Manipulation of in vivo iron levels can alter resistance to oxidative stress without affecting ageing in the nematode C. elegans

Highlights ► Here we test whether iron-catalyzed oxidative damage contributes to organismal ageing. ► We develop new methodologies to measure free iron in vivo in C. elegans. ► Moderate iron supplementation can increase oxidative damage without reducing lifespan. ► Iron chelation or increasing ferritin levels increase resistance to oxidative stress but do not increase lifespan. ► Our findings argue against the oxidative damage theory of ageing.


The oxidative damage theory of ageing
The mechanisms underlying the ageing process remain poorly defined, although many theories have been proposed (Medvedev, 1990). Prominent among these is the idea that ageing results from the accumulation of damage to biomolecules caused by reactive oxygen species (ROS) (Harman, 1956;Sohal and Weindruch, 1996). Thus, far, extensive experimental probing of this using animal models has not, on the whole, resulted in its clear validation (Muller et al., 2007;Perez et al., 2009).
A number of tests of the oxidative damage theory have used the short-lived nematode Caenorhabditis elegans (Van Raamsdonk and Hekimi, 2010). In this organism in particular, the results of many such tests have not supported the theory. For instance, the gene sod-2 encodes the major mitochondrial superoxide dismutase (SOD) (Doonan et al., 2008), an important antioxidant enzyme that removes the superoxide free radical (O 2 À ). Deletion of sod-2 might be expected to increase oxidative damage levels and decrease lifespan. It does indeed increase oxidative damage but does not decrease lifespan, and can even increase it (Van Raamsdonk and Hekimi, 2009). In the same vein, treatment with compounds with SOD activity can increase resistance to oxidative stress, but does not increase C. elegans lifespan (Keaney and Gems, 2003;Keaney et al., 2004;Kim et al., 2008;Uchiyama et al., 2005), though it should be noted that one initial study did report an increase in lifespan (Melov et al., 2000). Furthermore, treatments that increase ROS production can increase rather than decrease lifespan (Heidler et al., 2010;Schulz et al., 2007;Yang and Hekimi, 2010). Taken together, these studies raise doubts about whether oxidative damage contributes to age-associated increases in pathology and mortality in C. elegans (Gems and Doonan, 2009). pool of ''free'' (uncomplexed) iron, which becomes toxic at high concentrations, particularly due to its ability to generate oxidative stress via the Fenton reaction. Here Fe(II) is oxidized by H 2 O 2 to Fe(III) generating the highly reactive hydroxyl radical OH (Gutteridge and Halliwell, 2000;Halliwell and Gutteridge, 1984). The iron-catalyzed Fenton reaction is a major source of OH in biological systems (Fridovich, 1978;Keyer et al., 1995;Liochev, 1999;Meneghini, 1997), though other transition metals (e.g. copper) can also catalyze this reaction. Moreover, oxidative stress can disrupt iron homeostasis. For example, O 2 À can cause release of iron from Fe-S proteins, catalyzing increases in OH levels via the Fenton reaction, further increasing ROS levels (Meneghini, 1997;Puntarulo and Cederbaum, 1996).
The oxidative damage theory predicts that free iron levels and iron homeostasis more broadly are likely to influence ageing rate (Galaris et al., 2008;Mwebi, 2005;Terman et al., 2006). More specifically, it suggests that elevated levels of free iron will increase the rate of ageing due to increased ROS production, while robust control of iron homeostasis will protect against it. It has therefore been speculated that iron chelation treatment might be protective against ageing (Polla et al., 2003;Polla, 1999). It is indeed the case that lowering levels of free iron can result in resistance to oxidative stress. For example, iron chelation results in resistance to H 2 O 2 toxicity in Escherichia coli strains with mutations affecting SOD and oxidative damage repair enzymes (Imlay et al., 1988;Keyer et al., 1995).

The iron storage protein ferritin protects against oxidative stress
Ferritins are iron storage proteins that sequester large amounts of Fe(II), rendering it unavailable for Fenton chemistry, and as such are effective antioxidants (Levi and Arosio, 2004;Vile and Tyrrell, 1993). In vertebrates, ferritins assemble into 24 subunit protein nanospheres, each of which can store up to 4500 iron atoms in its central cavity (Crichton and Ward, 1992). Within this cavity, Fe(II) is oxidized to Fe(III) by the ferroxidase activity of heavy chain (H) ferritin. Here, iron interacts with oxygen, is oxidized to Fe(III) and then migrates to the cavity where it nucleates and aggregates to form the iron core.
C. elegans has two ferritins, FTN-1 and FTN-2 that contain predicted ferroxidase active sites. Expression of ftn-1 and, to a lesser degree, ftn-2 is induced by iron, and mutation of ftn-1 results in hypersensitivity to iron toxicity, consistent with the role of ferritins in iron sequestration (Gourley et al., 2003;Kim et al., 2004). The main site of ftn-1 expression is the intestine, while ftn-2 is expressed in the body wall muscle, hypodermis and pharynx (but not the intestine) (Kim et al., 2004).
In this study, we explore whether iron homeostasis and free iron levels influence resistance to oxidative stress and ageing in C. elegans.

Transgenic line construction
To obtain transgenic C. elegans containing multiple copies of the ftn-1 gene, a 5990 bp PCR product containing the ftn-1 gene (including 3860 bp upstream and 1048 bp downstream of the coding region) was injected into wild-type worms (Evans, 2006). This was co-transformed with the marker construct coel::GFP, which causes bright green fluorescence in the coelomocytes (Miyabayashi et al., 1999), macrophage-like cells found within the nematode pseudocoelom. The ftn-1 gene was amplified from C. elegans gDNA template using the following primers. Forward primer: ftn-1.5in TGTAGGGTTTGATTGTGGTTTG, reverse primer: ftn-1.4in_rev2 AAATTCGGAAATGTCGCAGC.

Ferric ammonium citrate (FAC) and deferoxamine (DF) supplementation
All iron (FAC) and iron chelator (DF) treatments were performed on NGM plates supplemented with FAC (C 6 H 8 O 7 ÁnFeÁnH 3 N) and the iron chelator DF (Desferal)(C 25 H 48 N 6 O 8 ). The appropriate amount of FAC (5-50 mM) and DF (100 mM) was added to the molten NGM agar just prior to pouring the plates.

GFP measurements
To quantify the GFP expression in the Pftn-1::GFP reporter strain, young adult worms were picked in 96 well V-shaped microtitre plates (Greiner) and GFP was measured in a GENios Plus microplate reader (Tecan) at excitation and emission wavelengths of 395 nm and 535 nm, respectively.

EPR sample preparation and measurement
To obtain a measure of free iron in vivo in C. elegans, we used continuous-wave electron paramagnetic resonance (cw-EPR) spectroscopy to detect Fe(III). EPR detects and fingerprints molecular species containing unpaired electrons using a combination of a fixed frequency microwave radiation, n, and an applied magnetic field, B 0 . Transitions appear at magnetic field positions determined by the equation: hn = gm 0 B 0 , where h is Planck's constant, m 0 is the Bohr magneton, and g is the g value or more properly, the g tensor. This has three principle components, the values of which depend on the precise electronic structure and thus are unique for every system. When all 5 d electrons in Fe(III) are in different orbitals, known as a high-spin state, one of these components has a value around 4.3 and is readily observed in the spectrum. EPR has previously been used to detect increased free iron levels in SOD knockout strains of E. coli and yeast (Keyer and Imlay, 1996;Srinivasan et al., 2000), and has been used to estimate free iron levels in C. elegans (Pate et al., 2006).
C. elegans whole worm samples were prepared for EPR as previously described (Pate et al., 2006). Synchronized populations of young adult C. elegans were raised on agar plates, and then collected in M9 buffer, washed several times and centrifuged for 3 min at 4000 rpm at 4 8C. The supernatant was removed and the pellet of worms resuspended in 15% glycerol. Deferoxamine was added to a final concentration of 2 mM, and worms were then incubated for 15 min at room temperature. During this incubation the samples were transferred to EPR tubes using 310 mm glass Pasteur pipettes. EPR tubes were then placed on ice to hasten the accumulation of worms at the bottom of the tube. The supernatant was then removed and the sample flash frozen and stored in liquid nitrogen until use.
We introduced a new step of normalising the EPR spectra to the manganese signal of the worms (Fig. S1). This internal control represented a more reliable way to compare the individual samples with each other than the counting method previously used (Pate et al., 2006). EPR measurements were performed on a Bruker EMXplus spectrometer operating at 9.4 GHz (X-band) equipped with a 4122SHQE resonator, with an Oxford Instruments ESR900 cryostat for measurements at 10 K. Measurements were performed with a magnetic field sweep from 0 to 6000 Gauss, a microwave power of 2 mW, a modulation amplitude of 10 Gauss and a modulation frequency of 100 kHz. The high spin rhombic ferric iron peak was detected at g = 4.3.
Ferrous iron signals are too broad to be detected using EPR, so worms were treated with DF, which converts Fe(II) to Fe(III). Thus, the relative magnitude of the peak at g = 4.3 is a measure of the relative levels of both free Fe(II) and Fe(III). The previous study showed that the free iron pool in C. elegans consists of 50% Fe(II) and 50% Fe(III).

Carbonylated protein detection
Protein oxidation (carbonylation) levels of young adult worms were measured much as previously described (Yang et al., 2007) using an Oxyblot assay kit (Millipore) according to manufacturer's protocol. 1-day-old worms were collected and homogenized using a Bioruptor (Cosmo Bio Co., Ltd., Tokyo, Japan) in 2-ml microcentrifuge tubes containing equal amounts of suspended worms and CelLytic (Sigma) and 1Â protease inhibitors (Roche). The resulting homogenate was spun at 20,000 rpm for 30 min at 4 8C, and the supernatant collected. Samples (15 mg) of total protein extract were incubated for 15 min at room temperature with 2,4dinitrophenylhydrazine to form the carbonyl derivative dinitrophenylhydrazone before SDS-PAGE separation. After transfer onto nitrocellulose, carbonylated proteins were detected by anti-dinitrophenol antibodies. Blots were developed using the SuperSignal West Pico chemiluminescent substrate (Perbio Sciences). Films were scanned and the density of each band or the entire lane was quantified by densitometry using ImageQuant TL (GE Healthcare Europe GmbH).

Lifespan measurements
Age-synchronous worms (hermaphrodites) were transferred to test plates (NGM alone, or supplemented with iron or iron chelator, or RNAi plates) and scored every second day. /glp-4(bn2)/ or /rrf-3(pk1426)/ worms were transferred to fresh plates every day during the first week. In some trials, 5-fluoro-2 0 -deoxyuridine (FUdR) was used to prevent progeny production. In these cases, FUdR was applied to the surface of plates with E. coli lawns already grown, to a final concentration of 10 mM one day prior to the start of the lifespan experiment. The L4 stage was used as t = 0 for lifespan analysis. When a worm failed to move after touching, it was removed from the plate and scored as dead. Loss where animals crawled off the plate, bagged (i.e. internal hatching of embryos) or showed uterine rupture were treated as censored values.

Oxidative stress assays
Worms were transferred as late L4s or young adults to NGM plates containing 10 mM tert-butyl hydroperoxide (t-BOOH) and scored for survival every 1-2 h (Tullet et al., 2008).

RNA-mediated interference (RNAi)
RNAi by feeding was performed as previously described (Kamath et al., 2001). RNAi E. coli feeding clones were derived from the Ahringer Library Kamath and Ahringer, 2003) kindly provided by Dr. Steven Nurrish. Worms were maintained for two generations on the RNAi feeding clones prior to testing. For simultaneous RNAi of ftn-1 and ftn-2, agar plates were spread with 50:50 mixtures of E. coli HT115 bearing the respective feeding plasmids.

Quantitative real-time RT-PCR
RNA was isolated from synchronized worms in the L4/young adult stage using TRIzol (Invitrogen). 1 mg of RNA was transformed into cDNA using the Super-Script TM II RNase H reverse transcriptase kit (Invitrogen). 1 ml of this cDNA was used as template for RT-PCR using ABI SYBR Green PCR mix (Applied Biosciences) and an ABI 7000 RT-PCR machine.

Statistics
Statistical comparisons between survival of nematodes under different treatments were performed with the JMP 7.0.1 programme (SAS Institute Inc.) using the log rank test. For analysis of the GFP-expression measurements, protein carbonyl measurements, brood size and EPR measurements, the Student's T test (two-tailed) was used.

Moderate increases in 'free' iron can cause oxidative stress without reducing lifespan
To test the effect of increased free iron levels in C. elegans, we supplemented their media with iron (ferric ammonium citrate, FAC). We first tested effects of additional iron on relative levels of free iron within the worms using cw-EPR spectroscopy (Fig. S1).
The results implied that supplementation with FAC to 15 mM increased in vivo free iron levels by 328% (p = 0.0004) (Fig. S2A and  B). Individual measurements of worms exposed to 5 mM and 50 mM FAC showed increases in free iron of 34% and 540%, respectively ( Fig. S2A and B, Table S1).
It was previously shown that iron supplementation increases ftn-1 expression, measured as mRNA levels (Gourley et al., 2003). We tested whether increased iron led to elevated expression of a Pftn-1::GFP transcriptional reporter, in C. elegans strain GA631 (Ackerman and Gems, 2012). Exposure to 25 mM FAC increased Pftn-1::GFP fluorescence by 146% (p = 0.0002). Thus, all other things being equal, Pftn-1::GFP expression can provide an indication of free iron levels in vivo.
Increased free iron levels are predicted to increase levels of hydroxyl radicals due to Fenton chemistry. Consistent with this, iron supplementation from L4 stage caused reduced resistance to peroxide stress (tert-butyl hydroperoxide, t-BOOH) in 1-day-old adults, suggesting increased production of tert-butyl hydroperoxyl radicals. Addition of 15 mM FAC caused a decrease in mean survival time of 16.4% (p < 0.0001) in wild-type worms on 10 mM t-BOOH (Fig. S2E, Table S2). Next we tested whether 15 mM iron lead to an increase in protein oxidation in 1-day-old adults. We found that it caused a 99.5% increase (p = 0.02) in levels of protein carbonylation ( Fig. S2C and D).
Our results suggest that supplementation with 15 mM FAC increased levels of free iron in vivo which increases levels of ROS and of protein oxidation in C. elegans. To test whether this could affect ageing, we exposed worms to 15 mM FAC throughout their life. This resulted in a 27.7% decrease (p < 0.0001) in lifespan relative to non-supplemented controls (Fig. S2F, Table S3).
The shortening of lifespan by 15 mM FAC could reflect either an acceleration of the ageing process, or the action of a mechanism unrelated to normal ageing. To probe this, we asked whether there exist levels of iron supplementation that increase peroxide sensitivity and levels of protein oxidation without shortening lifespan (Tables S1-S5). We found that given administration of 9 mM FAC to C. elegans, EPR detected an increase in free iron levels in vivo of 192.8% (p = 0.03) ( Fig. 1A and B, Table S1). 9 mM FAC increased Pftn-1::gfp expression by 75.3% (p = 0.01, Fig. 1C) and also caused a decrease in t-BOOH resistance (Fig. 1F, Table S4) and an 18% increase in protein oxidation (p = 0.02) in one day old adults (Fig. 1D and E), implying that it increases ROS production. However, 9 mM FAC did not shorten lifespan (Fig. 1G, Table S5). This suggests that under standard, non-FAC supplemented culture conditions, free iron levels do not contribute to ageing.

Reducing iron levels increases oxidative stress resistance but not lifespan
If free iron contributes to C. elegans ageing under standard culture conditions, then lowering iron levels ought to increase lifespan. To test this idea, we took two different approaches to reduce irons levels: administration of an iron chelator, and forced over-expression of ftn-1. The iron chelator deferoxamine (DF, 100 mM from late L4 stage onwards) reduced Pftn-1::GFP expression in 1-day-old adults by 32.7% (p = 0.01, Fig. 2A), suggesting that DF lowers in vivo free iron levels in C. elegans. EPR did not detect a change in free iron level after iron chelation (data not shown), but this may reflect a lack of sensitivity in the technique (see Section 4).
The transgene arrays wuEx187 and wuEx188, which carries the injection marker coel::GFP alone, were crossed into a Pftn-1::GFP background to try to gauge whether free iron levels are reduced by ftn-1 over-expression. However, wuEx187 actually increased Pftn-1::GFP expression (data not shown), suggesting the presence of as yet uncharacterized complexity in the regulation of ftn-1 expression.
However, both iron chelation and ftn-1 over-expression resulted in increased peroxide resistance, consistent with reduced free iron levels in vivo. Iron chelation increased survival time on t-BOOH by 14% (p = 0.008) (Fig. 2C, Table S6) and ftn-1 overexpression by 11% (p = 0.005) (Fig. 2I, Table S7). ftn-1 overexpression (but not iron chelation) also reduced protein oxidation levels by 31% (p = 0.03) (Fig. 2B, G, and H). However, neither   Table S8 and S9). These findings are consistent with the view that, under standard culture conditions, free iron levels affect oxidative stress resistance, but not ageing.
Deletion of the sod-4 extracellular superoxide dismutase (SOD) gene was previously shown to enhance both daf-2(m577) Age and Daf-c, suggesting that sod-4 exerts redox effects on insulin/IGF-1 signaling (Doonan et al., 2008). One possibility is that the enhancement of daf-2(m577) Age by ftn-1(0) reflects a similar mechanism. Consistent with this, ftn-1(0) enhanced the Daf-c phenotype, increasing dauer formation at 23 8C from 6% to 23%, although this difference did not reach statistical significance, likely due to the typically high variation between individual Daf-c assays (p = 0.12) (Table S14).
The effects of loss of ftn-1 on lifespan described here are hard to interpret, since depending on genetic background, and mode of abrogation, reduced ftn-1 expression either increases, decreases or has no effect. However, our results suggest that deletion of ftn-1 alters signaling in daf-2 mutants, thereby enhancing Daf-c and Age. They also provide further evidence that ftn-1 contributes to resistance to oxidative stress in a daf-2 mutant background.

Increased free iron can cause oxidative stress without reducing lifespan
In this study we have investigated the role of iron and ironmediated oxidative stress on ageing in C. elegans. We show that high levels of iron cause peroxide sensitivity, oxidative damage and reduced lifespan. Thus, iron at high levels (15 mM FAC or greater) can reduce lifespan, perhaps by increasing oxidative damage. However, many compounds are toxic at high concentrations. The critical question here is: are free iron levels a determinant of ageing under standard culture conditions?
We first asked whether there exist concentrations of iron that increase peroxide sensitivity and oxidative damage without reducing lifespan, and found that this was the case for 9 mM FAC. Our results suggested that at 9 mM FAC, increased Fenton chemistry leads to increased ROS production, increased oxidative damage, but this is not sufficient to shorten lifespan. This in turn suggests that levels of Fenton chemistry at FAC concentrations of 9 mM or below (i.e. standard iron concentrations) do not contribute to ageing.

Iron chelation and ftn-1 over-expression increase resistance to oxidative stress but do not increase lifespan
Overall, our results imply that both iron chelation and ftn-1 over-expression lead to a reduction in levels of free iron in vivo. Both treatments resulted in resistance to peroxide, and ftn-1 overexpression reduced levels of protein oxidation, consistent with lower levels of Fenton chemistry in vivo. However, neither treatment resulted in an increase in lifespan. This suggests that under standard conditions, neither levels of free iron or protein oxidation are critical determinants of ageing. The usefulness of these conclusions depends in part on the degree to which the influence of free iron levels on ageing is similar between nematodes and mammals. However, our findings do not support the view that iron chelation is an effective intervention to slow animal ageing.

Effects of loss of function of ftn-1 on stress resistance and lifespan
Our studies of effects of iron supplementation and treatments aimed at reducing free iron levels both imply that under normal culture conditions free iron is an important determinant of oxidative stress resistance but not of ageing. To test this further, we examined the effect of loss of function of ftn-1 on oxidative stress resistance and lifespan in daf-2(+) and daf-2(m577) mutants. Here, however, largely different results were obtained depending on whether ftn-1 function was abrogated by RNAi or by mutation. In daf-2(+) worms, ftn-1 RNAi reduced resistance to t-BOOH but had no effect on lifespan; by contrast, ftn-1(0) had no effect on resistance to t-BOOH and reduced lifespan. In daf-2(m577) worms, both ftn-1 RNAi and ftn-1(0) caused some reduction in resistance to t-BOOH, but only ftn-1(0) had an effect on lifespan -and it increased rather than decreased it.
How could similar interventions produce such different results? RNAi greatly reduces ftn-1 expression levels, while ftn-1(0) removes it entirely. RNAi appears to reduce ftn-1 function sufficiently to increase free iron, and thereby reduce resistance to t-BOOH. Deletion of ftn-1 may disrupt iron homeostasis to a greater extent, such that viability is reduced. The absence of an effect of ftn-1(0) could imply that the severity of the mutation induces compensatory changes in expression of other genes, such as ftn-2. It could also reflect effects of ftn-1 RNAi on expression of the homologous ftn-2 gene. That ftn-1(0) does not reduce resistance to t-BOOH also suggests that the cause of this reduction in lifespan may not be increased iron levels.
ftn-1(0) does not affect resistance to t-BOOH in daf-2(+) worms, but it markedly reduces it in daf-2(m577) worms. This implies that the high level of overexpression of ftn-1 in daf-2 mutants contributes significantly to their resistance to oxidative stress.
The enhancement of daf-2(m577) Daf-c and age by ftn-1(0) suggests that complete loss of ftn-1 impacts the signaling pathways that control both dauer larva formation and ageing. This effect of ftn-1 may affect either insulin/IGF-1 signaling (IIS) itself, one some other pathway upstream or in parallel to IIS. This could either involve an effect on redox status, as influenced by free iron levels, or some other ferritin-mediated process (e.g. affecting function of signaling-associated ferroproteins).

EPR as a tool for measuring in vivo 'free' iron in C. elegans
In this study, we used cw-EPR spectroscopy as a tool to measure levels of 'free' iron in vivo in C. elegans. This proved an effective means to detect increases in free iron levels resulting from iron supplementation. However, in a number of interventions where more subtle changes in iron levels were expected, no changes in free iron levels were detectable by EPR (data not shown). These interventions included iron chelation, ftn-1 over-expression and daf-2(m577). This suggests that, despite using the most sensitive commercial X-Band spectrometer available, EPR as applied here may be relatively insensitive as a tool for measuring changes in free iron levels close to the normal physiological range. Potentially, a way round this limitation would be to measure free iron content in an even larger numbers of worms or perform the EPR at higher magnetic fields/frequencies.

Conclusions
In this study we have used an animal model (C. elegans) to test the idea that ROS production catalyzed by free iron contributes to organismal ageing. We have used several approaches to manipulate free iron levels, and find that in a number of cases these can influence peroxide resistance and damage levels without affecting lifespan. Severe loss of ftn gene function can reduce lifespan in some contexts, but it remains unclear whether this is a function of increased oxidative stress. The results presented here demonstrate the importance of ferritins in organismal resistance to oxidative stress, which may reflect effects of ferritins on free iron levels. However, they also suggest free iron levels are not a determinant of ageing in C. elegans under standard culture conditions. This is consistent with a number of other studies in recent years that imply that oxidative damage (and perhaps molecular damage) do not cause ageing in this organism.