• Open Access

Carbonylated proteins are eliminated during reproduction in C. elegans

Authors

  • Jérôme Goudeau,

    1. Ecole Normale Supérieure de Lyon – CNRS – Université de Lyon Claude Bernard, Molecular Biology of the Cell Laboratory/UMR5239, 46, Allée d’Italie, 69364, Lyon Cedex 07, France
    Search for more papers by this author
  • Hugo Aguilaniu

    1. Ecole Normale Supérieure de Lyon – CNRS – Université de Lyon Claude Bernard, Molecular Biology of the Cell Laboratory/UMR5239, 46, Allée d’Italie, 69364, Lyon Cedex 07, France
    Search for more papers by this author

Hugo Aguilaniu, Ecole Normale Supérieure de Lyon – CNRS – Université de Lyon Claude Bernard, Molecular Biology of the Cell Laboratory/UMR5239, 46, Allée d’Italie, 69364, Lyon Cedex 07, France. Tel.: +334 7272 8794; fax: +334 7272 8080; e-mail: hugo.aguilaniu@ens-lyon.fr

Summary

Oxidatively damaged proteins accumulate with age in many species (Stadtman (1992) Science257, 1220–1224). This means that damage must be reset at the time of reproduction. To visualize this resetting in the roundworm Caenorhabditis elegans, a novel immunofluorescence technique that allows the detection of carbonylated proteins in situ was developed. The application of this technique revealed that carbonylated proteins are eliminated during C. elegans reproduction. This purging occurs abruptly within the germline at the time of oocyte maturation. Surprisingly, the germline was markedly more oxidized than the surrounding somatic tissues. Because distinct mechanisms have been proposed to explain damage elimination in yeast and mice (Aguilaniu et al. (2003) Science299, 1751–1753; Hernebring et al. (2006) Proc Natl Acad Sci USA103, 7700–7705), possible common mechanisms between worms and one of these systems were tested. The results show that, unlike in yeast (Aguilaniu et al. (2003) Science299, 1751–1753; Erjavec et al. (2008) Proc Natl Acad Sci USA105, 18764–18769), the elimination of carbonylated proteins in worms does not require the presence of the longevity-ensuring gene, SIR-2.1. However, similar to findings in mice (Hernebring et al. (2006) Proc Natl Acad Sci USA103, 7700–7705), proteasome activity in the germline is required for the resetting of carbonylated proteins during reproduction in C. elegans. Thus, oxidatively damaged proteins are eliminated during reproduction in worms through the proteasome. This finding suggests that the resetting of damaged proteins during reproduction is conserved, therefore validating the use of C. elegans as a model to study the molecular basis of damage elimination.

Introduction

Most living species use diatomic oxygen as a final electron acceptor to produce adenosine triphosphate through the mitochondrial respiratory chains. In eukaryotes, respiratory chains consist of redox cascades that couple fluxes of electrons through the respiratory complexes to the translocation of protons across the inner mitochondrial membrane (Mitchell, 1961). ATP synthase (complex V) then converts ADP to ATP using the energy from the resulting biochemical gradient (Boyer, 1975). This redox cascade is often initiated by the oxidation of NADH or FADH2, and it is ended by the reduction of diatomic oxygen into water by complex VI. However, partial reduction of diatomic oxygen can lead to the formation of a superoxide radical, which is a reactive oxygen species (ROS; McCord & Fridovich, 1968, 1969). Although this event is rare, it can be harmful, and a handful of detoxifying enzymes are therefore specifically dedicated to ROS removal (McCord & Fridovich, 1968, 1969). Despite the activation of these defense mechanisms, biomolecules still get oxidized by ROS (Stadtman, 1992; Levine, 2002), and oxidative modifications can profoundly alter protein function and activity (Yan et al., 1997; Grune et al., 2004). These findings led to the theory that the accumulation of oxidative damage with age (Stadtman, 1992) may be a cause of aging (Harman, 1956). However, unequivocal experiments validating this theory have not yet been performed (Gems, 2009; Gems & Doonan, 2009).

Protein carbonyls are the products of metal-catalyzed oxidative attacks on specific amino acid chains (Stadtman & Levine, 2000), and they have been widely used as an aging marker for several reasons: first, the process of protein carbonylation is irreversible (Dalle-Donne et al., 2003), rendering it more threatening than other reversible oxidative modifications such as cysteine disulfide bond formation. Also, carbonylated protein levels increase with age in many species such as yeast (Aguilaniu et al., 2003), flies (Orr & Sohal, 1994), worms (Adachi et al., 1998) and mammals (Stadtman, 1992) and are associated with many age-related diseases (Dalle-Donne et al., 2003).

The observation that carbonylated protein levels increase with age also implies that these molecular insults do not transmit to the offspring, considering that the progeny are not progressively shorter lived than the parents. While meiosis rids the cells of corrupted nuclear DNA through the process of genetic recombination (Medvedev, 1981; Lombard et al., 2005), the mechanisms ensuring the elimination of corrupted mtDNA molecules, which do not undergo recombination in the germline (Elson & Lightowlers, 2006), and harmful cytosolic components during animal reproduction are still largely unknown (Hernebring et al., 2006; Fan et al., 2008; Stewart et al., 2008).

Unicellular organisms such as Saccharomyces cerevisiae or Schizosaccharomyces pombe segregate toxic damage away from the offspring (Aguilaniu et al., 2003; Erjavec et al., 2008; Liu et al., 2010). In these models, carbonylated proteins form aggregates that are segregated at the time of cytokinesis. Consequently, daughter cells are born free of oxidative products, which instead accumulate in the mother cell compartment. Recent work revealed that this segregation process encompasses a retrograde flow of aggregates of oxidatively damaged proteins from the daughter to the mother cell (Liu et al., 2010). This process involves a series of cytoskeletal components named the ‘polarisome’ (Liu et al., 2010) and depends on the lifespan modulator Sir2p (Aguilaniu et al., 2003; Erjavec et al., 2008; Liu et al., 2010). The uneven distribution of oxidized proteins in yeast does not depend on differential protein turnover rates in the daughter and the mother compartments (Aguilaniu et al., 2003). In 2006, a pivotal study revealed that carbonylated proteins are also eliminated during embryogenesis in mice, demonstrating that similar processes occur in the animal kingdom (Hernebring et al., 2006). However, the process by which damaged proteins are effectively cleared from the cytoplasm appears to be distinct from what has been observed in yeast. First, the authors did not observe any decrease in the concentration of damaged proteins until the late stages of development. At the blastocyst stage, high levels of damaged proteins were observed in the inner mass, while cells that underwent differentiation harbored markedly reduced levels of carbonylated proteins (Hernebring et al., 2006). Second, the authors observed that the activity of the 20S proteasome also increased abruptly at this stage (Hernebring et al., 2006), although the mechanism involved in this process has not yet been fully elucidated. The authors hypothesized that these two processes might be causally linked, which correlates with the fact that although other degradation pathways can eliminate oxidized proteins (Chondrogianni et al., 2002; Kaushik & Cuervo, 2006), the 20S proteasome is responsible for the degradation of the majority of these proteins (Grune et al., 1997, 2003; Friguet, 2006; Petropoulos & Friguet, 2006).

In this study, a new detection technique that allows the visualization of carbonylated proteins in situ in the nematode Caenorhabditis elegans was developed, and it revealed that the germline is more oxidized than surrounding somatic tissues. However, carbonylated proteins are eliminated during oocyte maturation to produce embryos free of damage. Our results show that the removal of protein carbonyls requires the activity of the proteasome but not the presence of SIR-2.1. The data presented here suggest that C. elegans can be used as a powerful and reliable model to study the mechanisms of damage elimination during reproduction in animals.

Results

Carbonylated protein levels increase with age in wild-type worms

Ishii and collaborators have shown that protein carbonyl levels increase by 2-fold over the lifespan of the roundworm C. elegans (Adachi et al., 1998; Ishii et al., 2002). Because worms are fertile until day 5 of their adult life, the sensitivity of the DiNitroPhenylHydrazine (DNPH)-based immunodetection of protein carbonyls was tested to ensure that it could detect changes over shorter periods of time (Levine et al., 1994). Carbonylated proteins were measured in extracts isolated from worms at days 2 and 5 of adulthood, and their levels were found to increase by approximately 25% during this period (Fig. 1). Quantitatively, this is in line with previous reports that have performed similar analyses over longer periods of time (Adachi et al., 1998; Ishii et al., 2002). This validates the use of this method to detect eventual removal during reproduction. There are several models to explain how toxic product levels are kept low in the progeny (Fig. 2). In the first one, oxidatively damaged proteins never accumulate within the germline. This model suggests that the germline is immortal and protected against environmental hazards (Medvedev, 1981) and that oxidized proteins only occur in somatic tissues (Fig. 2A). Alternatively, the germline accumulates oxidized products but, unlike somatic tissues, it retains the capacity to discard, damage and produce ‘rejuvenated’ offspring (Fig. 2B, C). This scenario corresponds to what is described in mice (Hernebring et al., 2006). Results obtained with mice suggest that this process might occur during development (Hernebring et al., 2006). To distinguish between these models, the aim of this work is to detect carbonylated proteins in situ in the adult worm.

Figure 1.

 Protein carbonylation levels increase during the course of the reproductive period of adult wild-type worms. (A) Example of the detection of carbonylated proteins in wild-type adult worms at day 2 and day 5 of adulthood. Equal amount of protein was loaded and the signal was normalized to the ACT-1 band shown at the bottom. (B) Quantification of Western blot analysis obtained from 4 independent samples. We find that the overall carbonylation level in adult worms is increased by 26% between day 2 and day 5 (unpaired two-tailed t-test P-value=0.001).

Figure 2.

 Hypothetical models of the resetting of oxidatively damaged proteins during reproduction. (A) In our first model, the germline is protected from oxidative damage where they never accumulate in contrast to somatic tissues. Alternatively, oxidative damage can be present in both somatic tissues and the germline, but the germline has the capacity of ridding itself from damage. This process can occur either abruptly (B) or progressively (C).

Detection of carbonylated proteins in situ in C. elegans

To detect carbonylated proteins in situ in C. elegans, we combined the freeze-crack procedure with immunostaining protocols described in yeast (Aguilaniu et al., 2003) and mice (Hernebring et al., 2006) that we adapted to worms (see Data S1). Worms were first treated with DNPH and then subjected to freeze-cracking according to classic protocols, followed by fixation by successive rounds of incubation in methanol (5 min) and acetone (5 min) at 4°C. Immunostaining was then performed as usual (Duerr, 2006). To avoid artifacts, the inclusion of several controls for each staining experiment is important. First, omitting the DNPH treatment should abolish or largely attenuate the signal, ensuring that the primary antibody does not detect non-specific targets (Fig. 3B). Second, the absence of the primary antibody should result in the absence of signal, indicating that the secondary antibody has no cross-species detection (Fig. 3C). Finally, the signal should increase in response to treatment with hydrogen peroxide, a powerful oxidant in all tissues (soma and germline; Fig. 3D). Data presented in this report were all obtained with satisfactory controls. Similarly to most immunodetection protocols, this technique is semiquantitative, and samples cannot be reliably compared unless prepared and treated simultaneously. The results surprisingly showed that the germline was markedly and reproducibly more oxidized than the surrounding somatic tissues (Fig. 3A, E). This is in clear contrast with the notion that the germline should be specifically protected from the environment (Medvedev, 1981) but is consistent with the data obtained with mice (Hernebring et al., 2006). Somatic tissues were poorly oxidized, and we were unable to distinguish cellular or subcellular structures (Fig. S1). Although immunostaining protocols are only semiquantitative, we stained worms at day 2 and day 5 of their adult life. These animals were prepared and stained at the same time to avoid artifacts. We observed that, with age, the concentration of carbonylated proteins tended to increase in the germline and the soma (the head and the tail; Fig. S2). This further validates our method and indicates that all tissues contribute to the overall increase described in Fig. 1. These observations clearly exclude our first model to explain how protein oxidation levels are low in the young (Fig. 2A).

Figure 3.

 Carbonylated proteins are reset during oocyte maturation in Caenorhabditis elegans. (A) Distribution of oxidatively damaged proteins in a day 2 wild-type adult worm. Upper panel shows that carbonylated proteins on a section obtained by freeze-cracking, while the lower panel shows DAPI staining of the corresponding animal. The quantification of the signal along the black line 1–2 on the upper panel is depicted on panel (E) by the green line (dotted black line shows an average of the signal). It shows that the germline is clearly more oxidized than surrounding somatic tissues. Within the germline, protein carbonyl levels are uniformly distributed until oocyte maturation. The green line on panel F depicts by the quantification of the signal along the black line 3–4 (dotted black line shows an average of the signal). Levels of oxidized proteins suddenly drop during oocyte maturation. Positions of the spermatheca and of the oocytes are indicated in white below the animal. s indicates the spermatheca and the position of the oocytes are marked (-1, -2, etc.) (B) Similar staining was performed omitting the DiNitroPhenylHydrazine treatment. The signal was quasi null indicating that the primary antibody is specific to derivatized proteins. (C) When animals were not incubated with the primary antibody, they also harbor no signal, showing that our secondary antibody does not cross react with C. elegans. (D) Finally, we detect a clear increase in the signal when animals were treated with the powerful oxidant H2O2 (8 mm) prior to the immunostaining. Taken together, our results show that our detection technique is valid and that the resetting observed is real. These pictures are representative of all positive experiments realized in the laboratory. This experiment was repeated more than five times on independently prepared samples. Statistical analysis of these pictures can be found in Table S1.

Oxidatively damaged proteins are eliminated during oocyte maturation

Our results show that carbonylated proteins levels drop abruptly within the gonad, before fertilization (Figs 3A, F and S3). The resetting occurs in the last two or three mature oocytes in relation to the spermatheca (Figs 3A, F and S3). This observation excludes our third model and fits the second model (Fig. 2B). This finding contrasts those reported for mice, where this process occurs later during development (Hernebring et al., 2006). More precisely, the resetting occurs during oocyte maturation, soon after cellularization has been completed. In the worm, oocytes are incompletely cellularized until the first stage of oocyte maturation, which suggested that the detection of low levels of carbonylated proteins could potentially be the result of altered accessibility of the antibodies because of the presence of an extra membrane in these cells. However, treatment with oxidizing agents efficiently elevated the levels of damaged proteins to a similar extent in all tissues (germline and soma), including cellularized oocytes (Fig. 3D), indicating that the antibodies could efficiently reach these cells. In addition, three different treatments were found to abolish the resetting of carbonylated proteins in these cells (Fig. 6). Taken together, our data show that the resetting of carbonylated proteins observed in cellularized oocytes is not because of the presence of an extra plasma membrane. Next, we tested the impact of several germline-related processes on the elimination of carbonylated proteins.

Figure 6.

 Protein carbonyls resetting in Caenorhabditis elegans requires the proteasome activity in the germline, but not the presence of SIR-2.1 Upper panels shows carbonylated proteins in age-matched (day 2 of adulthood) feminized mutants fem-1(hc17) (A), animals in adult reproductive diapause for 2 days (B), sir-2.1(ok434) mutants (C), rrf-1(pk1417) animals treated with either pas-6 RNAi (D) or pbs-1 RNAi (E). Positions of the spermatheca and of the oocytes are indicated in white below the animals. s indicates the spermatheca and the position of the oocytes are marked (-1, -2, etc.). Mid panels show DAPI staining of the corresponding animals. In the case where resetting was detected, a line (1–2) is drawn across oocytes where resetting was detected (-2 or -3). When no resetting could be detected, a line was drawn from oocyte -4 to oocyte -2. Signal intensity was then measured along this line. On the lower panels, the relative quantification of the signal along these lines is depicted in green. The dotted black line represents an average line. While the resetting of carbonylated proteins is normal in the sir-2(ok434) mutants, it is no longer observed in rrf-1(pk1417) animals treated with either pbs-1 RNAi or pas-6 RNAi. Although pas-6 RNAi elevates overall levels of carbonylated proteins, pbs-1 does not, confirming that the proteasome activity is required for a proper resetting of carbonylated proteins. Feminized mutants fem-1(hc17) also fail to remove oxidized proteins from oocytes. This suggests that fertilization is required for efficient resetting. Finally, oocytes of animals that have entered adult reproductive diapause (Angelo & Van Gilst, 2009) for 2 days are oxidized. Because these animals contain sperm, this observation excludes sperm as the signal for resetting. The pictures shown are representative of experiments realized at least twice independently. Statistical analysis of these pictures can be found in Table S1.

Yolk trafficking is not required for the resetting of carbonylated proteins

Once the process of cellularization is completed, yolk is imported into oocytes through the RME-2 receptor (Grant & Hirsh, 1999). Yolk consists of lipids, carbohydrates and proteins. The vitellogenins are the protein components of yolk. They are first synthesized in the intestine but are quickly secreted in the pseudoceolom. The RME-2 receptor allows yolk to be endocytosed into oocytes (Kimble & Sharrock, 1983). The vitellogenin homolog YP170 is encoded by vit-2, and a GFP-tagged version of the VIT-2 protein allows the visualization of yolk in vivo. To ascertain that the resetting of carbonylated proteins occurred at the stage of oocyte cellularization, carbonylated proteins were stained in worms expressing a VIT-2::GFP. Convincingly, the resetting of carbonylated proteins coincided exactly with the endocytosis of yolk, confirming that the resetting of protein oxidation indeed occurs in maturing oocytes (Fig. 4) and prompting testing of whether the proper removal of carbonylated proteins along the gonad arm requires yolk endocytosis within the maturing oocytes. Vitellogenins are redox-sensitive proteins, and vitellogenin-6 is a preferential target for carbonylation in aging nematodes (Nakamura et al., 1999). The observation that vitellogenin-6 becomes increasingly oxidized over time (Nakamura et al., 1999) implies that vitellogenins are not strongly oxidized in young animals or in oocytes. In addition, the oxidation of vitellogenins, which can also act as apoB-like proteins, was suggested to promote germline development (Shibata et al., 2003). Because vitellogenins are the most abundant proteins in maturing oocytes, their average level of oxidation may very well affect the total levels of carbonylated proteins and explain our observation. To test this possibility, the endocytosis of yolk into matured oocytes was disrupted by inactivating the yolk receptor, RME-2, by RNAi. Surprisingly, this treatment effectively prevented yolk from entering the oocyte (Fig. 5B), but it did not alter the elimination of damaged proteins (Fig. 5A). These findings suggest that the presence of yolk within maturing oocytes is not required for the resetting of carbonylated proteins.

Figure 4.

 Carbonylated proteins are reset concomitantly with yolk endocytosis. Panel A shows a staining of carbonylated proteins in VIT-2::GFP worms. Positions of the spermatheca and of the oocytes are indicated in white below the picture. s indicates the spermatheca and the position of the oocytes are marked (-1, -2, etc.). Panel B show the remaining GFP signal after the immunostaining treatment. As previously described, yolk is actively imported in maturing oocytes. Panel C shows a merge picture of carbonylated proteins with VIT-2::GFP. From this data, it is clear that yolk import and resetting of carbonylated protein occur concomitantly. In panel D, a relative quantification of the GFP and the carbonyl signal (green and red lines, respectively) along the line (1–2) is shown. This picture is representative of experiments realized at least twice independently. Statistical analysis of these pictures can be found in Table S1.

Figure 5.

 Yolk is not required for the resetting of carbonylated proteins during oocyte maturation. Panel A shows carbonylated proteins in animals expressing VIT-2::GFP and treated with rme-2 RNAi. Positions of the spermatheca and of the oocytes are indicated in white below the animal. s indicates the spermatheca and the position of the oocytes are marked (-1, -2, etc.). As depicted in the insert, quantification of the signal along the line (1–2) shows a clear resetting of the signal). Importantly, as shown in panel B, rme-2 RNAi was efficient as witnessed by the absence of GFP signal is absent from mature oocytes. This picture is representative of experiments realized at least twice independently.

The elimination of carbonylated proteins is linked to oocyte production rate

In the next experiments, we tested the fem-1(hc17) feminized mutants that lack sperm and do not fertilize their oocytes (Kadandale & Singson, 2004). We show that day 2 fem-1(hc17) mutants failed to remove oxidized proteins from matured oocytes (Fig. 6A and Table S1). This data suggests that the process of damage elimination is not coupled to ovulation per se. Rather, fertilization seems to be more important. Animals under adult reproductive diapause (arrested by nutrient deprivation at the L4 stage (Angelo & Van Gilst, 2009)) were also analyzed after 2 days of arrest. These animals do not lay eggs, contain arrested oocytes and functional sperm. Animals that entered adult reproductive diapause (Angelo & Van Gilst, 2009) failed to remove carbonylated proteins from their oocytes, suggesting that this process is also independent of the presence of sperm. Taken together, our data suggest that maintaining the oocyte production rate, linked to fertilization, is key to the removal of oxidized proteins (Fig. 6B and Table S1).

Proper resetting of carbonylated proteins requires the activity of the proteasome and does not require SIR-2.1

The next experiments were designed to test whether the resetting of carbonylated proteins in C. elegans shared common mechanistic features with other organisms for which damage elimination during reproduction has been reported. Two distinct mechanisms have been described in yeast and mice (Aguilaniu et al., 2003; Hernebring et al., 2006; Liu et al., 2010). In yeast, the asymmetric distribution of damaged proteins is mediated by the cytoskeleton, and it requires the presence of the longevity regulator Sir2p (Liu et al., 2010). To test whether this mechanism was relevant in nematodes, oxidized proteins were labeled in mutant animals that lack the worm homolog to the yeast SIR2, SIR-2.1. Similar to wild-type, the sir-2.1(ok434) mutant animals removed carbonylated proteins from maturing oocytes (Fig. 6C and Table S1), indicating that the resetting of oxidatively damaged proteins does not require the presence of SIR-2.1 in C. elegans. This is in contrast with findings in both Saccharomyces cerevisiae (Aguilaniu et al., 2003) and Schizosaccharomyces pombe (Erjavec & Nystrom, 2007).

In mice, the elimination of damaged proteins occurs during development, and this process is accompanied by a sudden increase in the activity of the 20S proteasome (Hernebring et al., 2006). To examine whether these two processes are linked in C. elegans, the effect of proteasome activity on the elimination of carbonylated proteins was assessed by inhibiting proteasome activity using RNAi against pbs-1 and pas-6 (Fig. S4). PBS-1 encodes a beta subunit of the 20S proteasome, and PAS-6 encodes a type 1 alpha subunit of the 26S proteasome. Although RNAi against pbs-1 or pas-6 reduced proteasome activity by 40 and 55%, respectively (Fig. S4), these treatments induced larval arrests, indicating that the normal activity of the proteasome is essential for development. The treatment of adults resulted in sick animals that showed aberrant morphological features, thus compromising the analysis. This treatment also sharply decreased lifespan (Fig. S5A). To bypass this issue, proteasome inhibition was limited to the germline by using rrf-1(pk1417) mutant animals in which the RNAi machinery is only active in the germline. RRF-1 encodes an RNA-directed RNA polymerase (RdRP) homolog required for somatic, but not germline RNAi (Sijen et al., 2001). Treating rrf-1(pk1417) animals with RNAi directed against pas-6 from the L1 stage ultimately resulted in an enlarged maturing oocyte (Fig. 6D). Although few eggs were laid, no viable progeny could be obtained. In contrast, when rrf-1(pk1417) animals were treated with RNAi against pbs-1 from the L1 stage, oocyte size was not affected, few eggs were laid but viable progeny could be obtained. Both of these treatments fully abolished the resetting of carbonylated proteins (Fig. 6D,E and Table S1). Our data indicate that eliminating carbonylated proteins during the course of reproduction in worms requires the activity of the proteasome in the germline. Because a similar mechanism exists in mice (Hernebring et al., 2006), our data suggest that C. elegans is a valid model for studying the resetting of cytoplasmic damage in the course of reproduction.

The effect of protein carbonylation resetting on egg laying, hatching and lifespan

To monitor the effect of inheriting oxidatively damaged proteins, the offspring of rrf-1(pk1417) mutant worms treated with pbs-1 RNAi (as mentioned above, pas-6 RNAi leads to sterility) were analyzed. First, we observed that rrf-1(pk1417) mutants subjected to pbs-1 RNAi from the L1 stage exhibited a 20-fold decrease in egg laying compared to animals grown on the empty vector (Fig. 7A). Next, the fate of eggs laid by rrf-1(pk1417) animals grown on either pbs-1 RNAi or the empty vector (eggs from these two different cohorts were deposited on plates seeded with empty vector carrying bacteria) were analyzed. While all eggs gave rise to viable progeny under control conditions, only 15% of the eggs laid by rrf-1(pk1417) animals grown on pbs-1 RNAi hatched normally (Fig. 7B). The observation that oocytes produced by rrf-1(pk1417) animals grown on pbs-1 RNAi contain high levels of carbonylated proteins implies that the inheritance of oxidatively damaged proteins might explain, at least in part, the low rate of both egg laying and hatching in these animals. However, we cannot exclude the possibility that pbs-1 RNAi treatment of the parents affects hatching of the progeny through other means. Post-hatching (early L1 larval stage) overall levels of carbonylated proteins were compared in animals issued from highly or normally carbonylated oocytes. These levels were found to be comparable, indicating that the initial high load of damaged proteins had been removed (or diluted) during the early developmental stages (Fig. 7C). Finally, the lifespan of the progeny of rrf-1(pk1417) animals grown on either pbs-1 RNAi or the empty vector were compared. When this analysis was performed on control (empty vector) bacteria, the lifespan of the two cohorts was similar (Fig. 7D).

Figure 7.

 The progeny of rrf-1(pk1417) animals treated with pbs-1 RNAi is largely lethal. Panel A shows that the number of eggs laid by rrf-1(pk1417) animals treated with either pbs-1 RNAi or the empty vector. Eggs from an average of 30 animals were analyzed. Panel B shows that the proportion of eggs (laid by rrf-1(pk1417) animals treated with either the empty vector or pbs-1 RNAi) that hatch. Hatching was analyzed on an average of 30 eggs. Panel C shows the overall levels of protein carbonyl levels in early L1 larvae (immediately after hatching) issued from oxidized or non-oxidized oocytes. These levels are similar suggesting that most of the developmental elimination of protein carbonylation occurs prior to hatching (unpaired two-tailed t-test P-value=0.89). Panel D shows that the lifespan of the progeny of rrf-1(pk1417) animals treated with either pbs-1 RNAi or empty vector are similar (mean lifespan of 16.5 and 17.5 days, respectively, P = 0.88). These lifespan were realized on the empty vector. Lifespan analyses were performed at least twice independently. P values were calculated using the log rank (Mantel–Cox) analyses.

Taken together, our data suggest that animals in which the proteasome was partially deactivated in the germline produce oxidized oocytes and lay few eggs that mostly fail to hatch. However, when the progeny survive, they exhibit a normal level of protein oxidation as early as the L1 stage (Fig. 7C), suggesting that damaged proteins are either eliminated or diluted out during very early embryonic development.

Finally, the lifespan of these animals is normal. Our data therefore show that animals issued from oxidized oocytes can enjoy a normal lifespan and supports the notion that most of the developmental protein quality control occurs prior to hatching in the worm.

Discussion

Protein carbonyls are reset during oocyte maturation in C. elegans

In this work, a novel technique for the detection of carbonylated proteins in situ in the roundworm C. elegans is described for the first time, and its use leads to two important findings. First, levels of oxidatively damaged proteins were found to be higher in the germline than in surrounding somatic tissues (Fig. 3). This contradicts the notion that the germline is protected against environmental hazards but is in line with previous findings (Hernebring et al., 2006). There are several possible explanations for this surprising discovery. In the first place, Reactive Oxygen Species act as necessary second messengers mediating cell proliferation in mammalian cells, as complete scavenging of ROS leads to cell cycle arrest (Sarsour et al., 2009), and the germline is the only dividing tissue in adult C. elegans. In addition, increasing ROS levels by RNAi against SOD-1 promotes germline development in C. elegans (Shibata et al., 2003). Taken together, these data favor the notion that ROS levels are higher in the germline of C. elegans, which may in turn lead to higher steady states of oxidatively modified biomolecules in this tissue. Further studies and developing valid ROS detection techniques in the worm are necessary to critically test this hypothesis.

A resetting mechanism that requires normal activity of the proteasome in the germline

Second, a sudden drop in carbonylated protein levels was observed during oocyte maturation; the decrease in oxidatively damaged proteins occurred concomitantly with the endocytosis of yolk through the RME-2 receptor, which marks the beginning of oocyte maturation (Fig. 4). Although these two processes occur simultaneously, the deactivation of rme-2 by RNAi prevents yolk accumulation in maturing oocytes but does not affect the elimination of oxidatively damaged proteins (Fig. 5). The mechanism mediating the removal of carbonylated proteins was investigated to assess its similarity to those observed in other species studied thus far. Unlike what was shown in the yeasts S. cerevisiae and S. pombe (Aguilaniu et al., 2003; Erjavec et al., 2008), the longevity-ensuring gene SIR-2.1 is not involved in the elimination of oxidatively damaged proteins in C. elegans (Fig. 6C). RNAi against either pbs-1 or pas-6 showed that proteasome activity in the germline is required for this process (Fig. 6D, E and Table S1), but our data does not exclude the involvement of other activities (Kaushik & Cuervo, 2006). Based on results obtained with mice showing an increase in proteasome activity in cells that eliminate oxidized proteins (Hernebring et al., 2006), we hypothesized that maturing oocytes may harbor a higher proteolytic activity in C. elegans than the rest of the germline (Fig. 8). It is indeed less costly to increase the activity of proteasome within small compartments such as cellularized oocytes when compared to a large syncytium. Supporting this notion, we show that feminized mutants or wild-type animals in adult reproductive diapause (Angelo & Van Gilst, 2009) fail to produce oxidation-free oocytes. In these cases, the oocyte production flux is either stopped (for adult reproductive diapause) or greatly reduced (for fem-1(hc17) mutants (Kadandale & Singson, 2004)), and normal cytoplasmic turnover is likely altered. To firmly test this hypothesis, it is crucial to develop novel techniques for the sensitive detection of proteolytic activity in vivo in the worm germline. Interestingly, our data also indicate that fertilization, rather than ovulation, is coupled to the elimination of toxic products from the oocytes. Indeed, both feminized mutants, which do not fertilize oocytes because they lack sperm, and animals that have entered into an adult reproductive diapause, which produce oocytes and contain sperm but do not fertilize (Angelo & Van Gilst, 2009), produce oxidized oocytes (Fig. 6A, B and Table S1). One possible explanation for our data would be that fertilization produces signals that augment proteolytic activity in maturing oocytes. Testing this hypothesis will be possible with the development of sensitive assay for proteolytic activity in vivo in C. elegans.

Figure 8.

 Model proposed for resetting oxidatively damaged proteins during reproduction. In our model, we propose that an increased proteolytic activity (symbolized with white arrows) in matured oocytes compared to that in the syncytium may explain the observation that carbonylated proteins are less abundant in maturing oocytes and that diminishing the proteasome activity abolishes the resetting of the signal. Moreover, this model is in line with what was observed in mice where resetting carbonylated proteins and increased proteasome activity were concomitant (Hernebring et al., 2006).

Finally, using pbs-1 RNAi, the consequences of inheriting oxidatively damaged proteins were studied. We show that animals that suffer a partial deactivation of the proteasome in the germline lay few eggs (5% of the control, Fig. 7A) that hatch poorly (only 15% of these eggs hatch; Fig. 7B), which could be because of the inheritance of abnormally high levels of oxidatively damaged proteins. Interestingly, animals that hatched harbor low levels of carbonylated proteins soon after hatching (early L1 stage), suggesting that most toxic molecules are eliminated during early development (prior to hatching). Not surprisingly, these animals have a normal lifespan (Fig. 7D).

Future perspectives: developing new methods for detecting ROS, proteasome activity and oxidative damage in vivo in C. elegans

Damage elimination occurs in yeast (Aguilaniu et al., 2003; Erjavec et al., 2008) and in mice (Hernebring et al., 2006), although the mechanisms differ in each species. Our work shows that this process also occurs in C. elegans, thus providing a new model organism for the study of this process, as well as arguing in favor of its universality. The mechanisms mediating the process in the worm resemble those described in mice, suggesting that in animals, the mechanisms ensuring the production of damage-free progeny are conserved. The development of a novel technique for the visualization of this process in C. elegans presents many advantages. First, C. elegans is easily manipulated genetically and the combination of common genetic techniques with our staining protocol will likely enable the elucidation of the molecular basis of damage elimination. Second, the development of new techniques aiming at detecting (i) stressors (i.e. ROS); (ii) damage (i.e. oxidized macromolecules); and (iii) clearance machinery (i.e. proteasome activity) in vivo is in progress and will undoubtedly contribute to our understanding of the process of damage elimination. Finally, the genetics of aging is better characterized in the worm. It is therefore likely that the involvement of known longevity pathways in rejuvenation mechanisms will be investigated in the near future. An important perspective is to develop a technique to detect oxidized proteins in vivo. In yeast, carbonylated proteins were shown to colocalize with the heat shock protein HSP104 (Erjavec et al., 2007). Using a GFP-tagged version of HSP104, it is essentially possible to model how most oxidized proteins are handled in vivo. This has not yet been achieved in the worm and clearly represents an important step toward the optimization of the technique that will best suit the power of worm genetics.

How is the propensity to age-related disease reset?

Finally, our approach therefore provides an alternative research path to understand the molecular basis of youthfulness by asking the following question ‘what needs to be removed from germ cells to produce youthful, rejuvenated progeny?’ Indeed, offspring must be rejuvenated to ensure the perpetuation of the species. The rejuvenation process is likely to be complex and must guarantee the inheritance of some factors and the non-inheritance of others. In this work, we show the consequences of the inheritance of oxidatively damaged proteins, but there are probably a number of other critical factors that must also be cleared away from the progeny. In other words, our approach aims at defining ‘youth’ in molecular terms. It will be of particular interest to find out how the propensity to develop age-related diseases is also reset during reproduction. Indeed, the probability of developing an age-related disease (neurodegenerative disorders or cancer or metabolic syndromes) is quasi null in newborns.

Experimental procedures

Worm maintenance and strains

N2 Bristol was used as the wild-type strain. Nematodes were grown and maintained under standard conditions (Brenner, 1974). C. elegans strains are listed in the following table:

GenotypeOriginStrain
  1. *CGC = Caenorhabditis Genetics Center.

rrf-1(pk1417)ICGC*NL2098
sir-2.1(ok434)IVCGCVC199
sqt-1(sc103)II; bIs1 X [vit-2::GFP + rol-6(su1006)]CGCDH1033
fem-1(hc17)IVCGCBA17

Lifespan assay

Lifespan assays were conducted according to standard protocols (Kenyon et al., 1993). All assays were performed at 20°C, starting from the first day of adulthood. Animals were transferred away from progeny to new plates every day or every other day until animals stopped laying eggs. Worms crawling off the plate, exploding, bagging or contaminated were censored. We used the Biopylife software (Insa de Lyon, Lyon, France) to perform statistical analysis and to determine mean lifespan. P values were determined using log rank (Mantel–Cox) statistics.

Proteasome activity

Adult worms, treated with either PBS-1 RNAi or the empty vector, were homogenized with a Precellys 24 (Bertin) homogenizer using a cytosolic extraction buffer containing: Tris–HCl (50 mm, pH 7.5), sucrose (250 mm), MgCl2 (5 mm), ATP (2 mm), DTT (1 mm), EDTA (0.5 mm) and digitonin (0.025%) (Bulteau et al., 2001). Homogenates were centrifuged at 20 000 g for 15 min at 4°C. The supernatants (cytosolic fraction) were collected and the pellets were discarded. Protein quantification was made using the Bradford method (Biorad kit #500-0006, Hercules, CA, USA). Peptidase activities of the proteasome were evaluated using appropriate fluorogenic substrates. Chymotrypsin-like activities of the proteasome were assayed using the fluorogenic peptides LLVY-MCA (25 μm). The assay buffer was composed of Tris–HCl (50 mm, pH 7.5), KCl (40 mm), MgCl2 (5 mm), DTT (1 mm) and the appropriated peptide substrate. Enzymatic kinetics was carried out for 30 min at 20°C using 40 μg of cytosolic protein fractions in a temperature-controlled microplate fluorimetric reader (Mithras LB 940; Bertold, Stuttgart, Germany). The excitation/emission wavelengths were 380/460 nm. The rate of proteolysis was determined as the mean slope by comparing the linear response of fluorescence with time. Reactions were performed in the presence (20 μm) and absence of the specific proteasome inhibitor N-Cbz-Leu-Leu-leucinal (MG132), to test the specificity of the measured activity.

Western blotting

Protein extracts were prepared from 2 mL pellet of synchronized day 2 or day 5 adult nematodes grown on OP50 plates. Importantly, prior to the protein extraction, worms were separated everyday from their progeny, to avoid contamination with younger worms. To do so, worms were washed at least four times with M9 buffer and allowed to settle, starting from their first day of adulthood. Protein concentration was determined using the Bradford method (Biorad kit #500-0006, Hercules, CA, USA) in all protein samples and equal amount of proteins were loaded on a gel. All gels were duplicated to perform a blue coomassie staining to ensure that loading was indeed equal (not shown). All protein samples were resolved by electrophoresis through 10% gradient SDS polyacrylamide gel. Proteins were detected by immunoblot using the OxyBlot Protein Oxidation Detection Kit (Chemicon, Millipore, Billerica, MA, USA) following the provided protocol (Aguilaniu et al., 2003). Detections were accomplished using the ECL Plus Western Blotting Detection System (Amersham, Buckinghamshire, UK) following the provided instructions. Band intensities were quantified with ImageJ and normalized to the intensity of the signal obtained with ACT-1 antibody. Oxyblot analyses were repeated four times with lysates from separate nematode preparations.

Protocol for in situ detection of carbonylated protein in adult C. elegans

A detailed protocol is provided in Data S1 (Supporting information). All animals that were stained in this work were age matched at day 2 of adulthood with the exception of day 5 animals showed in Fig. S2 (Supporting information). The basis of our staining protocol is a mix between a classical protocol for worm freeze-cracking immunofluorescence (Duerr, 2006) and previously described in situ protocols for protein carbonylation (Pompella et al., 1996; Aguilaniu et al., 2003; Hernebring et al., 2006). We will describe what we think are the most important aspects of the newly developed protocol. First and prior to fixation, worms were treated with 2,4-DNPH to derivatize protein carbonyl groups (Levine et al., 1994). This reaction only occurs at very low pH and is then neutralized by adding alkaline solution (neutralization solution). Worms were then froze-cracked and fixed with cold methanol and acetone. Once worms are properly prepared, they are incubated with a primary antibody that recognizes the dinitrophenyl-hydrazone-carbonyl moieties and subsequently to a secondary antibody labeled with the fluorescent probe, Alexa Fluor, which is excited at 555 nm and emits at 565 nm. Indeed, endogenous fluorescence of worms is low at this wavelength. All staining presented in this works has been repeated at least three times independently and representative pictures are shown. It is important to note that our protocol does not stain the fertilized eggs because the freeze-cracking protocol for the entire worm is not compatible with that of eggs that are protected by an eggshell. It is therefore impossible to simultaneously visualize carbonylated proteins in situ of both adults and eggs. Consequently, eggs appeared black on our pictures (see Figs 3D, 5, 6C, S2 and S3) even when treated with H2O2.

Acknowledgments

We thank members of the Aguilaniu laboratory and Brian B Rudkin for proofreading our manuscript. We are grateful to the Caenorhabditis Genetics Center for kindly providing strains. We thank the PLATIM (Plateau Technique Imagerie/Microscopie IFR 128). This work was supported by grants from the ANR (ANR-06-JC-JC-0062) to HA, the Fondation pour la Recherche Medicale (FRM) to HA, CNRS PIR longévité program to HA, the Region Rhône Alpes to HA, the Ecole Normale Supérieure de Lyon and the CNRS to HA, doctoral grant from the ministère de l’enseignement supérieur et de la recherche to JG.

Ancillary