A conserved role for the 20S proteasome and Nrf2 transcription factor in oxidative stress adaptation in mammals, Caenorhabditis elegans and Drosophila melanogaster

Cells, small organisms and animals frequently experience protein damage as a result of oxidative stress. The levels of oxidative stress to which all are exposed is not static, but varies based on both external factors, such as air pollution (Halliwell et al., 1992; Menzel, 1994), pesticides (Abdollahi et al., 2004), ozone (Cross et al., 1992a; Cross et al., 1992b), exposure to radiation including UV (Hu and Tappel, 1992) and ionizing radiation (Leach et al., 2001), and internal factors, such as mitochondrial activity (Kappus, 1987) and immune responses (Baeuerle et al., 1996). Although oxidative stress exposure is variable, the ability of cells and organisms to cope with such stress can also be transiently varied to meet changing needs; this process is called oxidative stress adaptation (Davies, 2000; Pickering et al., 2010; Pickering et al., 2012; Wiese et al., 1995), or conditioning hormesis (Gems and Partridge, 2008; Ristow and Zarse, 2010). In previous studies using cultured mammalian cells, oxidative stress adaptation has typically been assayed using a pre-treatment and challenge model; in this model, exposing cells to a mild oxidative stress produces a transient increase in tolerance to a subsequent higher (and normally toxic) stress (Davies, 2000; Pickering et al., 2010; Pickering et al., 2012; Wiese et al., 1995). Such oxidative stress adaptation lasts for no more than 48 h (if the stimulating oxidant is removed or metabolized) and involves direct enzyme activation, upregulated expression of more than 50 protective and transient growth-arrest genes, and downregulation of a similar number of housekeeping and proliferative genes.

In mammalian cell culture (murine embryonic fibroblasts), an important part of this adaptive response is an increase in the 20S proteasome. Blocking the induction of 20S proteasome expression [with small interfering RNA (siRNA) or antisense RNA] significantly limits the effectiveness of the adaptive response (Pickering et al., 2010). The induction of the 20S proteosome is regulated by the stress-responsive transcription factor Nrf2 [nuclear factor (erythroid-derived 2)-like 2] (Pickering et al., 2012). In mammalian cells, adaptive increases in the levels and activity of the 20S proteasome, as well as the Pa28αβ (or 11S) proteasome regulator and the immunoproteasome, play a vital role in the proteolytic removal of oxidatively damaged proteins. This enables cells to cope with oxidative stress, and permits normal cell function to continue (Crawford et al., 1996; Davies, 1993; Davies, 2000; Pickering et al., 2010; Wiese et al., 1995).

Whole-organismal adaptation to oxidative stress has been documented in the nematode worm Caenorhabditis elegans. Exposure of the worm to a mild heat shock, or a mild dose of an oxidant (such as brief exposure to 100% oxygen or the quinone juglone), results in increased tolerance to a future, more severe, heat or oxidant assault. Interestingly, such adaptation can also result in an increased lifespan (Cypser and Johnson, 2002; Lithgow et al., 1995; Przybysz et al., 2009). Similarly, in the common fruit fly, Drosophila melanogaster, a mild oxidant pre-treatment, or irradiation, increases the tolerance of the fly to a future, normally toxic oxidant challenge (Moskalev et al., 2009). In the present study, experiments were conducted to test whether transient adaptation to hydrogen peroxide (H₂O₂) in C. elegans and D. melanogaster follows the same precepts that operate in mammalian cells in culture, and whether adaptation in worms and flies also depends on induction of the 20S proteasome, and if this induction is regulated by (orthologs of) the Nrf2 transcription factor. Although Pa28αβ and immunoproteasome are important for adaptation in mammalian cells (Pickering et al., 2010), neither is present in C. elegans or D. melanogaster, and while both organisms have a ortholog of the Pa28γ regulator, its function remains largely unclear (Rechsteiner and Hill, 2005). For these reasons, attention was focused on the 20S proteasome and Nrf2 in worms and flies.

Strains were maintained at 20°C as described by Brenner (Brenner, 1974). A bacterial lawn of OP50 Escherichia coli was spread on the plates for feeding. The N2 Bristol strain was used as the control for all experiments. skn-1(zu67) was obtained from the Caenorhabditis Genetics Center, University of Minnesota. For experiments investigating SKN-1::GFP localization, the strain LD1 containing the integrant idIs7, generously provided by the Blackwell lab (Harvard Medical School, Boston, MA, USA), was crossed into glo-1(zu391) to minimize auto-fluorescence. M9 solution (35 mmol l^–1 Na₂HPO₄, 22 mmol l^–1 KH₂PO₄, 85 mmol l^–1 NaCl, 19 mmol l^–1 NH₄Cl) was used as media for washing worms and for treating worms with H₂O₂.

In all experiments age-matched worm cultures were used. It was found that egg preparation using an NaOH/bleach solution, as is commonly used for age synchronization (Stiernagle, 2006), itself produced a stress response. Therefore, eggs layed over a 24 h period were used for synchronization.

Two to three day old (juvenile/young adult) worms were collected and washed with M9 medium; worms were washed 3 times to remove any contamination from the bacterial lawn. Worms were then transferred to Eppendorf tubes containing 1 ml of M9 medium, plus 0, 0.1, 1 or 10 μmol l^–1 H₂O₂. Worms were incubated in this medium with mild shaking for 1 h then returned to normal plates for 24 h to permit adaptation to occur. Unless otherwise stated, 1 μmol l^–1 H₂O₂ was used as an adaptive pre-treatment for C. elegans.

Three to four day old adult worms were collected and washed with M9 medium; worms were washed 3 times to remove any contamination from the bacterial lawn. Worms were then transferred to Eppendorf tubes containing 1 ml of M9 medium plus 100 μmol l^–1 to 160 mmol l^–1 H₂O₂. Worms were incubated in this medium, with mild shaking, for 1 h then returned to normal plates. Survival was either scored instantly or 24 h later, by response to prodding with a pick. This experiment was typically performed blind, by another member of the group encoding and randomizing the plates immediately before they were scored. Where pre-treatment challenge assays were performed, pre-treatment was typically carried out 24 h prior to challenge and 80 mmol l^–1 H₂O₂ was used for the challenge dose unless otherwise stated.

Age-matched idIs7;glo-1(zu391) worms were prepared as described above. L4 animals were used for imaging experiments as they have lower levels of intestinal auto-fluorescence. Animals were collected 48 h after egg laying and washed 3 times with M9 medium to remove any bacterial contamination. Worms were then transferred to Eppendorf tubes containing 1 ml of M9 medium ±1 μmol l^–1 H₂O₂ or ±10 mmol l^–1 H₂O₂. Worms were incubated ±H₂O₂ for 20 min and then paralyzed by the addition of 2,3-butanedione monoxime. Worms were then mounted on slides containing an agarose pad. SKN-1::GFP nuclear localization was scored by microscopic analysis using a 500–1000× magnification with GFP illumination (An and Blackwell, 2003; An et al., 2005).

Drosophila melanogaster were cultured on a standard agar/dextrose/corn meal/yeast media (Ren et al., 2009) at 25°C. Unless otherwise stated, w[1118] flies were used in all assays. Flies were collected over a 48 h period from pre-cleared bottles, and allowed 4 days to mature so that at initiation of assays, flies were 4–6 days old.

Samples of 20 flies were transferred to vials containing half a Kimwipe soaked in 1 ml of 5% sucrose and 4.4 mol l^–1 H₂O₂. Survival was then scored ever 6 or 8 h following initiation of the challenge. Flies were scored as dead once they became completely immobile.

Samples of 20 flies were transferred to vials containing half a Kim-wipe soaked in 1 ml of 5% sucrose plus 0, 10 or 100 μmol l^–1 or 1 mmol l^–1 H₂O₂ for 8 h. Flies were then returned to normal vials for 16 h to permit adaptation to occur. Flies were then challenged with a toxic dose of H₂O₂. In the case of experiments using cnc-C^RNAi flies, we found that, for effective adaptation to occur, flies needed to be incubated on H₂O₂-soaked Kimwipes for 24 h instead of 8 h. In addition, because of the longer incubation time in the experiments with cnc-C^RNAi flies, 10 μl of either ethanol or RU-486 was added to the Kimwipes during this treatment.

Approximately 1000–2000, 3–4 day old adult worms were collected per sample and washed with M9 medium. Worms were washed 3 times with M9 medium to remove any bacterial contamination. Worms were then transferred into a solution containing RIPA buffer (catalog no. 89901, Thermo Fisher, Waltham, MA, USA), supplemented with protease inhibitor cocktail (catalog no. 11836170001, Roche, Nutley, NJ, USA). We then lysed the worms by freezing them once, and subjecting them to 3×10 s bursts of sonication, followed by 3 cycles of incubation on ice for 5 min, and vortexing. Lysates were centrifuged at 10,000 g to remove un-lysed fragments. Protein content was quantified with a BCA protein assay kit (Pierce, Rockford, IL, USA) according to the manufacturer’s instructions for western analysis.

Twenty flies were collected per sample. Flies were then transferred into a solution containing RIPA buffer supplemented with protease inhibitor cocktail, as above. Flies were frozen, then homogenized using an electronic pestle after which lysis was performed by 3×5 min incubations on ice, followed by vortexing. Samples were then centrifuged at 10,000 g to remove cuticle fragments and unlysed cells. Protein content was next quantified with the BCA Protein Assay Kit as for C. elegans.

A 40 μg sample of protein was run on an SDS-PAGE gel and transferred to a PVDF membrane using standard western blot techniques.

Two-hundred adult worms were collected per sample and washed 3 times with M9 medium to remove bacterial contamination. Worms were then transferred into a solution containing proteolysis buffer (50 mmol l^–1 Tris, 25 mmol l^–1 KCl, 10 mmol l^–1 NaCl, 1 mmol l^–1 MgCl₂, 1 mmol l^–1 DTT, pH 7.5). Worms were frozen once, then homogenized, after which cell lysis was achieved by three freeze–thaw cycles in dry ice for 5 min, followed by a room temperature water bath for 5 min. After this, samples were centrifuged at 10,000 g to remove cuticle fragments and unlysed cells. Protein content was quantified with the BCA Protein Assay Kit (Pierce) according to the manufacturer’s instructions. Samples were then diluted to the appropriate concentration based on BCA assays results.

Ten flies were collected per sample. Flies were then transferred into a solution containing proteolysis buffer (50 mmol l^–1 Tris, 25 mmol l^–1 KCl, 10 mmol l^–1 NaCl, 1 mmol l^–1 MgCl₂, 1 mmol l^–1 DTT, pH 7.5). Flies were frozen once, then homogenized using a pestle, after which lysis was performed by three freeze–thaw cycles performed through incubation on dry ice for 5 min, followed by a room temperature water bath for 5 min. After this, samples were centrifuged at 10,000 g to remove cuticle fragments and unlysed cells. Protein content was quantified with the BCA Protein Assay Kit (Pierce) according to the manufacturer’s instructions. Samples were then diluted to the appropriate concentration based on BCA assays results.

Aliquots of 5.0 μg to 1 mg (depending on the assay) were transferred, in triplicate, to 96-well plates, and 2 μmol l^–1 of N-succinyl-Leu-Leu-Val-Tyr-AMC (catalog no. 80053-860, VWR, Chester, PA, USA) was added to each well. Plates were incubated at 37°C and mixed at 300 r.p.m. for 4 h. Fluorescence readings were taken at 10 min intervals using an excitation wavelength of 355 nm and an emission wavelength of 444 nm. Fluorescence units were converted to moles of free 7-amino-4-methylcoumarin (AMC), with reference to an AMC standard curve of known amounts of AMC (catalog no. 164545, Merck, Whitehouse Station, NJ, USA), following subtraction of background fluorescence. In some experiments, cells were treated with 20 μmol l^–1 of the proteasome inhibitor lactacystin (catalog no. 80052-806, VWR), 30 min prior to incubation and addition of substrates. Lactacystin was dissolved in DMSO at a 100× concentration and combined with samples at a concentration of 0.1%.

Two to three day old (juvenile/young adult) worms were cultured on plates containing a bacterial lawn expressing RNAi against the proteasome subunits pas-5, pbs-3, pbs-5 and pbs-6. In addition, some plates contained bacteria expressing RNAi against skn-1 or an empty vector control (L4444). Worms were incubated on these plates for 6 h, and then collected and washed 3 times with M9 medium to remove any contamination from the bacterial lawn. Worms were then transferred to Eppendorf tubes containing 1 ml of M9 medium ±1 μmol l^–1 H₂O₂. Worms were incubated, with mild shaking, for 1 h then returned to RNAi plates for 24 h. Worms were collected and washed 3 times with M9 medium to remove any bacterial contamination. Worms were then transferred to 48-well plates containing 500 μl of M9 medium + 8 mmol l^–1 H₂O₂. Worms were incubated, with mild shaking, for 1 h, then survival was scored immediately (as above).

Flies expressing RNAi against two proteasome subunits were purchased from the Vienna Drosophila RNAi center (VDRC, Vienna, Austria). These were prosβ1^RNAi (w[1118]; P[GD13913]v35923), and prosβ2^RNAi (w[1118]; P[GD10938]v24749). In addition we received the kind donation of flies expressing RNAi against D. melanogaster cap ’n’ collar transcription factor (an ortholog of mammalian Nrf2; cnc-C) and dkeap-1 from Dr Dirk Bohman (Sykiotis and Bohmann, 2008; Sykiotis and Bohmann, 2010). Males from these lines (or w[1118] as a control) were crossed with virgin females containing the Act-GS-255B driver (Fischer et al., 1988; Ford et al., 2007). Parents were removed 4 days after initiation of the cross. Progeny were then collected over a 48 h period after eclusion. The exception to this was the cnc-C^RNAi line where a low birthrate required collections over a 72 h period. The Act-GS-255B driver is activated by feeding flies RU-486. Flies were cultured in normal vials containing either 50 μl of stock RU-486 (20 mg ml^–1) or ethanol, which had been added to vials and air dried 24 h prior to the assay. Flies were incubated in these vials for 5–7 days. In experiments involving H₂O₂ adaptation, flies were removed from ±RU-486 vials 24 h before the end of RU-486 treatment. Flies were then transferred to vials containing half a Kimwipe soaked in 1 ml of 5% sucrose ±100 μmol l^–1 H₂O₂ for 8 h. The flies were then returned to RU-486 or ethanol vials for the remaining 16 h of RU-486 treatment.

To determine whether C. elegans are capable of H₂O₂ oxidative stress adaptation, a pre-treatment/challenge assay, similar to that developed in mammalian culture, was employed (Pickering et al., 2010; Pickering et al., 2012; Wiese et al., 1995). In this assay, age-matched young adult worms were exposed to low adaptive doses of H₂O₂ for 1 h, and 1 h later were assayed for survival in response to toxic doses of H₂O₂ (40–160 mmol l^–1) using standard survival assays (see Materials and methods) (Larsen, 1993). Worms that were not pre-treated exhibited a progressive decline in survival with increasing H₂O₂ challenge (Fig. 1A). A challenge dose of 80 mmol l^–1 produced a 70% loss (only 30% survival) of non-pre-treated worms. Survival was increased to 45–50% with H₂O₂ pre-treatment (Fig. 1B). From the above results it appeared that mild H₂O₂ pre-treatment produces the expected enhanced resistance to oxidative stress in C. elegans.

The challenge doses used in the above experiment, however, were highly toxic and produced 0% survival 24 h after treatment, even in pre-treated samples (Fig. 1A; data not shown). To better test oxidative stress adaptation, it seemed reasonable to examine longer-term survival with less severe H₂O₂ challenges. To test this, the assays in Fig. 1A were repeated, using milder H₂O₂ challenge doses (200 μmol l^–1 to 3.2 mmol l^–1). A 24 h recovery period was then permitted after H₂O₂ challenge before scoring survival. Using this protocol, a similar pattern of declining survival with increasing H₂O₂ challenge was observed (Fig. 1C). It was found that a 1.6 mmol l^–1 H₂O₂ pre-treatment would produce a 16% survival increase upon subsequent challenge, and survival upon challenge was increased to over 30% with a 10 μmol l^–1 H₂O₂ pre-treatment (Fig. 1D).

It is difficult to directly measure the activity of individual proteases in an extract, so instead their activity is approximated though the capacity of the extract to degrade peptide substrates (‘proteolytic capacity’). It has previously been shown that an increase in proteolytic capacity is an important component of the adaptive response to oxidative stress in mammalian cell culture (Grune et al., 2011; Pickering et al., 2010; Pickering et al., 2012). To determine whether a similar increase in proteolytic capacity occurs in C. elegans, the proteolytic activities of whole-worm extracts were examined 24 h after pre-treating worms with H₂O₂ (as per Fig. 1B,D), by measuring degradation of the short peptide substrate Suc-LLVY-AMC. H₂O₂ pre-treatment produced a 5-fold adaptive increase in proteolytic capacity in extracts from animals that were pre-treated with mild H₂O₂ compared with those extracts that were not pre-treated (Fig. 1E). Interestingly, pre-treatment with either 1 or 10 μmol l^–1 H₂O₂ increased both survival and proteolytic activity, whereas 0.1 μmol l^–1 pre-treatment did not significantly increase survival or proteolytic activity. This appears to be a non-linear response in which a mild H₂O₂ (0.1 μmol l^–1) pre-treatment yields a weak increase in proteolytic activity but a much stronger increase in oxidative stress tolerance.

These results suggest that as an overall phenomenon, oxidative stress adaptation and the corresponding increase in proteolytic activity are similar in C. elegans to that observed in mammalian cells (Grune et al., 2011; Pickering et al., 2010; Wiese et al., 1995). In the mammalian system, upregulation of the 20S proteasome has been shown to play an important role in this response through an increase in the capacity of cells to degrade oxidized proteins (Pickering et al., 2010; Pickering et al., 2012). To see whether this was also the case in C. elegans, the level of 20S proteasome protein under H₂O₂ adaptation was measured. To do this, animals were pre-treated with a variety of mild H₂O₂ doses. Subsequently, worm lysates from these pre-treated animals were analyzed for the abundance of either the 20S proteasome subunit PAS-7 or the 19S regulator subunit RPN-10 (proteasome regulatory particle non-ATPase-like 10, a key component of the 26S proteasome) using western blotting (Fig. 2A,B). This H₂O₂ pre-treatment produced a 2- to 3-fold increase in 20S proteasome (PAS-7) levels with no significant change in the levels of the 19S regulator of the 26S proteasome (RPN-10). These results suggest that the 20S but not the 26S proteasome may be involved in oxidative stress adaptation.

Because the 20S proteasome was induced by H₂O₂ treatment it was next determined whether 20S proteasome induction is important for the adaptive response. To test this, conditioning hormesis was assayed in animals in which the 20S proteasome was first knocked-down by RNAi. The effects of knock-down of four 20S subunit genes were tested. These corresponding RNAi clones were available from an RNAi library (Kamath et al., 2003): pas-5, pbs-3, pbs-5 or pbs-6. In this experiment, worms were cultured with RNAi directed against the 20S proteasome subunits pas-5, pbs-3, pbs-5 or pbs-6, or the empty vector (L4440) for 6 h to block new synthesis of proteasome genes without significantly reducing baselines levels. Worms were then transferred into 1 μmol l^–1 H₂O₂ for 1 h to induce adaptation. Worms were subsequently returned to their RNAi plates for a further 24 h, and then collected for adaption experiments, performed as in Fig. 1B. RNAi knockdown of the four proteasomal subunits did not significantly alter survival on 80 mmol l^–1 H₂O₂ compared with control RNAi treatments; however, RNAi knockdown of any one of these proteasome subunits severely blunted the enhanced survival of animals that were pre-treated with adaptive doses of H₂O₂ (Fig. 2C). RNAi knockdown of pbs-5 appeared to be the most effective, whereas pas-5, pbs-3 and pbs-6 appeared to have partial effects. This may be due to differences in the efficacy of RNAi in these experiments or it may reflect differences in the functions of these subunits in the proteasome. These results indicate that the 20S proteasome plays an important role in the adaptive increase in oxidative stress tolerance exhibited by H₂O₂ pre-treated C. elegans.

Nrf2 has previously been found to be an important regulator of oxidative stress adaptation in mammalian cell culture (Pickering et al., 2012). At least part of the reason for this is the Nrf2-dependent increase in 20S proteasome synthesis during mammalian cell oxidative stress adaptation (Pickering et al., 2012). Studies by the Blackwell lab have demonstrated that SKN-1 is a functional ortholog of Nrf2 (An and Blackwell, 2003). Like Nrf2, SKN-1 is normally maintained at low levels in the cytoplasm, but under conditions of oxidative stress it translocates to the nucleus (An and Blackwell, 2003; An et al., 2005; Kahn et al., 2008; Oliveira et al., 2009). It has been shown that high doses of H₂O₂ (10 mmol l^–1) can cause the cytoplasmic to nuclear translocation of SKN-1 fused to GFP (SKN-1::GFP) in C. elegans intestinal cells (An and Blackwell, 2003; An et al., 2005). Also, a report by Przybysz and colleagues has shown that with depletion of SKN-1, worms become more susceptible to the oxidant juglone (Przybysz et al., 2009). The SKN-1::GFP line (An and Blackwell, 2003), was crossed with a glo-1 mutant strain to produce worms with reduced non-specific fluorescence. Using transgenic animals expressing the SKN-1::GFP reporter, the number of fluorescent SKN-1::GFP intestinal nuclei in H₂O₂-treated animals was compared with that in untreated controls. First, it was confirmed that exposure to 10 mmol l^–1 H₂O₂ caused a 3-fold increase in the number of SKN-1::GFP fluorescent nuclei (Fig. 3A,B). Treated worms had an average of 10.0±1.4 nuclei exhibiting expression of GFP, compared with 3.7±1.2 in untreated worms (Fig. 3B). It was next tested whether lower adaptive treatment concentrations could also produce an increase in SKN-1::GFP nuclear localization. The experiments of Fig. 3A,B were repeated, using an adaptive dose of 1 μmol l^–1 H₂O₂, and a smaller but still significant increase in the number of SKN-1::GFP fluorescent nuclei was observed relative to untreated controls (Fig. 3C,D). Treated worms had an average of 6.1±1.1 nuclei exhibiting expression of GFP, compared with 2.1±0.6 in untreated worms (Fig. 3B). These results suggest that low adaptive doses of H₂O₂ that confer enhanced survival may activate SKN-1 by promoting its translocation to the nucleus.

To determine whether SKN-1 is important in H₂O₂-induced oxidative stress adaptation, experiments were performed to test whether enhanced survival following pre-treatment through H₂O₂ adaptation is dependent on the presence of SKN-1. Knockdown of SKN-1 by RNAi did not alter survival of non-pre-treated animals compared with control RNAi-treated animals upon challenge with 80 mmol l^–1 H₂O₂. However, the adaptive response with 1 μmol l^–1 H₂O₂ pre-treatment was blunted following treatment with RNAi against skn-1 (Fig. 3E). Thus, not only does the adaptive response in C. elegans involve an increase in SKN-1 nuclear localization but also SKN-1 activity is important for oxidative stress adaptation, implicating SKN-1 nuclear translocation as a crucial event in oxidative stress adaptation. However, as it was unclear how effective the SKN-1 RNAi was at causing depletion of skn-1, the experiment was repeated using a skn-1(zu67) mutant (Bowerman et al., 1992). The skn-1(zu67) mutant encodes an early stop codon that is predicted to eliminate activity of the two SKN-1 isoforms (SKN-1a and SKN-1c) that are expressed in the intestine and are required for stress responses (An and Blackwell, 2003). Despite pre-treatment of this mutant line with a range of doses of H₂O₂, no adaptive increase in oxidative stress tolerance was observed upon H₂O₂ challenge (Fig. 3F). This result is therefore highly supportive of an important role for skn-1 in regulating the adaptive response.

Given the role of mammalian Nrf2 in regulating proteasome function (Pickering et al., 2012), it is possible that SKN-1 may also regulate proteasome activity in response to adaptive doses of H₂O₂. This idea is consistent with the finding that at least half of the proteasome subunit genes in the C. elegans genome contain SKN-1 binding sites and are inducible by SKN-1 (Oliveira et al., 2009). Furthermore, it has been demonstrated that RNAi depletion of SKN-1 causes a marked reduction in the proteolytic capacity of the worm (Kirkwood, 2005). To determine whether SKN-1 regulates the oxidative stress-induced adaptive increase in the 20S proteasome, western blotting was used to measure the levels of a 20S proteasome subunit, PAS-7, following 1 μmol l^–1 H₂O₂ pre-treatment, in wild-type (N2) worms and skn-1(zu67) mutant worms. H₂O₂ pre-treatment of control animals caused a 60% increase in the level of PAS-7 from whole-worm lysates (Fig. 3G). In contrast, only a 30% increase in PAS-7 levels was observed following H₂O₂ treatment in lysates from the skn-1 mutant animals (Fig. 3G), and PAS-7 is one of the subunits previously shown to be inducible by skn-1 (Oliveira et al., 2009). These results suggest that skn-1 regulates the oxidative stress-induced increase in the 20S proteasome in C. elegans, similar to the effect of Nrf2 in mammalian cell cultures.

Previous studies have shown that pre-treatment of D. melanogaster with mild doses of oxidants or radiation can result in a subsequent increase in tolerance to oxidative damage (Moskalev et al., 2009). First, we investigated whether H₂O₂ pre-treatment in D. melanogaster could be used to induce an increase in proteolytic capacity as has been seen in other model systems. Because of the greater size and complexity of the fly compared with mammalian cell cultures, and even nematode worms, it was not possible to simply incubate flies in H₂O₂. Instead, flies were maintained in vials without standard food, containing a Kimwipe soaked in 5% sucrose with or without an adaptive dose of H₂O₂ (Grover et al., 2009). For these experiments, flies were collected 0–48 h after eclosion. Flies were then allowed to mature for 4 days, after which they were transferred into vials containing Kimwipes soaked in 5% sucrose with or without 10–1000 μmol l^–1 H₂O₂ for 8 h. It was confirmed that the sucrose/H₂O₂ solution was ingested by the flies over the 8 h incubation period by uptake of a visible dye (10% blue food color, Amerifoods, Los Angeles, CA, USA) (supplementary material Fig. S1). The flies were returned to normal vials for 16 h to permit adaptation, and then homogenized. The proteolytic capacity of the fly lysates was then measured by degradation of the fluorogenic peptide model substrate Suc-LLVY-AMC. As shown in Fig. 4A, a progressive increase in proteolytic capacity was observed in females following H₂O₂ pre-treatment. In contrast, no increase in proteolytic capacity was observed when male flies were pre-treated with H₂O₂: instead, 100 and 1000 μmol l^–1 H₂O₂ pre-treatment was found to decrease the proteolytic capacity of lysates from male flies (Fig. 4A).

This difference in adaptive response between male and female flies might be due to different food consumption rates. Female Drosophila consume approximately 3 times as much food as their male counterparts, have a much larger gut, and have more stem cells in their digestive tract (Wong et al., 2009). It was confirmed that during the 8 h H₂O₂ pre-treatment, male flies consumed considerably less of the sucrose/H₂O₂ solution than did female flies, and this might mean that male flies have a lower exposure to the adaptive dose of H₂O₂ and so are less responsive to H₂O₂ pre-treatment (supplementary material Fig. S1). Although it is clear that male flies consumed less than half as much of the sucrose/H₂O₂ solution as did females (see supplementary material Fig. S1), we can think of no reason why this difference in ingestion would negatively affect proteolytic capacity. An alternative hypothesis is that the observed differences are due to sexual dimorphism in stress responses between male and female flies and, consistent with this idea, sexual dimorphism has previously been observed in thermotolerance and oxidative stress resistance in Drosophila (Sørensen et al., 2007; Waskar et al., 2009; Weber et al., 2012).

To determine whether the increased proteolytic capacity of female flies (Fig. 4A) might be due to increased proteasome activity, selected samples were pre-exposed to the proteasome-selective inhibitor lactacystin, 30 min before proteolytic capacity was measured (Pickering et al., 2010). No increase in proteolytic capacity was evident in lactacystin-treated lysates, indicating that increased proteasome activity underlies the elevated proteolytic capacity (Fig. 4B).

Because H₂O₂ treatment increased proteolytic capacity in female fly extracts, but not male fly extracts, it was next tested whether stress resistance and proteasome levels would correlate with tolerance to oxidative stress in an adaptation experiment. The first step was to develop an appropriately toxic challenge for the flies. Instead of subjecting flies to a single challenge as in mammalian cell culture and C. elegans, the assay employed for D. melanogaster involved incubating the flies in vials containing the toxic dose of H₂O₂ and scoring survival over time, as previously described (Grover et al., 2009; Sykiotis and Bohmann, 2008; Sykiotis and Bohmann, 2010). Survival was scored every 8 h, for up to 104 h. The survival curves for male and female flies cultured in vials containing Kimwipes with 0, 1, 2, 4 or 8 mol l^–1 H₂O₂ are shown in supplementary material Fig. S2.

Having determined an optimal H₂O₂ challenge concentration of 4.4 mol l^–1 (15%) (supplementary material Fig. S2), adaptation experiments were performed using flies prepared and pre-treated as in Fig. 4. Then, 24 h after initiating pre-treatment (16 h after terminating pre-treatment), flies were transferred to vials containing Kimwipes soaked in 5% sucrose and a toxic challenge dose of 4.4 mol l^–1 (15%) H₂O₂. Survival was scored every 8 h, for up to 80 h. Non-pre-treated female flies survived an average of 48 h following initiation of the challenge. By comparison, female flies pre-treated with 100 μmol l^–1 H₂O₂ (which was found to be the optimum pre-treatment) survived an average of 64 h. This represented a significant increase in survival time (Fig. 5A,B). In contrast, no change in survival time was seen following pre-treatment of male flies; both pre-treated and non-pre-treated male flies survived an average of 40 h following initiation of the challenge (Fig. 5A,B), thereby supporting a functional correlation between proteolytic activity and adaptation.

Next, we tested whether 20S proteasome expression is induced by H₂O₂ treatment in female flies. To do this, female flies were pre-treated for 8 h with 100 μmol l^–1 H₂O₂, which appeared to be the optimum treatment regime. Then, 24 h after the start of pre-treatment, female flies were homogenized and the lysates were analyzed by western blotting. The western blots were incubated with antibodies directed against the α-subunit of the 20S proteasome, the CG13349 subunit of the 19S regulator of the 26S proteasome, and β-tubulin as a loading control. As presented in Fig. 5C, a 50–60% increase in the expression of the 20S proteasome α-subunit was observed (reflecting increased overall 20S proteasome levels) but there was no change in the expression of the 19S regulator (indicating no increase in the 26S proteasome). Having seen an increase in the female fly 20S proteasome as a product of oxidative stress adaptation, we next asked whether this induction was required for adaptation to occur. To test this, two RNAi lines directed against the prosβ1 and prosβ2 20S proteasome subunits were used. These lines were crossed to the Act-GS-255B strain (Ford et al., 2007) enabling the RNAi to be conditionally expressed in the presence of the drug RU-486 in all the somatic tissues of the adult fly (Shen et al., 2009). Additionally, the Act-GS-255B strain was crossed with w[1118] flies as controls for potential effects of RU-486. The female progeny of these three crosses were then cultured with or without RU-486 for 7 days. RU-486 activates the Gene-Switch transcription factor, thereby inducing expression of the appropriate RNAi. One day prior to the end of the 7 day incubation, the female flies were transferred to vials containing Kimwipes soaked in 5% sucrose with or without 100 μmol l^–1 H₂O₂ for 8 h, and then returned to vials containing RU-486 or ethanol for the remaining 16 h. At the start of the assay, the flies were transferred to vials containing 5% sucrose and a toxic dose of 4.4 mol l^–1 H₂O₂. Fly survival was then scored every 8 h for the next 96 h. As a minor confounding issue, it was observed that RU-486 induced a small increase in tolerance to H₂O₂ challenge. This was controlled for by including controls with or without RU-486 in all experiments. In w[1118] × Act-GS-255B flies, an H₂O₂ pre-treatment-induced increase in tolerance to oxidative stress was observed, both with and without RU-486 treatment (Fig. 6A,B). The experiment was then repeated using the progeny from the two RNAi lines crossed with Act-GS-255B. In the prosβ2 RNAi flies, it was found that flies not treated with RU-486 adapted to a similar extent to that seen for control female flies following H₂O₂ pre-treatment. In contrast, H₂O₂ pre-treatment in female flies that were also treated with RU-486 failed to increase tolerance to oxidative stress. In fact, H₂O₂ pre-treatment actually reduced oxidative stress tolerance in RU-486-treated females (Fig. 6C,D). Similarly, prosβ1 RNAi flies not treated with RU-486 adapted to H₂O₂ pre-treatment to a similar extent to control flies. When flies were treated with both RU-486 and H₂O₂, however, not only did they lose the ability to adapt but also their tolerance to oxidative stress was severely reduced (Fig. 6E,F).

In both mammalian cells (Pickering et al., 2012) and C. elegans (Fig. 3), it was observed that Nrf2/SKN-1 is important for oxidative stress adaptation, and that it regulates the H₂O₂-induced increase in the 20S proteasome. We wished to test whether this was also the case in D. melanogaster. The D. melanogaster ortholog of Nrf2/SKN-1 is CNC-C, which has also been shown to be an important regulator of oxidative stress tolerance (Sykiotis and Bohmann, 2008; Sykiotis and Bohmann, 2010). First, to test whether CNC-C is involved in adaptation, a line expressing RNAi against cnc-C (Sykiotis and Bohmann, 2008; Sykiotis and Bohmann, 2010) was crossed with Act-GS-255B, and adaption experiments were performed on the progeny as for the proteasome RNAi lines. As before, an increase in oxidative stress tolerance was observed with H₂O₂ treatment in flies not treated with RU-486 (Fig. 7A), while no adaptive increase in oxidative stress tolerance was observed in RU-486-treated flies (Fig. 7B). In fact, H₂O₂ treatment caused a reduction in oxidative stress tolerance (Fig. 7A–C) in RU-486-treated female flies.

DKEAP-1 is a repressor of CNC-C and has a similar function to that of Keap-1, which functions as a repressor of Nrf2 in mammalian cells (Sykiotis and Bohmann, 2008). As before, the dkeap-1 RNAi line was crossed with Act-GS-255B and the progeny were then treated with RU-486 for 5 days. The flies were subsequently challenged with a toxic dose of H₂O₂ and survival was measured as before. It was found that knock-down of dkeap-1 caused a significant increase in resistance to oxidative stress (Fig. 7D–G). We were next interested to determine whether the induction of the 20S proteasome is dependent on CNC-C/Nrf2. To test this, cnc-C^RNAi × ACT-GS-255B flies were prepared and pre-treated as in Fig. 7A. The flies were then lysed and the proteins analyzed by western blotting with antibodies directed against α-subunits of the 20S proteasome, or β-tubulin as a loading control. It was observed that H₂O₂ treatment produced an increase in the expression of the 20S proteasome subunit that was suppressed in flies treated with RU-486. These results are supportive of a role for cnc-C in regulating the H₂O₂-induced increase in the 20S proteasome (Fig. 7H,I).

In mammalian cells, it has been shown that H₂O₂-induced adaptation to oxidative stress occurs through a pathway that is strongly dependent on an Nrf2-mediated increase in the 20S proteasome (Pickering et al., 2010; Pickering et al., 2012). The model organisms C. elegans and D. melanogaster have also been shown to be capable of adaptation to oxidative stresses through mild exposure to heat shock, 100% oxygen or the redox quinone juglone in worms, and heat shock or irradiation in flies (Cypser and Johnson, 2002; Lithgow et al., 1995; Moskalev et al., 2009; Przybysz et al., 2009). The present results demonstrate that both model organisms are capable of such adaptation with H₂O₂ pre-treatment. As in mammalian cells (Pickering et al., 2010), it was found that this adaptive response involves an increase in proteolytic activity and increased expression of the 20S proteasome, but not of the 26S proteasome. Moreover, it was found that the increase in 20S proteasome expression in both worms and flies, as in mammalian cells, is important for the adaptive response, and that this increase is mediated by the SKN-1 and CNC-C orthologs of the mammalian Nrf2 transcription factor, respectively. These studies indicate conservation of the adaptive response to oxidative stress across a wide biological diversity, and demonstrate that stress mechanisms operative in cell culture also apply in disparate intact organisms.

Stress adaption, or conditioning hormesis, has been widely reported in C. elegans, but it is typically studied in terms of changes in lifespan, with only a few examples of changes in oxidative stress tolerance (Cypser and Johnson, 2002; Lithgow et al., 1995). It was shown in previous reports that exposure to high (10 mmol l^–1) concentrations of H₂O₂ induces SKN-1 and causes its translocation to the nucleus (An and Blackwell, 2003; An et al., 2005; Oliveira et al., 2009). It appears that much lower (1 μmol l^–1) adaptive concentrations of H₂O₂ also induce SKN-1 and cause its translocation to the nuclei in cells throughout the intestine of C. elegans. In addition, using skn-1 RNAi and skn-1 mutants, it was shown that this induction and translocation is important for the adaptive increases in proteasome expression and oxidative stress resistance.

Oxidant- or radiation-induced adaptation to oxidative stress in D. melanogaster has previously been reported (Moskalev et al., 2009). Here, it was confirmed that upon H₂O₂ pre-treatment flies become more resistant to oxidative stress and their cellular proteolytic capacity increases. This increase in proteolytic capacity was shown to be largely dependent on the 20S proteasome, which is consistent with mammalian cell culture studies (Pickering et al., 2010). Finally, these data indicate that CNC-C, the Drosophila ortholog of mammalian Nrf2 and C. elegans SKN-1, mediates adaptive increases in stress resistance, at least in part through induction of 20S proteasome expression.

An important aspect of this study is the use of short-term RNAi treatments as a means of blocking new protein synthesis without significantly altering the baseline expression level of the target protein. This was achieved in C. elegans by feeding worms RNAi for 24 h and in flies by feeding with RU-486 for 5–7 days in adulthood. One concern in the use of RNAi to deplete a protein from an animal is that this approach may cause additional unwanted physiological changes such as either inhibition of normal cell function or production of a stress response. Here, such unwanted physiological changes were reduced or avoided by employing short-term RNAi treatments.

One long-term goal for the present work was to develop viable animal models for aging, consisting of short-lived organisms that recapitulate major stress-adaptive mechanisms that operate in mammals. The present data demonstrate that C. elegans and D. melanogaster, both of which have long been among the most important organisms for aging models, share with mammalian cells a capacity for oxidative stress adaptation that relies on increased 20S proteasome synthesis, mediated by well-conserved (Nrf2/SKN-1/CNC-C) signal transduction pathways.

 
• AMC

  7-amino-4-methylcoumarin

 
• CNC-C

  D. melanogaster cap’n’collar transcription factor (an ortholog of mammalian Nrf2)

 
• H₂O₂

  hydrogen peroxide

 
• Nrf2

  nuclear factor (erythroid-derived 2)-like 2

 
• PA28αβ

  proteasome activator 28αβ

 
• PAS-5

  proteasome α-subunit 5

 
• PAS-7

  proteasome α-subunit 7

 
• PBS-3

  Proteasome β-subunit 3

 
• PBS-5

  Proteasome β-subunit 5

 
• PBS-6

  proteasome β-subunit 6

 
• RPN-10

  proteasome regulatory particle non-ATPase-like 10

 
• SKN-1

  C. elegans SKiNhead-1 transcription factor (an ortholog of mammalian Nrf2)

 
• Suc-LLVY-AMC

  succinylated peptide N-succinyl-leucine-leucine-valine-tyrosine-7-amino-4-methylcoumarin (used as a peptide substrate to measure proteolytic capacity)

We thank Dr Keith Blackwell, from Harvard Medical School, Boston, MA, USA, Dr Dirk Bohmann from the University of Rochester Medical Center, Rochester, NY, USA, and Dr Bruce Bowerman from the Fred Hutchinson Cancer Research Center, Seattle, WA, USA, for their kind donation of C. elegans stains and Drosophila lines used in this paper.