells, indicating that redox activity of the ribosome-bound iron is responsible for the strand breaks in the rRNA. In summary, our findings provide evidence that oxidative stress–associated rRNA cleavages can occur through rRNA strand scission by redox-active, ribosome-bound iron that potentially promotes Fenton reaction–induced hydroxyl radical production, implicating intracellular iron as a key determinant of the effects of oxidative stress on ribosomes. We propose that iron binding to specific ribosome elements primes rRNA for cleavages that may play a role in redox-sensitive tuning of the ribosome function in stressed cells.

Reactive oxygen species (ROS) 2 are omnipresent stressors for all biological systems. Low levels of ROS play roles in many biochemical processes and act as signaling molecules, but at supraphysiological levels, ROS can damage cellular components, including lipids, proteins, and nucleic acids (1,2). Through oxidation of rRNA and proteins, ROS can influence the functionality of ribosomes, the complex ribonucleoprotein machines responsible for protein synthesis. For example, translation by the bacterial ribosome is inhibited by oxidation of the rRNA bases within the peptidyl transferase center (3). Reversible oxidation of cysteine residues in ribosomeassociated proteins under conditions of increased ROS generation has been proposed to attenuate translation (4). Oxidative damage to rRNA has been hypothesized to play a role in neurodegenerative disease (5,6). However, despite the central role of the ribosome in protein synthesis, our understanding of the impact of ROS on ribosome functions is still very limited.
In our previous work (7), we found that low, survivable levels of oxidative stress in the budding yeast Saccharomyces cerevisiae lead to rRNA cleavages in a subset of cytoplasmic ribosomes. One prominent cleavage, located in the ES7 region of 25S rRNA in the large (60S) ribosomal subunit, starts at early stages of the adaptive oxidative stress response that helps cells to cope with ROS accumulation. The ES7 region is located on the surface of the 60S subunit and is one of the largest rRNA expansion clusters in the eukaryotic ribosome (8,9). ES7cleaved ribosomes are capable of reinitiating translation, suggesting that this type of rRNA cleavage does not inactivate ribosomes and may instead play a role in modulating their function during stress (7).
Aside from adaptable oxidative stress conditions, excessive ROS accumulation is a hallmark feature of apoptosis (10,11). Induction of programmed cell death in various organisms has been shown to involve degradation of rRNA, presumably carried out by nucleases activated as part of apoptotic mechanisms (12)(13)(14). The identity of eukaryotic ribonucleases that initiate ribosome degradation during programmed cell death has remained uncertain.
In this study, we took advantage of the rapid yeast rRNA analysis techniques we developed recently (15) to survey genes encoding ribonucleases and proteins with function in redox regulation for possible effects on the ES7 cleavage in 25S rRNA. Unexpectedly, we found that the cellular control of iron but not ROS levels was the major determining factor in rRNA fragmentation during oxidative stress. Genetic and biochemical analyses confirmed that oxidant-induced cleavage of yeast rRNA is a nonenzymatic, iron-dependent process. Our results identify a previously unknown mechanism of site-specific changes in the ribosome that may have a biologically significant role in diverse oxidative stress conditions and during programmed cell death.

Deletion of the monothiol glutaredoxin gene GRX5 causes severe rRNA degradation in yeast exposed to low-level oxidants
Previously, we identified cleavage in the expansion segment ES7 of the 25S rRNA as one of early molecular responses to elevated ROS levels in the budding yeast S. cerevisiae (7). To better understand cellular factors that affect ES7 cleavage, we screened a panel of ϳ150 yeast strains deficient for antioxidant and oxidative stress-related genes as well as genes that encode various RNases. After growing cells to mid-log phase, we induced oxidative stress with low doses of inorganic peroxide (H 2 O 2 , 0.25 mM) or the redox-cycling agent menadione (vitamin K3, 50 M) and isolated total RNA from the cells (15). Fig.  1A shows a representative set of five strains that were either left untreated or treated with the oxidants and compared with the WT BY4741 control. A number of deletion strains in our test panel showed increased accumulation of stress-induced RNA fragments, including products of ES7 cleavage in 25S rRNA (7), consistent with the involvement of multiple antioxidant systems in the cellular neutralization of ROS. To compare the strength of this phenotype in different strains, we performed Northern hybridizations with the probe y540, which detects both full-length 25S rRNA and the well-resolved 5Ј fragment (ϳ600 nt) produced by ES7 cleavage (Fig. 1B) and determined the ES7 fragment/25S ratio in each lane (Fig. 1C). Control measurements in biological replicates showed that using this ratio minimizes variations resulting from loading and strain-specific differences in rRNA content. Among all tested mutants, deletion of the monothiol glutaredoxin gene GRX5 led to a particularly strong loss of rRNA stability in response to both H 2 O 2 and menadione (Fig. 1C).
To ensure that the observed 25S rRNA degradation phenotype was due to lack of GRX5 rather than any secondary mutations that may have accumulated in a passaged knockout library strain, we backcrossed grx5⌬ with WT BY4741 cells and generated several new grx5⌬ strains by tetrad dissection. The newly derived grx5⌬ strains were viable; however, they grew noticeably slower than the WT in fermentation medium (Fig. 1D). In agreement with a previous study (16), grx5⌬ strains were deficient in respiration (Fig. S1A). Genetic tests and analysis of mitochondrial DNA showed grx5⌬ cells to be 0 (Fig. S1, B-D), explaining their respiration deficiency. Testing the new grx5⌬ strains confirmed their high sensitivity to oxidants. For example, grx5⌬ cells were unable to grow in medium supplemented with 50 M menadione, a dose that caused only a transient growth delay in the WT (Fig. 1D). Consistent with the screen data ( Fig. 1C), Northern hybridization with the probe y540 showed severe 25S rRNA degradation in grx5⌬ cells after treatment with either 0.25 mM H 2 O 2 or 50 M menadione (Fig. S2A). The ES7-cleaved fragment was detectable in grx5⌬ cells after treatment with as little as 10 M menadione, whereas 25S rRNA integrity in WT cells was unaffected by this dose (Fig. 1E).
To better define the kinetics of the oxidant-induced 25S rRNA decay, we next analyzed rRNA at different time points after addition of 50 M menadione to the culture medium ( Fig.  2A). In grx5⌬ cells, the ES7-cleaved fragment was apparent after a 1-min treatment and further increased in intensity over time, paralleled by decay of the full-length 25S rRNA (Fig. 2). By comparison, menadione treatment in WT cells led to a slower onset of rRNA degradation and affected only a small fraction of ribosomes (Fig. 2). This indicated that cleavage of the ES7 region of 25S rRNA occurs virtually instantaneously upon oxidant exposure of grx5⌬ cells, arguing against lengthy mechanisms such as autophagic delivery of ribosomes to the vacuole (17) or induction of the apoptotic program (14). Northern hybridizations with probes for other RNA types revealed increased degradation of 5S rRNA, 5.8S rRNA, and tRNAs after treatment of grx5⌬ cells with menadione or H 2 O 2 (Fig. S2, C-F). However, large rRNAs (18S and 25S) were by far the most severely affected RNA species (Fig. S2, A and B), with only ϳ2% of intact 25S rRNA remaining after 2-h treatment with 50 M menadione (Fig. S2A). Hybridizations with additional 25S rRNA probes verified extensive degradation of the entire rRNA molecule in oxidant-treated grx5⌬ cells (Fig. S3).
Collectively, these data show that Grx5 plays a crucial role in maintaining ribosome integrity during oxidative stress. Lack of Grx5 greatly intensifies the effects of oxidants on ribosomes. This not only increases the number of ES7-cleaved ribosomes, but can also lead to runaway ribosome degradation after treatment with low oxidant levels that are survivable in Grx5-expressing cells.

The increase in intracellular ROS levels is not sufficient to explain the ribosome degradation phenotype in grx5⌬ cells
Grx5 belongs to the glutaredoxin family of enzymes, also known as GSH-dependent oxidoreductases, many of which help to maintain the redox state of proteins during oxidative stress (18). Because some glutaredoxins can also function as ROS scavengers (19,20), we first sought to determine how lack of Grx5 affected the cellular ROS load. We estimated H 2 O 2 levels in grx5⌬ cells by using for comparison a ctt1⌬ strain deficient for catalase, a key H 2 O 2 -inactivating enzyme. Amplex Red assays performed after treatment with 50 M menadione or 0.25 mM H 2 O 2 indeed revealed elevated H 2 O 2 levels in grx5⌬ cells as well as in the control ctt1⌬ strain (Fig. 3A). However, the observed increases in H 2 O 2 did not correlate well with the extent of rRNA degradation (Fig. 3B). For instance, grx5⌬ cells exhibited more rRNA degradation after treatment with 25 M menadione than the WT and ctt1⌬ cells after 50 M treatment (Fig. 3B) despite the lack of a comparable H 2 O 2 increase (Fig. 3A).
To assess the levels of the O 2 . radical in oxidant-treated grx5⌬ cells, we used the superoxide-specific probe dihydroethidium (DHE) (21). Oxidized DHE was analyzed by a flow cytometry assay, which revealed moderately increased superoxide levels in grx5⌬ cells compared with the WT cells (Fig. 3C). However, the O 2 . levels detected in grx5⌬ cells were less pronounced than in the control sod2⌬ strain lacking superoxide dismutase (Fig.  3C). Importantly, sod2⌬ cells displayed significantly lower rRNA degradation than similarly treated grx5⌬ cells (Fig. 3D). Thus, neither hydrogen peroxide nor superoxide radical levels reflect the severity of ribosome degradation during oxidant treatments of grx5⌬ cells, suggesting that additional factor(s) may play a role in this process.

rRNA undergoes chemical hydrolysis during oxidative stress Increased levels of labile iron is a key factor in 25S rRNA degradation in grx5⌬ cells
Previous studies have shown that Grx5 participates in ironsulfur cluster (ISC) protein biogenesis (22)(23)(24)(25)(26)(27), whereas the lack of Grx5 function causes iron accumulation in yeast cells (25,28). Iron can both potentiate ROS production and exacerbate ROS-induced toxicity in the cell (28,29). We therefore asked whether the disruption of iron homeostasis in grx5⌬ cells contributes to the oxidant-induced effects on rRNA.
First, we analyzed the iron content of the grx5⌬ cells and confirmed that it was significantly increased compared with the WT (Fig. 4A), in agreement with the results of a previous study (25). We also found that treatment of grx5⌬ cells with the membrane permeable Fe 2ϩ chelator 1,2-phenanthroline (PHL) (30) Remarkably, the addition of PHL prior to menadione treatment alleviated the oxidant-induced loss of viability in grx5⌬ cells (Fig. 4B, right). Analysis of RNA extracted from these cells showed protection of 25S rRNA from menadione-induced degradation by PHL ( Fig. 4C and S5A). Similar protective effects of PHL were observed with H 2 O 2 treatments (Fig. S5).
To address the contribution of iron through an independent approach, we used overexpression of the human ferritin MtF1 to decrease the labile intracellular iron pool in grx5⌬ cells. Human ferritins maintain proper iron homeostasis by seques-tering excess iron in intracellular complexes (31). Expression of the human mitochondrial ferritin MtF1 has been shown previously to effectively reduce free iron concentration in yeast cells (32). Consistent with the results obtained via chemical chelation of iron, grx5⌬ cells harboring the MtF1 expression construct displayed reduced degradation of 25S rRNA after oxidant treatment compared with cells transformed with an empty vector plasmid (Fig. 4D).
Based on these data, we conclude that the extent of oxidantinduced rRNA cleavages depends on intracellular iron levels. A defect in iron homeostasis, which leads to accumulation of excessive iron in cells deficient for Grx5 function, may thus explain why severe RNA damage is observed in grx5⌬ cells but not in strains defective for various ROS-scavenging enzymes (Fig. 1C).

Iron-dependent site-specific cleavage of 25S rRNA in vitro
The above results indicate that the severe degradation of ribosomes in grx5⌬ cells results from the combination of two factors: oxidants and high intracellular iron. Transition metal ions, including iron, can enhance ROS-induced damage in biomolecules, at least in part, through the generation of the highly reactive ⅐OH radicals in a Fenton reaction: This led us consider the possibility that the rRNA cleavages might be a result of chemical rather than enzymatic hydrolysis. These cleavages might be directed by redox-reactive iron bound at specific ribosomal sites; for example, through interactions with rRNA or ribosomal proteins in the proximity of the RNA backbone. Previous studies have shown that binding of redox-active metals to remove DNA and RNA molecules can give rise to oxidizing species that are not freely diffusible, thereby producing site-specific cleavages (33)(34)(35)(36). Studies have also shown that Fe 2ϩ can substitute for Mg 2ϩ in RNA (37), with highly structured RNA often exhibiting preferred sites for Fe 2ϩ binding (33). Examination of the ES7 region in yeast 25S rRNA, where the major iron-dependent cleavage takes place, revealed three Mg 2ϩ ions located 6 -8 Å away from the ES7 cleavage site (Fig. 5A), providing potential sites for exchange with Fe 2ϩ ions.
To test the hypothesis that the site-specific rRNA cleavages in vivo may be driven by redox-active iron bound to ribosomes, we sought to reconstitute the cleavage process in vitro. We devised reaction conditions based on earlier studies showing that ascorbic acid can induce localized redox reactions at the sites of metal binding (38,39). Cleavage of nucleic acids by ascorbate in the presence of molecular oxygen is thought to occur by a multistage mechanism involving metal-catalyzed oxidation of ascorbic acid with molecular oxygen, leading to H 2 O 2 formation (40) and reduction of the ligand-bound metal to the form capable of producing the highly reactive ⅐OH radicals from H 2 O 2 (38):  We assembled the in vitro reactions with ribosomes that were purified from WT yeast cells using centrifugation through a sucrose cushion to remove soluble cytosolic components. To minimize the generation of hydroxyl radicals in solution, which can attack solvent-accessible regions of nucleic acids with little selectivity (41, 42), we did not add any exogenous H 2 O 2 , thus relying on ascorbate-iron redox cycling as the principal ROS source in the system. Fig. 5B shows the Northern blot analysis of 25S rRNA from ribosomes treated with varying concentrations of ascorbic acid and 1 M Fe(NH 4 ) 2 (SO 4 ) 2 . The combination treatment efficiently induced ES7 cleavage, as evidenced by the accumulation of the ES7-cleaved fragment, and generated several additional minor bands on the gel that matched well the pattern of rRNA fragments appearing after oxidant treatment of grx5⌬ cells (Fig. 5B). Primer extension analysis of the 25S rRNA undergoes chemical hydrolysis during oxidative stress rRNA from the ascorbic acid/iron-treated ribosomes (Fig. 5C) confirmed that the ES7 cleavage in vitro corresponds precisely to cleavage induced by redox stress in live cells, including grx5⌬ cells and the previously studied thioredoxin peroxidase-deficient tsa1⌬ cells (7). One notable observation in these experiments was that ES7 cleavage could be obtained with ascorbic acid treatment alone (Fig. 5B), indicating that low levels of iron (or another redoxactive metal) were already present in ribosomes isolated from cells. Supplementing the ribosome cleavage reaction with Fe 2ϩ markedly increased the level of the ES7 cleavage product (Fig.  5B), showing that iron can indeed promote cleavage at this site. Conversely, addition of the iron chelator deferoxamine (DFO) completely blocked rRNA cleavage (Fig. 5B). Similar protective effects were observed with other chelating agents (EDTA, PHL, and BPS; Fig. 5D), consistent with their ability to displace iron from complexes with biomolecules. Because soluble Fe(EDTA) complexes can readily react with H 2 O 2 and generate diffusible ⅐OH radicals (43,44), this result further supports the idea that the site specificity of rRNA cleavages depends on the iron ion association with rRNA.

Discussion
Site-specific cleavages of rRNA have been observed previously in different organisms that experience various types of stress (12-14, 45-47). Although these cleavages were ascribed to cellular nucleases, in most cases the enzymes responsible for initiating rRNA strand breaks were not conclusively identified. The results presented here provide several lines of evidence that rRNA cleavages associated with oxidative stress in yeast cells can occur through a nonenzymatic mechanism based on rRNA strand scission by redox-active iron bound to the ribosome. These findings implicate intracellular iron as a key determinant of the effects of oxidative stress on ribosomes (Fig. 6).
Iron is an essential nutrient that acts as a co-factor for many enzymes and transport proteins and participates in a wide variety of physiological processes, including respiration, DNA metabolism, lipid biosynthesis, oxygen transport, and others (48,49). Eukaryotes have developed complex systems for iron uptake, utilization, and storage (48 -50). Disruption of these systems can result in either iron shortage or overload, with mitochondrial defects often playing a major role in iron homeostasis disturbances (49). For example, defects in the mitochondrial ISC biosynthesis pathways can increase iron uptake, resulting in excessive amounts of iron in the cytoplasm (51). In this study, we found that deletion of the yeast mitochondrial monothiol glutaredoxin gene GRX5, involved in ISC biogenesis (22)(23)(24)(25)(26)(27), made ribosomes highly susceptible to ROS-induced damage. This manifested in the extensive degradation of rRNA after exposure of cells to low-dose oxidants (Figs. 1, A-C, and 2 and Figs. S2 and S3) and was accompanied by a rapid decline in cell viability (Fig. 4B). Although grx5⌬ cells con-

rRNA undergoes chemical hydrolysis during oxidative stress
tain elevated levels of ROS (Fig. 3, A and C), we observed no direct correlation between ROS amounts and the extent of rRNA fragmentation (Fig. 3, B and D). Instead, we found that lowering the iron content of grx5⌬ cells through a cell-permeable chemical chelator or expression of the human mitochondrial ferritin MtF1 (32) suppressed cleavage formation in rRNA (Fig. 4, C and D, and Fig. S5) and improved cell survival (Fig. 4B). Crucially, we found that probing ribosomes for the presence of bound metals using an in vitro ascorbic acid redox cycling reaction recapitulated the 25S rRNA cleavage pattern observed in cells. Supplementing this reaction with inorganic iron was sufficient to increase the efficiency of the ES7 rRNA cleavage, whereas iron chelation completely blocked it (Fig. 5, B and D). These data imply that no additional nuclease is required to cleave the rRNA. Consistent with a role of iron in generating rRNA strand breaks, we observed a markedly increased rRNA degradation during oxidative stress in grx5⌬ and other mutants with excessive iron content but not in strains with defective ROS scavenging systems (Fig. 1C).

rRNA undergoes chemical hydrolysis during oxidative stress
One significant observation is that the iron-mediated rRNA cleavages both in vivo and in vitro were not random but rather site-specific. This argues for the existence of high-affinity iron binding sites within the ribosome structure. One possibility is that iron may be brought in proximity to the rRNA's sugar phosphate backbone by a protein capable of coordinating iron ions without inhibiting their redox potential. An alternative possibility is that rRNA itself captures iron ions; for instance, by substitution of Mg 2ϩ in permissive and accessible rRNA structures. Consistent with this scenario, structural data demonstrate three Mg 2ϩ ions inside of the ES7/ES7 CJ region of 25S rRNA and located ϳ6 -8 Å away from the ES7 cleavage site (Fig. 5A). These Mg 2ϩ ions are exposed to the solvent side of the large subunit, allowing their potential replacement with Fe 2ϩ . In support of iron binding by rRNA, previous studies have shown that Fe 2ϩ , being a divalent cation with similar ionic radii and geometric properties as Mg 2ϩ , is capable of structural and functional replacement of Mg 2ϩ in nucleic acids (33,37,(52)(53)(54)(55). It is clear, however, that not every Mg 2ϩ ion bound to the ribosome could be substituted by iron with the same efficiency. The number of Mg 2ϩ ions estimated to exist in the ribosome structure by structural studies (8) greatly exceeds the number of rRNA cleavages observed after oxidant treatments (Fig. S2). The site specificity of iron binding to rRNA could depend on local structural features, by analogy to the previously described differential binding of iron to DNA (34). In addition, sequence-and structure-specific interactions of RNA with iron have been observed in iron-responsive elements within mRNAs (56), group I intron RNAs (33), and, recently, bacterial ribosomes (57).
The reaction between ligand-bound Fe 2ϩ and H 2 O 2 is capable of generating hydroxyl radicals (⅐OH) through a Fenton mechanism and higher oxidation states of iron, including highly reactive ferryl (Fe 4ϩ ) species (44,58). Similarly to ⅐OH, ferryl oxygen [Fe 4ϩ ϭ O] 2ϩ is capable of nucleic acid strand cleavages, as seen with DNAzymes (59). The exact nature of the reactive species produced in a given reaction may depend on multiple factors, including pH, the nature of the substrate, and the chelation state of iron. Although discernible in a controlled chemical system, these parameters are often difficult to account for in a cellular setting (discussed in Ref. 44). Given the complex molecular environment of the ribosome, it is possible that different types of chemistry may underlie rRNA hydrolysis at specific ribosomal locations.
The function of iron as an effector for the ribosome in oxidative stress may be significant in different biological contexts. The finding that the ES7 rRNA cleavage occurs during adaptive stages of the oxidative stress response (7) suggests that ironmediated rRNA cleavages are not purely destructive but may serve a beneficial biological function in healthy cells. For example, these cleavages might temporarily down-regulate protein synthesis or alter some specific aspects of ribosome function during oxidative stress. By contrast, a failure to control intracellular iron levels, such as seen in strains defective in iron metabolism, predisposes ribosomes to rapid degradation even at low oxidant levels (Fig. 6). This may effectively prevent the synthesis of new proteins required for stress damage control and, hence, enforce a program of cellular self-destruction. In previous studies, ribosome degradation has been observed during apoptotic cell death (13,14,60). Recently, a form of programmed cell death in mammalian cells, termed ferroptosis (61), has been related to the induction of excessive lipid damage via iron-dependent peroxidation (reviewed in Ref. 62). Understanding the role of iron in ferroptosis is far from complete (48). It is possible that iron-dependent effects on the cell's ribosomes, analogous to those described here, may provide a convergence point for cell death mechanisms operating in diverse organisms.
Cells were treated with H 2 O 2 (Avantor Performance Materials), menadione (Enzo), and PHL (VWR Life Science). For in vitro reactions, we used Fe(NH 4 ) 2 (SO 4 ) 2 (Sigma), ascorbic acid (Alfa Aesar), DFO (BioVision), 4,7-diphenyl-1,10-phenanthrolinedisulfonic acid (BPS, Alfa Aesar), and EDTA (Sigma). In cells with balanced iron metabolism, ROS cause limited, position-specific cleavages in rRNA at the sites of bound iron. This might affect ribosome function and optimize translation to conditions of oxidative stress, potentially serving as a stress adaptation mechanism. Accumulation of labile iron, such as in cells with defects in mitochondrial iron metabolism, increases iron binding to ribosomes and leads to extensive rRNA fragmentation, causing global inhibition of translation, which, in turn, may accelerate cell death.

rRNA undergoes chemical hydrolysis during oxidative stress
The human mitochondrial ferritin gene MTF1 was cloned from human complementary DNA into the pNS6 vector using standard PCR-based techniques. The pNS6 vector was created by replacing the GAL1-GAL10 promoter region within PESChis (Stratagene) with a 1.5-kb ADH1 promoter region from the PUAD plasmid (a kind gift from Randy Strich) (64).

RNA extraction, Northern blotting, and primer extension
Total RNA was isolated from cells using one-step extraction with formamide-EDTA as described previously (15). To analyze large rRNA species (25S and 18S rRNAs), RNA was separated on 1.2% agarose gels containing 1.3% formaldehyde (65). Small rRNAs and tRNAs were separated on 8% polyacrylamide gels containing 8 M urea as described previously (66). To visualize total RNA, agarose gels were stained with SYBR Gold (Invitrogen) at room temperature for 30 min, and the fluorescent signal was scanned using a Typhoon 9200 imager (GE Healthcare) at 532 nm. RNA was transferred to nylon membranes (Hybond N, GE Biosciences), and individual rRNA and tRNA species were visualized by Northern hybridizations using 32 P-labeled oligonucleotide probes (67). The sequences of all probes used in this study are presented in Table S1. All hybridizations were analyzed using Typhoon 9200 in phosphorimaging mode and ImageQuant software (GE Healthcare). For quantification, the volume of the hybridization signal corresponding to the fragment(s) of interest was converted to phosphorimaging units, and the average image background value was subtracted. For primer extensions, a total of 3 g of RNA was separated on a guanidine thiocyanate-containing gel, and rRNA was extracted from excised bands as described previously (68). RNA was annealed with 2 pmol of the 32 P-labeled primer Prex1 and used for primer extensions (69). The Prex1 probe sequence can be found in Table S1. Reaction products were processed and analyzed on a 6% polyacrylamide/urea gel (7). Sequencing reactions were performed with SequiTherm EXCEL II (Epicenter) using 250 ng of the pJD694 plasmid containing rDNA (70) as a template and the Prex1 primer.

ROS detection
To measure H 2 O 2 levels, Amplex Red assays were performed with a kit from Thermo Fisher as described previously (7) with a few minor modifications. Briefly, 2 ϫ 10 7 mid-log phase cells were collected, washed with PBS, and resuspended in 300 l of the reaction buffer supplied with the kit. Cells were incubated for 30 min in a 30°C shaker and centrifuged, and 50 l of supernatant was taken into an Amplex Red assay reaction performed in triplicate. The fluorescence of the reaction product resorufin was measured at 590 nm using a Synergy HT microplate reader (BioTek).
For analysis of the mitochondrial superoxide, we used a technique described previously (71). Briefly, 2 ϫ 10 7 cells were harvested by centrifugation, resuspended in 500 l of PBS, and stained with 2.5 g/ml DHE (Molecular Probes) for 30 min in the dark at room temperature. Fluorescence was measured using a BD Accuri C6 flow cytometer (excitation at 488 nm, emission at 585/40 nm), and histograms were analyzed in FlowJo version 10.

Measurement of iron levels
To determine levels of total cellular iron, we used a technique described previously (72). Briefly, cells were collected and counted, and the same amounts of cells (5 ϫ 10 8 ) were washed and resuspended in 3% nitric acid. Cells were incubated for 16 h at 98°C, followed by iron chelation with BPS. The absorbance was measured at 535 nm (73).

In vitro redox cycling reaction
Pellets of purified ribosomes (see above) were resuspended in 100 l of buffer K supplemented with RNaselock (Thermo Fisher). The RNA concentration was estimated spectrophotometrically, and 5 g was used for one reaction. The reaction volume was adjusted to 100 l with buffer K. Solutions of ascorbic acid, Fe(NH 4 ) 2 (SO 4 ) 2 , DFO, PHL, BPS, and EDTA were freshly prepared as 100ϫ stocks. 1 l of each reaction component was placed on a tube wall, and the tubes were briefly centrifuged to start all reactions simultaneously and incubated on ice for 10 min. The reactions were stopped with 50 l of 0.1 M thiourea. RNA was precipitated by isopropanol, washed with 80% ethanol, and dissolved in 6 l of FAE (98% formamide and 10 mM EDTA) (15). Reaction products were resolved on agarose gels and analyzed by Northern hybridizations as described above.

Structural analysis
PDB file 4V88 (8) was obtained from the Protein Data Bank and visualized with PyMOL 2.1 (Schrödinger).

Cell viability and growth assays
For growth assays, yeast cultures were adjusted to A 600 ϳ0.1 with appropriate media (with or without 50 M menadione), 200 l/well was inoculated into 96-well plates in three replicates, and the cultures were grown for 36 h at 30°C with shaking. A 600 measurements were taken every 5 min and automatically recorded using a BioTek Synergy HT microplate reader.
For viability assays, overnight cultures were diluted with YPDA to A 600 ϳ0.2, grown for 4 h, and incubated with 80 M PHL for 1 h or left untreated. For oxidant treatment, cells were incubated with 50 M menadione for 30 min to 1 h at 30°C with shaking. Cells were washed, resuspended in fresh YPDA, and adjusted to the same concentration of 2 ϫ 10 6 cells/ml. Six 1:5 serial dilutions for each culture were plated on YPDA agar plates and incubated at 30°C for 3-5 days. rRNA undergoes chemical hydrolysis during oxidative stress