• Open Access

Quantification of in vivo oxidative damage in Caenorhabditis elegans during aging by endogenous F3-isoprostane measurement



Arjan B. Brenkman, University Medical Centre Utrecht, Department of Metabolic Diseases and Netherlands Metabolomics Centre, Lundlaan 6, 3508 AB, Utrecht, The Netherlands, Tel.: +31-8875-54921; fax: +31-8875-4295, e-mail: a.b.brenkman@umcutrecht.nl


Oxidative damage is thought to be a major cause in development of pathologies and aging. However, quantification of oxidative damage is methodologically difficult. Here, we present a robust liquid chromatography–tandem mass spectrometry (LC-MS/MS) approach for accurate, sensitive, and linear in vivo quantification of endogenous oxidative damage in the nematode Caenorhabditis elegans, based on F3-isoprostanes. F3-isoprostanes are prostaglandin-like markers of oxidative damage derived from lipid peroxidation by Reactive Oxygen Species (ROS). Oxidative damage was quantified in whole animals and in multiple cellular compartments, including mitochondria and peroxisomes. Mutants of the mitochondrial electron transport proteins mev-1 and clk-1 showed increased oxidative damage levels. Furthermore, analysis of Superoxide Dismutase (sod) and Catalase (ctl) mutants uncovered that oxidative damage levels cannot be inferred from the phenotype of resistance to pro-oxidants alone and revealed high oxidative damage in a small group of chemosensory neurons. Longitudinal analysis of aging nematodes revealed that oxidative damage increased specifically with postreproductive age. Remarkably, aging of the stress-resistant and long-lived daf-2 insulin/IGF-1 receptor mutant involved distinct daf-16-dependent phases of oxidative damage including a temporal increase at young adulthood. These observations are consistent with a hormetic response to ROS.


The universal phenomenon of aging is a complicated biological process with a mechanism that lacks overall understanding. The nematode C. elegans served as a paradigm in the field of aging ever since the discovery that single-gene mutations or environmental manipulations could double its lifespan (Kenyon, 2010). For instance, mutations in the insulin/IGF-1 receptor daf-2 were found to extend lifespan up to threefold, a process that is completely suppressed by the Forkhead transcription factor daf-16 (Kenyon, 2010). The insulin/IGF-1 pathway and its role in longevity are highly conserved and a remarkable hallmark of these and other longevity mutants is a phenotype of increased resistance to cellular stress, including oxidative stress (Kenyon, 2010). These findings are consistent with the Free Radical Theory of Aging, which was formulated back in 1956 and explains aging through accumulation of oxidative damage to macromolecules (Harman, 1956).

To minimize damage from ROS, stress protective response pathways have evolved. These are highly conserved and consist of complex systems of both defense and sensing mechanisms, which include superoxide dismutase (SOD), catalase (Ctl), peroxiredoxins, and glutathione peroxidase enzymes (Halliwell, 1999). SODs scavenge the highly reactive and local superoxide (O2•) radicals and dismutate them to hydrogen peroxide (H2O2). H2O2 can readily cross membranes and in turn is further reduced by catalase, peroxiredoxins and glutathione peroxidase. In the presence of transition metal ions, hydrogen peroxide can be converted via a Fenton reaction to another highly reactive radical, OH. In addition to the generation of damage, ROS have been shown to play a critical role in signaling as well and the balance between signaling and damage is a field of intense research.

A major obstacle, however, in understanding the contribution of ROS in the generation of oxidative damage or signaling in C. elegans in aging is the widely recognized methodological difficulties to measure ROS and its derived oxidative damage in vivo in living organisms (Van Raamsdonk et al., 2010; Muller et al., 2007; Gems & Doonan, 2009). Instead, the role of oxidative damage has therefore often been inferred through indirect assays, including resistance to pro-oxidants, knockout or transgenic analysis of ROS scavenging enzymes and respiration assays, each with their own pitfalls (Muller et al., 2007). As production of oxidative damage depends both on the levels of ROS and the scavenging capacity of the organism, there is a clear need for methods to directly quantify oxidative damage and ROS. Through design of unique fluorescent probes, recent studies quantified H2O2 and established the glutathione redox state in C. elegans (Back et al., 2012) and Drosophila (Albrecht et al., 2011) in vivo. Moreover, another study reported a mitochondrial targeted probe quantified by ratiometric mass spectrometry to measure H2O2 in mitochondria in living Drosophila, which illustrates the potential use of mass spectrometry toward quantification of ROS in signaling (Cocheme et al., 2011). Although these studies enable ROS measurements, methods to accurately determine endogenous oxidative damage in C. elegans are lacking.

The predominant mammalian poly-unsaturated fatty acid (PUFA), arachidonic acid (AA, C20:4, ω3), can give rise to F2-isoprostanes through extraction of an electron on one of the four double bonds by both superoxide and hydroxyl radicals (Morrow et al., 1990; Rokach et al., 1997; Pratico et al., 2004). Because of their chemical stability, these nonenzymatic lipid peroxidation products are regarded as the gold standard for oxidative damage quantification in mammalian plasma and urine (Pratico et al., 1998; Kadiiska et al., 2005). Furthermore, F3-isoprostanes derived from another PUFA, Eicosapentaenoic acid (EPA), were also recently identified as markers for oxidative damage in mammals (Gao et al., 2006; Chang et al., 2008; Song et al., 2009). Here, we present an approach for in vivo quantification of endogenous oxidative damage in C. elegans through a mass spectrometry-based stable isotope dilution assay of F3-isoprostanes. The method is characterized as highly specific, sensitive, accurate, and linear over a wide dynamic range and it enabled for the first time absolute quantification of endogenous levels of oxidative damage in C. elegans in vivo. We validated it on C. elegans strains knockout or transgenic for ROS scavenging sod and ctl genes which showed an expected increase in F3-isoprostanes levels at young adulthood as compared with wild-type for most, but not all mutants. Levels of oxidative damage could be assessed and quantified in the cytoplasm, mitochondria, and peroxisomes as well as in a small set of chemosensory neurons specifically. Unexpectedly, whereas mutants of sod-1, encoding the predominant cytoplasmic SOD, showed exceptional sensitivity to the pro-oxidant paraquat (PQ), the levels of oxidative damage at steady state were wild-type-like, underscoring the need to directly quantify oxidative damage in vivo. Moreover, mitochondrial clk-1 and mev-1 mutants which are affected in electron transport chain activity showed increased oxidative damage levels. Oxidative damage increased exponentially in the aging worm starting from the postreproductive phase. Surprisingly, we found that the long-lived daf-2 mutant has strongly increased daf-16 dependent levels of oxidative damage as compared with wild-type controls in young adulthood. However, these levels abated postreproductively and were significantly lower during aging than wild-type nematodes in a daf-16-dependent manner. These findings have important implications for how oxidative damage may contribute to aging.


F3-isoprostanes are produced from EPA in the nematode C. elegans

To identify isoprostanes in C. elegans, we searched for the presence of AA and other PUFA's. In contrast with mammals, we found EPA (C20:5, ω3) to be the predominant C. elegans PUFA (Fig. S1A) in agreement with previous reports (Watts & Browse, 2002). As noted above, EPA contains five double bonds and has recently been shown to give rise to F3-isoprostanes in response to in vivo oxidative damage (Gao et al., 2006; Chang et al., 2008). These observations initiated our search for the presence of F3-isoprostanes as markers of lipid peroxidation in C. elegans. Comparison of chromatograms of C. elegans homogenates to that of copper-induced in vitro oxidized esterified EPA, showed high similarity with corresponding multiple reaction monitoring (MRM) peaks at different retention times which may represent different F3-isoprostane isomers (Fig. 1A). Using tetradeuterated standards of two recently characterized F3-isoprostanes (Chang et al., 2008), which have the same chemical properties as endogenous F3-isoprostanes but with a higher m/z value, two of the endogenous peaks in the C. elegans homogenate were identified as 5-epi-8,12-iso-iPF3α-VI, and 8,12-iso-iPF3α-VI (Fig. 1B), suggesting that C. elegans contains at least two F3-isoprostane isomers [an alternative isoprostane nomenclature is also in use (Taber et al., 1997)]. The identity was confirmed by collision-induced dissociation (CID) mass spectral analysis of labeled and unlabeled F3-isoprostanes (Fig. S1B) and by coinjection of pure synthesized standards (Fig. S1C). To demonstrate that F3-isoprostanes are indeed formed from oxidation of EPA, we subjected eicosapentaenoyl PAF-C16, an EPA-containing glyceryl-3-phosphorylcholine, to in vitro lipid peroxidation and monitored the formation of F3-isoprostanes in time (Fig. 1C). Lipid peroxidation by Cu++ is a simple and reproducible in vitro model for studying oxidative damage to lipids (Maiorino et al., 1995). The mechanism of copper-induced lipid peroxidation is proposed to be as follows: LOOH + Cu++→LOO + Cu+ or LOOH + Cu+→LO + Cu++.

Figure 1.

Identification of two F3-isoprostane isomers in C. elegans. (A) EPA oxidation products are identified in a C. elegans homogenate by LC-MS/MS. Representative MRM chromatograms show peaks corresponding to F3-isoprostanes (m/z 351 > 115) in C. elegans homogenate (top panel) and in an esterified EPA (EPA-PAF-C16) standard, oxidized with CuCl2 (bottom panel). (B) Deuterated F3-isoprostane isomer standards coelute with F3-isoprostane peaks in C. elegans homogenate. Chromatograms showing MRM of d4-labeled 5-epi-8,12-iso-iPF3α-VI, and 8,12-iso-iPF3α-VI (m/z 355 > 115) coeluting with endogenous peaks (m/z 351 > 115). (C) F3-isoprostanes were formed in a time-dependent manner by Cu++ -induced EPA in vitro peroxidation, which is quenched by the antioxidant BHT. EPA-PAF-C16 was incubated with 50-μM CuCl2 for the indicated time points either in the presence or absence of BHT. (D) The antioxidants BHT and Trolox inhibited F3-isoprostane formation in a copper-treated C. elegans homogenate. A C. elegans homogenate was incubated with 50-μM CuCl2 for 24 h either in the presence or absence of 100 μM BHT or 10-mM Trolox. The significance of changes in panels C and D was confirmed by Welch's t-test and are shown as mean ± SD. (**< 0.01, *< 0.05).

The fatty acid peroxyl radicals (LOO) that are formed initiate a chain reaction of lipid peroxidation that results in the formation of chemically stable lipid peroxidation end products (Chamulitrat & Mason, 1989). As expected, F3-isoprostane levels increased in a time-dependent manner (Fig. 1C). To confirm that the F3-isoprostane formation results from oxidation, the antioxidant butylated hydroxytolueen (BHT) was added, which converts peroxyl radicals to hydroperoxides (LOOH) and subsequently terminates the chain reaction. Indeed, F3-isoprostane formation was inhibited by BHT (Fig. 1C). Incubation of C. elegans homogenate with Cu++ also induced lipid peroxidation which was blunted by both BHT and Trolox, a vitamin E-derived antioxidant (Fig. 1D). Thus, F3-isoprostanes can indeed be formed as nonenzymatic lipid peroxidation products in C. elegans homogenates and can be used as markers for oxidative damage. Furthermore, as shown in Fig. S2 and Table S1, this isotope dilution method is highly linear, sensitive, accurate, and precise and can be used for absolute F3-isoprostane quantification (for a detailed method description see Supplementary information).

F3-isoprostanes as in vivo markers of oxidative damage in C. elegans

Because of the lack of sensitive methods to measure in vivo oxidative damage in C. elegans, PQ survival is often used as a readout for sensitivity to oxidative damage (Van Raamsdonk & Hekimi, 2009). PQ undergoes a redox cycling reaction in vivo by accepting a single electron from NADPH after which the electron is donated to molecular oxygen resulting in the formation of O2•, the primary ROS species in the cell (Krall et al., 1988). We subjected young adult N2 worms to a 200-mM PQ solution and compared the F3-isoprostane levels over time with untreated controls (Fig. 2A, left panel). In parallel, the same experiment was performed on a smaller scale to monitor mortality (Fig. 2A, right panel). There is a clear time-dependent increase in F3-isoprostane production in response to PQ treatment as compared with untreated controls (Fig. 2A), confirming that PQ induced in vivo oxidative damage in C. elegans. An alternative protocol for exposure of nematodes to PQ is through supplementation on plate (Van Raamsdonk & Hekimi, 2009). A 4-mM PQ dose was sufficient to increase oxidative damage, which could be blunted through cosupplementation of trolox (Fig. 2B) further indicating that PQ induced in vivo oxidative damage. As biological effects of PQ have been reported using lower PQ concentrations (Yang & Hekimi, 2010), we next tested a PQ concentration range on plate and noted a significant increase in F3-isoprostanes after nematodes were exposed to low PQ concentrations of 0.05 mM (Fig. 2C). To determine the site of PQ-mediated oxidative damage production in the cell, we subjected two SOD mutant strains to a 50-mM PQ solution. sod-1 is the predominant Cu/ZnSOD present in the cytoplasm, which when mutated reduces total SOD activity by 80% (Doonan et al., 2008) whereas sod-2 is the predominant mitochondrial MnSOD (Hunter et al., 1997). Figure 2(D) shows a dramatic time-dependent F3-isoprostane increase in the sod-1 mutant as compared with sod-2 and N2, suggesting that PQ-induced oxidative damage is at least formed in the cytoplasm.

Figure 2.

F3-isoprostanes as in vivo markers of oxidative stress. (A) In vivo F3-isoprostane levels and death in C. elegans increased in a time-dependent manner in response to PQ treatment. N2 worms were treated in solution with 200-mM PQ for indicated times and analyzed for F3-isoprostanes by LC-MS/MS (left panel). In a parallel experiment, survival was monitored (right panel). (B) Oxidative damage induction by PQ is quenched by the antioxidant trolox. N2 nematodes were exposed to 4-mM PQ on plate either in the absence or presence of trolox for 3 days. (C) Exposure to low PQ concentrations induced F3-isoprostanes. N2 nematodes were exposed to indicate concentrations of PQ on plate for 3 days. (D) The sod-1 mutant showed a specific increase in F3-isoprostane levels in response to 50-mM PQ. N2, sod-1, and sod-2 worms were treated in solution with 50-mM PQ for indicated times and were analyzed for F3-isoprostanes (left panel). In a parallel experiment, survival was monitored (right panel). (E) Rotenone exposure predominantly induces F3-isoprostane formation in sod-2 mutants. N2, sod-1, and sod-2 worms were treated in solution with indicated concentration of rotenone for 24 h followed by F3-isoprostane analysis (left panel). Survival was scored in a parallel experiment (right panel). (F) Glucose oxidase (GOX) induces in vivo F3-isoprostanes in N2 worms. Worms were treated in solution with 50 mU GOX for indicated times and were analyzed for F3-isoprostanes (left panel). In a parallel experiment, survival was monitored (right panel). Data were expressed as means ± SD and significance was confirmed by Welch's t-test (**< 0.01, *< 0.05) for panels A, D–F and Student's t-test for panels B,C.

Next, we examined if our developed method enables measurement of oxidative damage from ROS produced in the mitochondria, which are considered to be the main site for ROS production (Paraidathathu et al., 1992). N2, sod-1 and sod-2 mutant strains were exposed to rotenone, a Complex I inhibitor of the mitochondrial electron transport chain which enhances mitochondrial ROS production (Li et al., 2003). An increase in F3-isoprostanes was observed in all three strains (Fig. 2E). In contrast with PQ treatment, however, F3-isoprostanes were predominantly formed in the sod-2 mutant, confirming that the site of ROS production by rotenone mainly occurs in the mitochondria. Next, we exposed N2 worms to an exogenous source of ROS in the form of H2O2, produced by the enzyme glucose oxidase (GOX). In the presence of D-glucose and oxygen, GOX catalyzes D-glucose oxidation into D-glucose-1,5-lactone and a constant flux of H2O2. After 3 h of GOX exposure, a sixfold induction of F3-isoprostane levels was measured which increased further to 8.5-fold induction after 7 h as compared with control worms (Fig. 2F). Thus, endogenous in vivo oxidative damage caused by ROS, produced at different cellular compartments in C. elegans, including the cytoplasm, mitochondria, and exogenously can be readily quantified using F3-isoprostanes.

Oxidative damage in ROS scavenging mutants, mitochondrial mutants, and during aging

H2O2 acts as an important signaling molecule that can diffuse over cell membranes to other compartments. To prevent hydroxyl radical formation by H2O2, C. elegans has three catalases which reduce H2O2 to water and oxygen. ctl genes are expressed at specific cellular locations, including the cytosol (ctl-1) and peroxisomes (ctl-2), whereas the location of ctl-3 gene expression has not been reported. All three mutant strains showed increased in vivo oxidative damage as compared with N2 (Fig. 3A), indicating that oxidative damage originating from hydroxyl radicals produced in the cytoplasm and peroxisomes can be readily detected and quantified by this method. Moreover, it suggests that functional redundancy from other scavenging enzymes does not prevent an increase in oxidative damage. Conversely, transgenic expression of all three ctl genes in the wuIs151(ctl-1 ctl-2 ctl-3 + myo-2::GFP) strain reduced oxidative damage levels to below wild-type levels (Fig. 3B).

Figure 3.

F3-isoprostanes are increased in ROS-scavenging mutants, mitochondrial mutants, and aging worms. (A) Catalase (ctl) mutants showed increased F3-isoprostane levels compared with wild-type. The experiment was performed five times in triplicate and is shown as mean ± SEM (*< 0.05). The significance of changes was confirmed by multiple comparisons testing of differences using Restricted Maximum Likelihood estimation (see Methods). (B) F3-isoprostane levels of ROS-scavenging mutant and transgenic worms. N2, sod-2(gk257); sod-3(tm760), wuIs151(ctl-1 + ctl-2 + ctl-3 + myo-2::GFP), wuIs152[sod-1 (genomic) + pRF4 (rol-6)], mutants were analyzed as in (A). The significance of changes was confirmed by Welch's t-test and are shown as mean ± SD (*< 0.05). (C) An increase in F3-isoprostanes is measured in most SOD mutant strains. N2 plus sod mutants were analyzed as in (A). (D) The mitochondrial mutants clk-1(qm30) and mev-1(kn1) showed increased oxidative damage as compared with N2. Analysis was performed as in (A). The significance of changes was confirmed by Students t-test and are shown as mean ± SD (*< 0.05). (E) Aging of C. elegans is associated with increasing F3-isoprostane levels. Synchronized populations of N2 (left panel) and sod-1 (right panel) animals were analyzed at indicated ages as in (A). Scatter plots show individual data points with the mean indicated by the horizontal line.

In addition to the major cytoplasmic Cu/ZnSOD and mitochondrial MnSOD genes sod-1 and sod-2, respectively, C. elegans contains three additional isoforms: sod-3, a minor mitochondrial MnSOD; sod-4 which is an extracellular Cu/ZnSOD responsible for up to 5% of all SOD activity in adult worms (Doonan et al., 2008); and finally sod-5, a minor cytoplasmic SOD. A significant 10%–30% increase in oxidative damage for most strains was detected as compared with wild-type (Fig. 3C), suggesting that loss of SOD isoforms increases intrinsic oxidative damage. Interestingly, despite the small contribution of SOD-5 to total SOD activity and its expression limited to the ASI, ASK, and ADL neurons of C. elegans (Doonan et al., 2008), a mutation of sod-5 resulted in a dramatic increase in oxidative damage (Fig. 3C). This observation emphasizes the importance of oxidative damage defense and its critical regulation in these neurons. Moreover, whereas the sod-1 mutant showed high sensitivity to PQ (Fig. 2D), no significant increase in oxidative damage was observed (Fig. 3C). Conversely, a mutant with extra sod-1 copies [wuIs152(sod-1 (genomic) + pRF4 (rol-6)] showed no decrease in oxidative damage (Fig. 3B). These findings suggest that the very high cytoplasmic SOD-1 levels are not required to protect against baseline ROS production, but rather protect the organism in specific conditions, such as in the harsh natural C. elegans environment.

To further address the sensitivity of our method, we tested a set of mitochondrial mutants, including clk-1(qm30) and mev-1(kn1) for which an altered ROS metabolism is expected. clk-1 encodes for a ubiquinone biosynthetic enzyme (Van Raamsdonk & Hekimi, 2009) and mutant clk-1(qm30) is marked by a mild respiratory defect and extended lifespan (Van Raamsdonk & Hekimi, 2009). In contrast, mev-1 encodes the succinate dehydrogenase cytochrome b and mutant mev-1(kn1) displays high sensitivity to PQ, decreased SOD activity and is short-lived (Adachi et al., 1998). As shown in Fig. 3D, both clk-1(qm30) and mev-1(kn1) mutants showed increased oxidative damage as compared with N2 in line with their impaired electron transport function. Mutant mev-1(kn1) moreover showed a higher oxidative damage increase than clk-1(qm30) as compared with N2.

Finally, we tested if oxidative damage increased over age. As depicted in Fig. 3(E), an age-dependent increase in oxidative damage was found for both N2 and sod-1 in conditions starting from the postreproductive phase (i.e., after day3/4 postdevelopment). Consistent with our findings, the sod-1 mutant did not show increased oxidative damage levels as compared with N2 while aging (Fig. 3E, right panel).

Oxidative damage in longevity mutants

We next determined oxidative damage levels in a model for longevity by the highly conserved insulin/IGF-1 signaling pathway. In C. elegans, insulin signaling regulates metabolism, development, and longevity (Kenyon, 2010). Mutations in the homologue of the mammalian insulin receptor gene daf-2, arrest development in the dauer stage but weak or temperature sensitive mutations in daf-2 can develop reproductively, but show increased energy storage and longevity (Kenyon, 2010). The lifespan extension caused by daf-2 mutation is hypothesized to at least partially result from increased protection against stress, including oxidative stress (Lee et al., 2003; Kenyon, 2010; Zarse et al., 2012). Both the longevity and stress resistance response from the insulin/IGF-1 pathway requires the activity of the daf-16 gene, which encodes a Forkhead transcription factor (Kenyon, 2010).

To determine oxidative damage in these longevity mutants, we analyzed the levels of F3-isoprostanes in young adults of daf-2(e1370), when the longevity pathway is active (Dillin et al., 2002). As predicted by the Free radical theory of aging, we anticipated this mutant to have lower levels of oxidative damage. However, contrary to expectation, we found a more than 90% increase in oxidative damage in the long-lived daf-2(e1370) mutant which was partially suppressed in the daf-16(mu86); daf-2(e1370) double mutant (Fig. 4A). Next, we determined the ROS damage levels of these mutants during aging, by synchronizing nematodes and harvesting them at various time points up to day 11 of adulthood (Fig. 4B). Oxidative damage of both N2 and the double-mutant daf-16(mu86); daf-2(e1370) increased during aging (Fig. 4C, 4E). Increased levels of oxidative damage were observed after reproductive age for N2 and daf-16(mu86); daf-2(e1370), the latter which furthermore showed a steep oxidative damage increase at day 8, just before most animals collapsed around their maximal lifespan (Fig. 4E). These findings suggest that oxidative damage increases with age. Interestingly, while daf-2(e1370) has much higher oxidative damage as young adult than the N2 or daf-16(mu86); daf-2(e1370) strain (Fig. 4A, D), levels rapidly decreased with age (Fig. 4D). However, at day 11, when about 75% of N2 and about 95% of daf-2(e1370) survived, age-matched levels of oxidative damage in daf-2(e1370) were significantly lower as compared with N2 (Fig. 4D). Thus, the young adulthood of the daf-2 mutant is marked by an unexpected daf-16-dependent temporal phase of oxidative damage, whereas during postreproductive aging, the daf-2(e1370) has significantly less oxidative damage than the N2. Finally, we tested the mutant age-1(hx546). age-1 encodes a phosphoinositide-3 kinase which signals in the Insulin/IGF-1 pathway downstream of daf-2 but upstream of daf-16. The mutant age-1(hx546) also extends lifespan in a daf-16-dependent manner albeit to a lesser extend as the daf-2(e1370) (Kenyon, 2010). Interestingly, at young adulthood age-1(hx546) did not show significantly increased oxidative damage, whereas at day 4 levels of oxidative damage were decreased as compared with N2 (Fig. 4F). These findings suggest that the temporal oxidative damage phase occurs upstream from age-1. Our findings furthermore indicate that oxidative damage levels of young adults do not necessarily reflect levels at older age, which highlights the need for longitudinal analysis to establish the role of oxidative damage during aging.

Figure 4.

Oxidative damage in Insulin/IGF-1 mutant during aging. (A) The daf-2 mutant has increased levels of F3-isoprostanes relative to wild-type, partially rescued by daf-16 at day 1 of adulthood. Young adult N2, daf-2(e1370) and daf-16(mu86); daf-2(e1370) mutants were analyzed for F3-isoprostane levels. The experiments was performed three times in triplicate and is shown as mean ± SEM (*< 0.05). (B) Lifespan curves of Insulin/IGF-1 mutants. All experiments were performed at least twice with >60 animals per condition. Arrows denote time of harvesting for analysis in (C–E). (C, D, E) Oxidative damage of N2 (C), daf-2(e1370) (D) and daf-16(mu86); daf-2(e1370) mutants (E) during aging. The experiment was carried out as in (B) and animals were harvested at indicated time points. Scatter plots show individual data points with the mean indicated by the horizontal line. (F) N2 and age-1(hx546) mutants were analyzed for F3-isoprostane levels at indicated time points. The experiments was performed in sixfold and is shown as mean ± SD (*< 0.05).


Here, we describe the development of a robust, sensitive, and straightforward assay to quantify in vivo oxidative damage in C. elegans through LC-MS/MS-based quantification of endogenous F3-isoprostanes. These chemically stable products are produced by ROS-dependent lipid peroxidation of the predominant PUFA, EPA, in C. elegans. It is unlikely that F3-isoprostanes originated from the provided bacterial food source, Escherichia Coli, as C. elegans synthesizes EPA de novo by desaturation and elongation of saturated fatty acids from its bacterial food source (Watts & Browse, 2002). The formation of endogenous oxidative damage could be detected and quantified in whole animals, specifically in the cytoplasm or in organelles, such as mitochondria and peroxisomes. Moreover, even oxidative damage produced by only a few specific chemosensory neurons in the C. elegans head could be quantified, indicating high sensitivity and broad applicability of the method.

Several methods of oxidative damage quantification have been developed for C. elegans in the past but are generally accepted to lack sensitivity, specificity, or linearity (Muller et al., 2007; Doonan et al., 2008; Gems & Doonan, 2009). These have included application of fluorescent probes either in live animals or isolated mitochondria [e.g., 2′-7′-dichlorofluorescin (H2DCF)] (Harding et al., 2003; Schulz et al., 2007). These methods suffer from methodological difficulties as it has been shown that these molecules can undergo photoreduction and produce reactive species (Marchesi et al., 1999; Rota et al., 1999). Fluorescent probes are prone to autofluorescence and require uptake by the animal or organelle or depend on promoter-driven expression, possibly introducing artifacts (e.g., distribution) and toxicity. These systems furthermore rely on optically accessible systems, which restrict their usage. Oxidative damage assessment in isolated mitochondria can also suffer from artifact introduction in the isolation process (Picard et al., 2010). Finally, protein carbonylation is described as a marker for oxidative damage to proteins. This method relies on postlysis chemical derivatization and antibody-based analysis. Linearity, specificity, and sensitivity are limited for this method and it could suffer from artificial carbonylation backgrounds introduced by oxygen, trace metals, and nucleic acids (Luo & Wehr, 2009). In contrast with these and other reported methods, F3-isoprostanes are produced endogenously and through the use of labeled internal standards and LC-MS/MS detection, they allow for detection with high specificity, sensitivity, and absolute quantification. Another distinguishing feature is that our method allowed thorough validation. Indeed, F3-isoprostane quantification is highly sensitive, detecting oxidative damage in less than 150 animals per sample, highly accurate, precise, and linear over a 2000-fold concentration range. In this respect, it is interesting to note that both clk-1 and mev-1 mutants showed increased oxidative damage levels. Although this finding is consistent with their phenotype of altered electron transport function (Adachi et al., 1998; Van Raamsdonk & Hekimi, 2009), including decreased SOD activity, high PQ sensitivity, and strongly increased levels of protein carbonylation at older age for the mev-1(kn1) (Ishii et al., 2002), at young age both clk-1(qm30) and mev-1(kn1) mutants did not show increased levels of protein carbonylation (Ishii et al., 2002; Yang et al., 2007). This finding could be explained by a higher sensitivity of F3- isoprostane analysis. However, further investigation will be required as these findings could additionally or alternatively reflect a difference in the type of quantified damage (lipid peroxidation vs. protein oxidation).

PQ has been proposed to induce superoxide radicals in vivo (Krall et al., 1988) and time of survival in response to a lethal dose has often been used to determine oxidative damage resistance in C. elegans. We show that PQ-induced oxidative damage is indeed formed in vivo. In response to PQ, the sod-1 mutant showed highest oxidative damage compared with wild-type and sod-2, suggesting that PQ induced oxidative damage predominantly in the cytoplasm. Consistent with its proposed mitochondrial localization, the sod-2 mutant showed highest oxidative damage in response to the Complex I inhibitor rotenone. Surprisingly, in contrast with all other individual sod and ctl mutants, sod-1, which contributes to 80% of all of C. elegans SOD activity (Doonan et al., 2008; Gems & Doonan, 2009), showed no significantly increased steady-state oxidative damage. This could be the consequence of functional redundancy by other ROS-scavenging enzymes. However, sod-1 mutants showed extreme sensitivity to a high dose of PQ, in agreement with previous reports (Doonan et al., 2008; Van Raamsdonk & Hekimi, 2009). Thus, these findings demonstrate that the resistance to induced oxidative damage phenotype does not automatically correlate with steady-state oxidative damage, underscoring the critical need for direct quantification of oxidative damage in vivo. Rather, we now propose that sod-1 predominantly functions when challenged by adverse conditions, such as the harsh natural C. elegans environment. This notion is consistent with recent findings that a mutant which completely lacks SOD activity has normal lifespan yet high sensitivity to exogenous stressors (Van Raamsdonk & Hekimi, 2012). Interestingly, mutation of the second cytoplasmic SOD gene, sod-5, showed a 30% increase in oxidative damage. This is remarkable, given that its expression is predominantly localized to a small set of chemosensory neurons located in the nematode head and its expression contributes to less than 0.5% of all SOD mRNA (Doonan et al., 2008). Further functional studies will be required to determine the consequences of these oxidative damage levels.

Using our method, we could for the first time accurately quantify oxidative damage of several nematode strains during aging. We note that nematodes increased oxidative damage levels with age and, as could be noted for the daf-16(mu86); daf-2(e1370) double mutant, oxidative damage increased exponentially during aging being highest immediately before death of most nematodes. Surprisingly, the stress-resistant daf-2(e1370) insulin/IGF-1 receptor longevity mutant has a significantly higher level of oxidative damage at young adulthood than N2 wild-type nematodes and daf-16(mu86); daf-2(e1370) mutants. This is unexpected since daf-2 mutants showed high resistance to a pro-oxidant challenge (Lee et al., 2003) and, if oxidative damage would be the primary cause of aging, conflicts with its longevity phenotype. However, longitudinal analysis of synchronized aging daf-2(e1370) populations of nematodes revealed that this increased oxidative damage is temporarily, restricted to young adulthood and followed by a decrease in oxidative damage to levels that are significantly lower than the wild-type during postreproductive aging. These findings suggest that despite an initial period of high oxidative damage in young adulthood, impaired insulin/IGF-1 mutants accumulate less oxidative damage during postreproductive aging. Importantly, for all strains tested, oxidative damage was only found to increase during postreproductive aging. Thus, for the study of aging theories in C. elegans, we therefore propose to perform longitudinal analysis, rather than restricting to e.g., young adults only.

The longevity phenotype of lowered insulin/IGF-1 signaling is suppressed by daf-16 (Kenyon, 2010) and we found that the temporal increase in oxidative damage in the daf-2 mutant at least partially requires daf-16, suggesting that this oxidative damage phase contributes to its longevity phenotype. In this respect, it is interesting to note that the antioxidant N-acetyl-L-cysteine was found to decrease daf-2 lifespan but not that of wild-type nematodes (Yang & Hekimi, 2010; Zarse et al., 2012). In fact, our observations are consistent with the concept of mitohormesis which postulates that a temporal increase in ROS species originating from mitochondria triggers an adaptive cellular stress response of increased ROS defense and ultimately, an increased lifespan (Schulz et al., 2007; Zarse et al., 2012). Interestingly, whereas Zarse et al. showed increased hydrogen peroxide and mitochondrial ROS production after acute impairment of daf-2 signaling by RNAi, we extend these observations by showing ROS damage in the constitutive impaired Insulin/IGF-1 signaling mutant daf-2 (e1370). For insulin/IGF-1 signaling, this increase in ROS is thought to be triggered by a metabolic switch in response to decreased glucose levels as a consequence of impaired insulin/IGF-1 receptor function (Schulz et al., 2007; Zarse et al., 2012). Although further studies in C. elegans will be required to determine the contribution of oxidative damage to longevity, with respect to the concept of mitohormesis, our findings suggest that the transient ROS burst initiates early in adult life and importantly, is daf-16 dependent, suggesting that the metabolic switch is governed by this transcription factor. The transient increase in ROS is likely blunted later in life through increased expression of antioxidant genes, including SODs, catalases, glutathione S-transferases, and thioredoxins, in response to this ROS peak. Indeed, daf-16 regulates gene expression of antioxidant genes, including SODs, catalases, glutathione S-transferases, and thioredoxins (Murphy et al., 2003) and is activated by oxidative stress (Essers et al., 2004).

In conclusion, we characterize a novel ultrasensitive approach to quantify in vivo endogenous oxidative damage in C. elegans. Our method follows a short protocol that employs Triple-Quadruple MS technology, readily available in most metabolomics, proteomics, and pharmacology laboratories. This method can easily be scaled up for screening purposes and applied to other model organisms. We anticipate that this method could prove useful toward understanding the contribution of ROS and oxidative damage in pathologies and the biology of aging.

Experimental procedures


Synthetic 5-epi-8,12-iso-iPF-VI, 8,12-iso-iPF-VI, d4-5-epi-8,12-iso-iPF-VI and d4-8,12-iso-iPF-VI were synthesized as previously described(Chang et al., 2008). Stock solutions were prepared in 100% ethanol and stored at −20°C. Chemicals used included Butylated hydroxytolueen (BHT), Trolox and Glucose Oxidase (Sigma Aldrich), Paraquat (Acros organics), and Rotenone (Fluka).

Caenorhabditis elegans strains and growth conditions

sod-1(tm776), sod-2(gk257), sod-3(tm760), sod-4(gk101), sod-5(tm1146), sod-2(gk257);sod-3(tm760), wuIs151(ctl-1 + ctl-2 + ctl-3 + myo-2::GFP), wuIs152[sod-1 (genomic) + pRF4 (rol-6)] were prepared by the Gems laboratory (Doonan et al., 2008). These strains plus N2 (wild-type) and ctl-1(ok1242) ctl-3(ok2042), ctl-2(ok1137), daf-2(e1370), daf-16(mu86); daf-2(e1370), age-1(hx546), mev-1(kn1), and clk-1(qm30) were all obtained from Caenorhabditis Genetics Center (CGC, Minneapolis, USA). Strains were cultured on nematode growth media (NGM) agar at 20°C, containing Escherichia coli strain OP50 as food source. Worms were synchronized by bleaching to collect eggs as previously described (Dillin et al., 2002). During longitudinal analysis nematodes were placed on fresh plates after 4 days. Unless otherwise stated, all worms were collected at day 1 of adulthood.

Lifespan analysis

Lifespan analysis was performed as described previously (Dillin et al., 2002). In brief, synchronized nematodes were grown until L4, when transferred to fresh plates supplemented with 100 μM 2′fluoro-5′deoxyuridine (FUDR; Sigma, St Louis, MO, USA). The prefertile period of adulthood was chosen as T = 0 for lifespan analysis. Strains were grown for at least two generations before lifespan analysis was started. Lifespan analysis of Insulin/IGF-1 mutants was conducted at 25°C. An animal was scored dead when it no longer responded to (mechanical) stimulation. Animals that ruptured, bagged, or crawled off the plates were censored. The first two phenomena occurred on few occasions. Statistical significance was calculated using the log-rank (Mantel-Cox) method.

Sample preparation and LC-MS/MS analysis

Detailed methods for the sample preparation and LC-MS/MS experiments are found in the method description of the Supplementary Information.

In vitro peroxidation of EPA and C. elegans homogenates

Eicosapentaenoyl PAF-C16 (Cayman Chemicals) was stored as a 5-mM stock solution in 100% ethanol under nitrogen at −20°C to ensure minimal auto-oxidation. For analysis, EPA PAF-C16 was diluted to 5 μM in PBS.

Pro-oxidant treatment of C. elegans

Assays were performed either on plate or in liquid solution. For in liquid solution assays, worms were placed into 12-well plates containing 2 ml of either PQ (50 or 200 mM), rotenone (5 or 50 μM), glucose oxidase (50 mU), or glucose (4.5 g/L) dissolved in M9 buffer. After treatment, worms were washed three times with M9 buffer and analyzed. Parallel cultures were used to assay survival. Death was scored when worms failed to respond to mechanical stimulation. For on plate assays, pro- and antioxidants were added to NGM at indicated concentrations, casted into plates and dried in the dark. A fresh OP50 lawn was seeded on plates and allowed to grow overnight where after nematodes were placed on for indicated times.

Quantification and statistical analysis

All data were normalized to the amount of protein per sample, using Bradford (BioRad) analysis. Experiments were done in triplicate (unless stated otherwise). Welch's t-test was used to test for significant differences between groups at P < 0.05 taking into account unequal variance. In case of equal variance, a Student's t-test was performed. For experiments with multiple comparisons of differences between mutants and N2, Restricted Maximum Likelihood (REML) estimation was performed, taking the random variation between experiments into account. The Wald test was used to assess significance of fixed effects; subsequent pairwise comparison of single mutants with N2 was performed using the least significance difference at < 0.05.


A.B.B was cofinanced and E.C.S and C.L. were financed by grants from the Netherlands Metabolomics Centre which is part of the Netherlands Genomics Initiative / Netherlands Organisation for Scientific Research. We thank Tobias Dansen and Boudewijn Burgering and members of the Brenkman lab for stimulating discussion and critical reading of the manuscript. The Caenorhabditis Genetic Center (University of Minnesota, Minneapolis) is acknowledged for providing nematode strains.