FOXO/DAF-16 Activation Slows Down Turnover of the Majority of Proteins in C. elegans

Summary

Most aging hypotheses assume the accumulation of damage, resulting in gradual physiological decline and, ultimately, death. Avoiding protein damage accumulation by enhanced turnover should slow down the aging process and extend the lifespan. However, lowering translational efficiency extends rather than shortens the lifespan in C. elegans. We studied turnover of individual proteins in the long-lived daf-2 mutant by combining SILeNCe (stable isotope labeling by nitrogen in Caenorhabditis elegans) and mass spectrometry. Intriguingly, the majority of proteins displayed prolonged half-lives in daf-2, whereas others remained unchanged, signifying that longevity is not supported by high protein turnover. This slowdown was most prominent for translation-related and mitochondrial proteins. In contrast, the high turnover of lysosomal hydrolases and very low turnover of cytoskeletal proteins remained largely unchanged. The slowdown of protein dynamics and decreased abundance of the translational machinery may point to the importance of anabolic attenuation in lifespan extension, as suggested by the hyperfunction theory.

Graphical abstract

Dhondt et al. apply a stable isotope labeling by nitrogen in Caenorhabditis elegans (SILeNCe) approach to unravel individual protein turnover dynamics in the long-lived insulin/insulin-like growth factor (IGF-1) receptor mutant daf-2.

Introduction

Cellular protein quality can be maintained by proteolytic elimination of damaged proteins and replacing them with newly synthesized copies, a process called protein turnover (Ward, 2000). Protein turnover rates have been estimated using SILAC (stable isotope labeling by amino acids in cell culture) in prokaryotes and eukaryotes. The last decade has witnessed a growing interest in the analysis of whole-organism proteome dynamics in metazoans using the same approach (Claydon and Beynon, 2012). In recent work, SILAC was applied to monitor protein synthesis throughout life in adult Caenorhabditis elegans (Vukoti et al., 2015) and to investigate food intake (Gomez-Amaro et al., 2015).

Progressive decrease in protein synthesis and proteolytic clearance through the autophagosomal and proteasome systems with age results in a strong increase in protein half-life in many species, including nematodes (Grune, 2000; Lewis et al., 1985; Young et al., 1975). This finding led to the formulation of the protein turnover hypothesis, stating that the increase in protein dwell time with age results in the accumulation of damaged and misfolded proteins. The progressive decrease in general protein turnover might be responsible for the ultimate collapse of proteome homeostasis in aging cells, possibly also driving the aging process itself (Rattan, 1996; Ryazanov and Nefsky, 2002; Taylor and Dillin, 2011). In this vein, it is expected that increased protein turnover rates would help to maintain a young undamaged proteome and extend the lifespan. However, in yeast and C. elegans, genetically induced attenuation of protein synthesis extends, rather than shortens, the lifespan (Hansen et al., 2007; Kaeberlein et al., 2005; Pan et al., 2007; Syntichaki et al., 2007). Moreover, classical ³⁵S pulse-chase labeling and quantitative proteomic studies suggest that low overall protein synthesis is a hallmark of long-lived C. elegans, either by dietary restriction or by mutation in the insulin signaling pathway (Depuydt et al., 2013, 2016; Stout et al., 2013). Similar findings have been reported for diet-restricted mice (Price et al., 2012). Hence, why does reducing protein synthesis promotes lifespan extension? And how can this be reconciled with the protein turnover hypothesis, which predicts enhanced turnover rates in long-lived organisms?

The DAF-16 homologue Forkhead box O (FOXO)/DAF-16 transcription factor drives the increased longevity of the insulin/insulin-like growth factor (IGF-1) receptor mutant daf-2 (Kenyon et al., 1993). We hypothesized that DAF-16-dependent longevity in C. elegans is supported by differential protein turnover. Down-regulating turnover of the majority of proteins could save much energy, which, in turn, could be spent at prioritized maintenance of specific proteins that are crucial to extend the lifespan. To test this hypothesis, we used stable isotope labeling by nitrogen in Caenorhabditis elegans (which we now designate SILeNCe), an efficient method applied in C. elegans before (Geillinger et al., 2012; Krijgsveld et al., 2003). Here we present a SILeNCe dataset that reveals patterns of intracellular protein dynamics in the C. elegans model and shifts of these patterns that occur in the long-lived daf-2 mutant via DAF-16 activation.

Results

Worm Strains Used to Study Protein Turnover in Long-Lived daf-2

To understand the role of protein turnover in DAF-16-mediated lifespan extension, we chose to compare the normal-lived glp-4(bn2) daf-16(mgDf50); daf-2(e1370) triple mutant (reference strain) and the long-lived glp-4(bn2); daf-2(e1370) insulin/IGF-1 receptor mutant. Mutation in DAF-2 causes activation of the downstream transcription factor DAF-16, which, in turn, activates a life maintenance program, doubling the lifespan of C. elegans. To specifically study DAF-16, the major lifespan regulator, we compared two worm strains carrying the daf-2 mutation (causing lifespan extension via DAF-16 activation) and knocked out the daf-16 gene by mutation in the reference strain, thereby nullifying the lifespan extension phenotype. The use of a temperature-sensitive sterile mutant background (glp-4) enables us to focus on somatic cells and avoids purging of the ¹⁵N label via egg laying. The glp-4 mutation has only minimal effects on the lifespan of wild-type and daf-16 nematodes (TeKippe and Aballay, 2010). However, it acts as an enhancer of daf-2 longevity (McColl et al., 2005; McElwee et al., 2004), probably via surplus activation of DAF-16 (Arantes-Oliveira et al., 2002; Hsin and Kenyon, 1999; Lin et al., 2001), which makes it an even better strain to study the DAF-16 effect on lifespan. Mass spectrometry (MS) proteome profiling of glp-4(bn2) revealed only modest changes relative to the N2 wild-type (Krijgsveld et al., 2003). Recently, it was found that glp-4 encodes a valyl aminoacyl tRNA synthetase, suggesting reduced protein translation in the mutant (Rastogi et al., 2015). We verified this prediction with classical ³⁵S pulse-chase labeling and did not observe a difference in protein synthesis and degradation rates between the glp-4 mutant and other control strains, including wild-type N2 (Figure S1). The strategy of comparing glp-4; daf-2 with glp-4 daf-16;daf-2 has been used earlier by others (Depuydt et al., 2013, 2014; McElwee et al., 2007). Lifespan curves of these strains cultured under identical conditions as in this study have been published earlier (Depuydt et al., 2013). For clarity and simplicity, these strains will be designated as reference (glp-4(bn2) daf-16(mgDf50); daf-2(e1370)) and daf-2 (glp-4(bn2); daf-2(e1370)) for the remainder of this article.

DAF-16 Activation Increases Global Proteomic Half-Life

In the SILeNCe experiment, worms were fed ¹⁵N-labeled E. coli bacteria on the second day of adulthood, avoiding possible changes in relative protein composition related to development. From this moment, samples were taken at regular time points. Accurate MS-based quantitative proteomics were performed in a randomized and blind manner (Table S3). The resulting 43 MS datasets were analyzed using a custom R package to extract the proportions of heavy (¹⁵N) and light (¹⁴N) isotope peaks of the peptides, followed by estimation of the corresponding protein half-lives (Figure 1). Data were statistically evaluated using the moderated t test from the limma R package (Ritchie et al., 2015). The use of a conservative pipeline enabled us to generate a very solid dataset with the tradeoff of reduced protein diversity (Figure S2A). However, the proteins covered in our experiment make up 30.8% of total worm mass, as estimated from integrated proteomic C. elegans studies (Figure S2B; Wang et al., 2015). We observed significantly decreased peptide turnover rates for 185 (i.e., 54%) of the analyzed peptides in daf-2 compared with the reference strain (p < 0.05). Only five peptides exhibited significantly faster replacement rates in daf-2, and turnover of 155 (i.e., 45%) peptides was unaltered (Figure 2A). This observation was confirmed with a classical ³⁵S pulse-chase experiment showing overall reduced protein synthesis and degradation rates in daf-2 compared with the reference. Although this downtrend is somewhat less pronounced in the glp-4 genetic background, the global tendency is preserved. Moreover, loss of daf-16 in the daf-2 mutant background reverts protein turnover back to reference levels (Figure S1). The half-lives of redundant peptides from each protein were averaged. In total, we compared 245 overlapping non-redundant proteins from the reference and daf-2 (Table S1). Turnover rates vary widely, with half-lives ranging from 40 hr to more than 40 days. Mutation in daf-2 leads to a shift in median protein half-life from 103 hr to 173 hr (Figure 2B). According to the protein turnover hypothesis, the longer protein half-life in daf-2 should lead to increased damage accumulation. Hence, we specifically screened the liquid chromatography-tandem mass spectrometry (LC-MS/MS) datasets (0 hr time point) for signatures of protein damage via spectral counting. We observed a significantly lower level (∼32%) of methionine oxidation in long-lived daf-2 (Poisson-based generalized linear model, p < 0.001; Figure S3), which parallels earlier carbonylation data (Yang et al., 2007). This may be explained by the elevated expression of antioxidants (Honda and Honda, 1999) and high reductive capacity (Houthoofd et al., 2005) in daf-2 mutants. Other common types of protein damage, such as non-tryptic protein degradation and deamidation, do not differ between young daf-2 and reference worms. Our data show that decreased protein turnover rate does not necessarily correlate to increased accumulation of protein damage. Although it is still unclear to what extent low oxidative damage levels influence the downregulation of protein turnover in daf-2, this process is likely actively regulated as described by Essers et al. (2015).

Figure 1. Study Design.

A¹⁵N metabolic labeling approach was performed to study individual protein turnover rates in long-lived daf-2 C. elegans. Adult worms were sampled at regular time points after pulsing with ¹⁵N-labeled E. coli. Samples were blocked based on the time of the pulse, randomized, and blindly analyzed using LC-MS/MS. The resulting datasets were processed using a custom R package. Protein half-lives of 245 overlapping proteins between reference and daf-2 worms were estimated for five and four biological replicates, respectively.

Figure 2. Slowdown of Protein Turnover in daf-2.

(A) Volcano plot representing peptides with upregulated (red dots), downregulated (blue dots), and unchanged (black dots) turnover and their corresponding change in peptide half-life in daf-2. p Values were obtained from the moderated t test implemented in the limma R package and adjusted for multiple testing according to Benjamini and Hochberg's method.

(B) Histogram showing the distribution of protein half-lives for reference and daf-2 worms. The median protein half-life is 103 hr for the reference strain and 173 hr for daf-2.

Source data are available in Table S1.

Changes in Protein Half-Life Do Not Correlate with Changes in Abundance

We compared the relative changes in protein turnover with relative protein abundance shifts of the same strains published earlier by our group (Depuydt et al., 2013). Correlation analysis was performed on the 167 proteins in common between both datasets (Table S2). Overall, the relative changes in protein abundance because of mutation in daf-2 do not correlate with those in protein turnover (R² = 0.0054, p = 0.3447 for the Pearson correlation; Figure 3). This lack of correlation indicates that the change in abundance of a particular protein is not necessarily dictated by its relative turnover. This is showcased in the fermentation enzymes SODH-1 and ALH-1: both proteins show a similar increase in protein half-lives, although disparate increases in mRNA expression (McElwee et al., 2007) and protein levels (Depuydt et al., 2013) have been observed (Figure 3, inset). Furthermore, a contrasting pattern is found between mitochondrial and ribosomal proteins. The relative mitochondrial protein abundance is increased, whereas turnover rates are decreased in daf-2 mutants. Because mitochondrial proteins represent a large fraction of the total cellular protein, this decrease in mitochondrial turnover may represent a considerable energy saving in daf-2. The ribosomal proteins, another family of highly abundant proteins, follow the opposite pattern: both turnover and relative abundance are decreased in daf-2 mutants (Figure 3). Here, energy saving may be combined with an actual functional reduction, as discussed below.

Figure 3. Changes in Protein Turnover and Abundance Do Not Correlate in daf-2.

Scatterplot illustrating the fold change (log₂) of protein abundance and protein turnover in daf-2 relative to the reference strain (origin). Significant changes in protein turnover (p < 0.05) are signified with closed circles. p values were obtained from the moderated t test implemented in the limma R package and adjusted for multiple testing according to Benjamini and Hochberg's method. Ribo, ribosomal; Mito, mitochondrial; Cytosk, cytoskeletal; ER; Lyso, lysosomal; Nucl, nuclear; Cytopl, cytoplasmic; Extracell, extracellular; Chap, chaperones; Unclass, unclassified proteins. Inset: case study of SODH-1 and ALH-1 demonstrating discordance between transcript (McElwee et al., 2007) and protein (Depuydt et al., 2013) abundance versus protein half-life (fold change [log₂]) in daf-2.

Source data are available in Table S2.

Turnover of the Protein Synthesis Machinery Is Slowed Down in daf-2

The half-life of proteins associated with the protein synthesis machinery is significantly increased in daf-2 mutants. In reference animals, ribosomal proteins have half-lives of approximately 100 hr, whereas, in daf-2 mutants, these figures are increased 2- to 4-fold. Proteins of the small (40S) and large (60S) ribosomal subunits show similar turnover rates, and a comparable increase in half-life was observed for both subunits in daf-2 (Figures 4 and 5A). In addition, we found a similar pattern for translation factors, also involved in protein synthesis (Figure 5B). Endoplasmic reticulum (ER)-bound ribosomes synthesize proteins, which are translocated into the ER lumen, where they are subsequently folded and assembled with the aid of specialized proteins (Vázquez-Martínez et al., 2012). Interestingly, we found prolonged half-lives for all detected ER proteins predicted to be involved in these processes in the daf-2 mutant (Figure 5C). The average protein half-life of ER proteins is only 86 hr in reference worms and shifts significantly toward 173 hr in daf-2 mutants (two-sample Student's t test, p < 0.0001). In the same line, daf-2 mutants show a loose pattern of rough ER (RER) cisternae dispersed through the cytoplasm, sparsely studded with ribosomes (Figure 6), whereas, in the reference strain, abundant RER is densely stacked in Terasaki ramps to accommodate maximum protein synthesis within the confined cells (Heald and Cohen-Fix, 2014; Terasaki et al., 2013). However, despite the dispersed RER arrangement, decreased ribosomal load, and lower refreshment rates of ER-associated proteins, the half-lives of RER-processed proteins (secretory proteins and lysosomal proteins), remain unchanged in daf-2 mutants (Figures 4A and 4B, insets).

Figure 4. Protein Dynamics in Subcellular Compartments.

(A and B) Scatterplot comparing significant (A) and non-significant (B) changes of protein half-lives in daf-2 versus the reference strain. Functional groups are color-coded and summarized in pie chart insets. p Values were obtained from the moderated t test implemented in the limma R package and adjusted for multiple testing according to Benjamini and Hochberg's method. Error bars are SEM.

(C and D) Protein half-lives in different cellular compartments with differentiation of membrane-bound proteins (open symbols) and free proteins (filled symbols). Significantly changed half-lives are represented in (C) (two-sample Student's t test, p < 0.05), and non-significant changes are depicted in (D).

Source data are available in Table S4.

Figure 5. Individual Protein Half-Lives of daf-2 and Reference Replicates and Estimation of Mitochondrial Turnover Rates with the Photoswitchable Reporter Dendra2.

(A–C, E–H, and J–M) Heatmap representation of individual protein half-lives of daf-2 and reference replicates. The colors indicate a relative decrease (red) or increase (blue) of protein half-life compared with the median of the reference strain (white, 103 hr). Color limits signify the 5^th (41 hr) and the 95^th (291 hr) percentile. Each tile represents a biological replicate. p Values were obtained from the moderated t test implemented in the limma R package and adjusted for multiple testing according to Benjamini and Hochberg's method. Asterisks indicate significant differences between both strains. *p < 0.05, **p < 0.01, ***p < 0.001, respectively.

(D) Confocal images of jrIs5 transgenic worms expressing Dendra2 under the constitutive rps-0 promotor and targeted to the mitochondria with the mitochondrial localization signal of gas-1. Left: overview of Dendra2-expressing worms. Scale bar, 100 μm. Right: detail of the midsection of the body. Scale bar, 10 μm.

(I) Linear regression of the decline in red Dendra2 fluorescence after photoconversion in reference versus daf-2 strains (F-test, p < 0.0001).

Source data are available in Table S4.

Figure 6. Organization of the Rough Endoplasmic Reticulum.

(A and B) Transmission electron micrographs of intestinal cells of reference (A) and daf-2 worms

(B). Arrows indicate the RER. int, intestine; lys, lysosome; m, mitochondrion; g, glycogen; lip, lipid. Scale bars, 1 μm.

Lysosomal Protein Turnover: High in the Lumen but Low in the Membrane

Autophagy is the major degradation pathway of macromolecules and organelles, and this process involves lysosome-mediated catabolism (Yorimitsu and Klionsky, 2005). We identified four lysosomal proteins whose turnover is very fast (average half-life, 38 hr) and similar in the reference strain and daf-2 mutant, whereas the half-life of only one hydrolase, the cathepsin ASP-4 (1.6-fold down, p = 0.0436) is slightly altered (Figure 5J). Acidification of the lysosomal lumen depends on the activity of proton pumps, vacuolar-type H⁺ ATPase complexes, present in the lysosomal membrane (Mindell, 2012). As opposed to the fast intraluminal hydrolase turnover, we found very slow replacement rates for subunits of the V1 cytosolic complex in both reference and daf-2 worms (average half-life, 225 hr; Figure 5K). Only two subunits, VHA-8 and VHA-14, show a slight but significant reduction in protein half-lives in daf-2 (p = 0.0328 and p = 0.0319, respectively). Unexpectedly, the turnover rate of VHA-11, the subunit C homolog of the V1 complex, is very fast in reference worms (50 hr, one replicate) and exceptionally high in daf-2 worms (average half-life, 9 hr; three biological replicates). In general, however, we observed significantly longer protein half-lives for membrane-associated proteins compared with intraluminal proteins for both reference and daf-2 mutants (unpaired Student's t test, p < 0.0001 and p = 0.00021, respectively). A similar distinction between membrane-bound and free proteins was observed within mitochondria.

Mitochondrial Turnover Is Decreased in daf-2

Mitochondrial proteins may be vulnerable to oxidative damage because of their proximity to reactive oxygen species (ROS)-generating centers (Murphy, 2009). Therefore, it could be expected that mitochondrial proteins exhibit high refreshment rates to maintain their important vital function. Contrary to prediction, 65% of the identified mitochondrial proteins exhibit longer half-lives in daf-2 compared with the reference strain (p < 0.05), whereas the remaining proteins do not show any significant change. To corroborate this finding, we expressed Dendra2 in the mitochondria of all cells (Figure 5D). We measured the red Dendra2 fluorescence intensity of transgene worms at different time points after photoconversion for the reference and daf-2 worms and monitored the decrease of the relative red signal over time as a proxy of mitochondrial protein breakdown. Consistent with our SILeNCe data, the decrease of the red fluorescence is slowed down significantly in daf-2 mutants compared with the reference strain (F-test, p < 0.0001; Figure 5I). With this method, we estimated the half-life of the mitochondrially targeted Dendra2 protein to be 44 hr in the reference strain and 77 hr in daf-2 mutants. Interestingly, the turnover of Dendra2 is comparable with the fastest turnover rates observed for mitochondrial matrix-resident proteins (Figure 5E). This high turnover might be due to the fact that Dendra2 is a foreign protein and, therefore, probably more likely to be degraded by mitochondrial proteases. As in lysosomes, turnover of membrane-bound mitochondrial proteins is significantly slower compared with the turnover of free proteins for both strains studied, although this difference is less prominent in daf-2 mutants (Figures 4C and 4D; two-sample Student's t test, p < 0.0001 and p = 0.0096, respectively). Widely divergent protein half-lives were found for mitochondrial matrix proteins, ranging from 52–183 hr in the reference strain and 103–358 hr in daf-2 mutants (Figure 5E). In addition, mitochondrial protein turnover shows limited uniformity within biochemical pathways in reference worms. For instance, the citric acid cycle enzymes fumarase (FUM-1) and malate dehydrogenase (MDH-1) show half-lives of 81 and 162 hr, respectively. Nevertheless, an overall reduction in protein half-life of the carbohydrate metabolic enzymes is observed in daf-2 (Figure S4). On the other hand, proteins of the mitochondrial membrane, including subunits of the electron transport chain and adenine nucleotide transporters, display very slow turnover rates compared with most matrix proteins. This slow turnover is further decreased in daf-2 for most proteins (Figures 5F–5H).

Minimal Turnover of Cytoskeletal and Muscle-Related Proteins

Proteins of the muscle contractile apparatus, such as thin filament actins and myosin class II light-chain proteins, exhibit extremely long protein half-lives (on average 25 and 41 days, respectively; Figure 5L). These numbers suggest that myofilaments only undergo occasional turnover during the entire C. elegans lifespan. We observed slow turnover for the isoform myosin class II heavy-chain protein UNC-54 (half-life on average 298 hr), but we did not detected any label incorporation at all for the other three isoforms, which suggests a complete lack of turnover. On the other hand, the invertebrate-specific paramyosin UNC-15, which interacts with these heavy-chain proteins, shows remarkably fast turnover in both reference and daf-2 mutants (average half-life, 66 hr). Other proteins involved in muscle physiology show moderate protein half-lives in reference worms (140 hr on average) and are somewhat delayed in daf-2 mutants without reaching significance (202 hr on average). Only the putative creatine kinase W10C8.5 and the triosephosphate isomerase TPI-1 have significantly reduced turnover rates in daf-2 (p = 0.0062 and p = 0.0293, respectively). These proteins, involved in the supply of energy molecules in muscle cells (Ralser et al., 2007; Wallimann et al., 2011), show strong DAF-16-dependent upregulation (Depuydt et al., 2013). Increased protein levels in combination with lower rates of turnover suggest an upkeep of these proteins. Low turnover was also found for cytoskeletal proteins that reside in the cytoplasm. ACT-5, an intestine-specific actin, displays a long protein half-life analogous to muscle actins. The constituents of microtubules, alpha- and beta-tubulins, are less stable, with an average half-life around 141 hr in reference worms. Protein half-lives of beta-tubulins remain unchanged in daf-2 (168 hr on average), whereas turnover of alpha-tubulins is significantly attenuated (average half-life, 261 hr; p < 0.05).

Reduced Turnover of Stress Response Proteins in daf-2

Much evidence points to a strong correlation between longevity and cytoprotective mechanisms, linking lifespan extension to an increased tolerance to a variety of stressors (Shore et al., 2012). We observed a wide range of stress protein half-lives, ranging from 27–159 hr in the reference worms and 41–236 hr in daf-2 mutants (Figure 5M). Stress-induced cytosolic chaperones, including heat shock proteins (HSPs), show fast to moderate turnover rates in the reference strain (average half-life, 68 hr). Surprisingly, these proteins show reduced turnover in long-lived daf-2 mutants (average half-life, 132 hr), with a significant diminution for DAF-21, HSP-1, HSP-70, and CDC-48.2. These observations are in line with the extended protein half-lives for HSPs of the ER, as described above. Several HSPs are downregulated in daf-2 (Depuydt et al., 2013), including DAF-21 and HSP-1, which implies that, contrary to expectation, daf-2 worms invest less energy in the abundance and turnover of HSPs. In addition, other proteins involved in cytoprotection showed a similar pattern, including glutathione-S-transferase (GST) 36 (1.48-fold down, p = 0.0107), involved in xenobiotic detoxification (Ayyadevara et al., 2007), and superoxide dismutase (SOD) 1 (Figure 5M; 1.82-fold down, p = 0.0144). However, not all daf-2 stress proteins show decreased turnover; e.g., turnover of SIP-1 (average half-life, 33 hr), a small heat shock protein, is unaffected in daf-2 mutants. A similar pattern was observed for CLEC-63, an antimicrobial protein involved in the worms' innate immunity (average half-life, 49 hr). Our data suggest that, to some extent, daf-2 saves energy by reducing the turnover of specific cytoprotective proteins.

Dendra2 Photoswitching Validates the SILeNCe Approach

As an alternative approach to SILeNCe, we constructed reporters expressing the green-to-red photoswitchable fluorescent protein Dendra2 from the octocoral Dendronephthya (Gurskaya et al., 2006). Green-to-red fluorescence photoconversion is an irreversible process. Hence, surveying the decrease of red Dendra2 fluorescence over time (protein degradation) is an alternative estimate of protein half-live that is insensitive to label uptake or re-use. To validate the slowdown of ribosomal turnover and the short lysosomal protein half-lives assessed with SILeNCe (Figure 7A), we expressed Dendra2 fused to the ribosomal RLA-1 protein and lysosomal ASP-4 hydrolase in long-lived daf-2, respectively. We monitored the change in red fluorescence intensity with confocal microscopy after 96 hr (for RLA-1) and 48 hr (for ASP-4) after photoconversion. We observed a prolonged protein half-life of RLA-1 in the long-lived daf-2 compared with the reference worms fed with daf-16 RNAi (two-sample Student's t test, p = 0.0019; Figures 7B and 7C), which confirms the reduced turnover of this ribosomal subunit upon DAF-16 activation (Figure 7A). Likewise, we corroborated the preserved fast protein turnover of ASP-4 via the Dendra2 method (two-sample Student's t test, p = 0.6278; Figures 7B and 7D). Hence, estimating protein turnover using Dendra2 photoconversion validated our observations from the SILeNCe method.

Figure 7. Validation of the SILeNCe Method with the Dendra2 Approach.

(A and B) Protein half-lives of RLA-1 and ASP-4, estimated via SILeNCe pulse labeling (A) and the Dendra2 pulse-chase method (B). Red fluorescence decline was determined in glp-4 daf-2 worms expressing RLA-1∷Dendra2 and ASP-4∷Dendra2 translation reporters and treated with daf-16 (reference) and empty vector (L4440) RNAi (daf-2) (minimum six individually tracked worms per condition; two-sample Student's t test; p = 0.0019 and p = 0.6278 for RLA-1 and ASP-4, respectively). Error bars are SEM.

(C and D) Representative confocal images of glp-4 daf-2;rla-1p∷rla-1∷Dendra2 (C) and glp-4 daf-2; asp-4p∷asp-4∷Dendra2 (D) transgenic worms grown on daf-16 RNAi (reference) and empty vector L4440 (daf-2). Images were taken 96 hr (RLA-1) and 48 hr (ASP-4) after photoconversion. Scale bars, 20 μm.

Discussion

Proteins with long dwelling times are likely to accumulate all sorts of molecular damage during aging. Thus, one could expect that, by increasing the protein turnover rate, cellular damage accumulation is prevented and the lifespan is extended. However, several studies are in conflict with the turnover hypothesis, as they observe overall decreased protein synthesis rates in C. elegans longevity mutants, or, vice versa, translation inhibition results in lifespan extension (Depuydt et al., 2013; Hansen et al., 2007; Pan et al., 2007; Stout et al., 2013; Syntichaki et al., 2007; Van Raamsdonk and Hekimi, 2009; Yang et al., 2007). Therefore, we hypothesized that, in daf-2 mutants, energy is saved by downregulation of turnover of the majority of proteins and reinvested in prioritized turnover of specific proteins that are crucial to somatic maintenance. To that end, we investigated the turnover dynamics of individual proteins using a SILeNCe metabolic labeling method. Contrary to our hypothesis, we did not discover a delineated set of proteins with turnover priority in daf-2 mutants. The majority of the detected proteins (56%) exhibit prolonged half-lives in daf-2, whereas turnover of the remaining proteins is unchanged. Only three proteins (CPN-3, ASP-4, and VIT-6) display marginally significant higher turnover rates in daf-2, but they lack a clear biological relationship. This dramatic change in protein turnover in long-lived daf-2 was also observed in a parallel study (Visscher et al., 2016 [this issue of Cell Reports]).

One of our most notable observations is the drastic slowdown in turnover of the translation machinery in daf-2 mutants. This slowdown coincides with decreased levels of ribosomal proteins and enzymes with predicted function in ribosome assembly and biogenesis we and others observed earlier (Depuydt et al., 2013; Stout et al., 2013; Walther et al., 2015) and probably relates to the decreased protein synthesis rates in this mutant. In line with this, we also observed a dispersed organization of the RER cisternae with reduced ribosomal count in long-lived daf-2 worms, which is considered a sign of diminished protein translation (Hansen et al., 2007). Corroborating this is a recent work suggesting a specific translational block in daf-2 by specific binding of the long non-coding RNA tts-1 to the ribosomes (Essers et al., 2015). Our observation of decreased protein turnover in daf-2 mutants is not entirely surprising. The insulin/IGF1 signaling pathway is a main activator of anabolic metabolism; hence, it is conceivable that mutants in this pathway show reduced protein turnover. This reduction allows the worm to save much energy, which may be diverted to other processes that support longevity, such as the synthesis of trehalose, a chemical chaperone that stabilizes membranes and proteins, for which a role in daf-2 longevity has already been shown (Depuydt et al., 2016; Honda et al., 2010). Moreover, this mutant displays reduced protein degradation activity (Stout et al., 2013), although recent work challenges this finding (Walther et al., 2015). The overall slow protein turnover agrees with the hyperfunction theory, stating that aging is caused by excess biosynthesis (hypertrophy) (Blagosklonny, 2012; Gems and de la Guardia, 2013). Thus, inhibited insulin/IGF-1 signaling reduces anabolic pathways, which, in turn, lowers hypertrophy, decelerating the onset of its related pathologies.

Earlier, it has been shown that reduced insulin/IGF-1 signaling and dietary restriction rely on autophagy to extend the lifespan (Hansen et al., 2008; Meléndez et al., 2003). Hence, the preserved turnover rates of lysosomal proteins might be crucial to sustain autophagy in daf-2 mutants. However, our data do not support increased bulk protein degradation because most protein half-lives are prolonged or unchanged in daf-2. The turnover rates of lysosomal hydrolases are among the fastest found in the cell, which can be explained by the fact that the lysosomal lumen is a strong proteolytic environment in which accidental breakdown of lysosomal components is to be expected. Opposite to these hydrolases, subunits of the V-type ATPase proton pumps, present in the lysosomal membrane, undergo slow turnover, indicating that lysosomes, apart from their content, are not rapidly replaced as a whole. Also, V-type ATPase subunits are part of large complexes that may be intrinsically more stable and resilient to protein breakdown. One remarkable exception is the subunit C homolog VHA-11, which is known to reversibly dissociate from this complex in yeast cells (Iwata et al., 2004). It shows extremely fast turnover in the reference strain, which is even enhanced in daf-2 mutants. V1/V0 dissociation is thought to be an energy-conserving response (Kane, 2006), and recent work shows a general role of V-type ATPase complexes in controlling metabolic programs (Zhang et al., 2014). It is possible that VHA-11 functions as an energy-dependent switch, downregulating lysosomal acidification and protein degradation in the daf-2 mutant. More detailed evaluation of a lysosomal role in the regulation of protein metabolism in daf-2 is necessary.

Similar to lysosomes, we detected a clear discrepancy in turnover between free and membrane-bound mitochondrial proteins. In addition, the half-lives of mitochondrial proteins vary widely, and no uniformity is found within biochemical pathways. This variability in turnover suggests that mitophagy is not the predominant process for elimination of mitochondrial proteins in C. elegans under normal conditions. Matrix proteases and extra-mitochondrial degradation mechanisms are likely co-responsible for mitochondrial protein dynamics (Lau et al., 2012).

Our data disprove that daf-2 longevity is supported by enhanced mitochondrial turnover rates because the majority of protein half-lives in this organelle are downregulated. One may argue that this is because daf-2 worms activate their antioxidant defense systems (Honda and Honda, 1999), making the worms less dependent on high turnover of mitochondrial components to avoid damage accumulation. Although we found lower levels of methionine oxidation in the daf-2 LC-MS/MS data, it should be noted that levels of oxidative damage in mitochondria of young adult wild-type and daf-2 animals are similar (Brys et al., 2010) and that in vivo hydrogen peroxide production in the muscle is not altered in young adult daf-2 (Knoefler et al., 2012). Moreover, expression of sod is not required for lifespan extension in daf-2 mutants (Doonan et al., 2008). Despite their low turnover, daf-2 mitochondria are highly abundant (Depuydt et al., 2013) and show enhanced coupling (Brys et al., 2010). Our previous experiments suggest that, in this mutant, oxidative phosphorylation may be supported by anaerobic mitochondrial pathways (Brys et al., 2010). Although our data challenge extensive recycling of mitochondrial components in the long-lived daf-2, selective and specific autophagy of damaged mitochondria might be necessary for its increased lifespan, as suggested recently (Palikaras et al., 2015).

Proteins of the muscle contractile apparatus undergo extremely slow turnover in both reference and daf-2 worms. Despite their high abundance, half-lives of myosin class II heavy-chain proteins could not be estimated, which is indicative of rare replacement events. These results are consistent with those reported in recent work on protein synthesis during the C. elegans lifespan (Vukoti et al., 2015), reporting the slow appearance of newly synthesized motility-related proteins. Although reduced insulin/IGF1 signaling is often associated with muscle atrophy in mammalian systems (Sandri et al., 2004), our previous work showed a highly preserved biomass of striated body wall muscle in daf-2 compared with the reference strain (Depuydt et al., 2013). In addition, the upkeep of mitochondrial proteins further supports the maintenance of muscle functionality in these mutants. These findings do not exclude the possibility that non-filamental muscle proteins undergo more rapid and even enhanced turnover in daf-2, as suggested in a study using a LacZ reporter (Szewczyk et al., 2007). Cytoskeletal components that are not restricted to muscle cells show medium turnover rates, for which only alpha-tubulins are significantly attenuated in daf-2, suggesting slightly reduced cellular dynamics in these mutants compared with the reference strain.

Fast to moderate turnover was detected for proteins involved in cytoprotective mechanisms in reference worms. Although microarray studies show strong DAF-16-mediated upregulation of stress-related genes (McElwee et al., 2003; Murphy et al., 2003), several HSPs are downregulated at the protein level (Depuydt et al., 2013) and show a prolonged half-life in daf-2 mutants in this study. The slow turnover of these chaperones could be related to the recent observation that small HSPs are mostly enriched in the insoluble fraction of old daf-2 worms, playing a role in protective aggregate formation to maintain proteome balance (Walther et al., 2015). However, it is currently not clear whether these phenomena would be of any significance in the young worms used in our study. Nevertheless, some cytoprotective proteins do show unchanged fast turnover, implicating the complex and specific regulation of stress proteins in daf-2 mutants.

In summary, we observed an overall slowdown of protein turnover in long-lived daf-2 worms using a SILeNCe approach. Most prominently, this was reflected in, and probably caused by, drastic lowering of translation machinery turnover in daf-2. In agreement with our observations, researchers demonstrated extended ribosomal and mitochondrial half-lives in long-lived, calorie-restricted mice (Karunadharma et al., 2015; Price et al., 2012). Hence, it seems that high protein turnover is not essential to support lifespan extension, but it is still unclear whether the observed slowdown of protein turnover is a causal factor. Recent evidence showing that alleviation of protein synthesis inhibition in daf-2 mutants partially rescues its longevity phenotype strongly supports this causal relation (Essers et al., 2015). Reduced protein turnover may be related to some typical features of the daf-2(e1370) mutant, including reduced food uptake, altered amino acid metabolism, and a lower energetic flux. The Eat phenotype, and thus diminished amino acid uptake and likely lower tRNA loading, particularly typify daf-2 physiology, which possibly supports silenced anabolism and, consequently, decreased global proteome turnover (Davies et al., 2015; Depuydt et al., 2014; Fuchs et al., 2010). Although these aspects of daf-2 physiology could confound pulse-chase analysis of protein turnover, we validated that limited label uptake or amino acid re-utilization was not causal to the reduced turnover measured in our ³⁵S and SILeNCe experiments. This was done by the Dendra2 approach, which is not influenced by label uptake and reutilization. Of note, the absolute protein half-lives resulting from the Dendra2 method differ from those estimated by the SILeNCe approach. This discrepancy originates from differences in parameters used for the calculation of the protein half-lives. In the SILeNCe experiment, half-life is based on synthesis and degradation dynamics (¹⁵N incorporation versus ¹⁴N loss), whereas the Dendra2 method relies on degradation rates. Nevertheless, relative patterns of protein half-lives are consistent in both approaches. Low protein turnover does not lead to an inferior proteome in daf-2 worms, as shown by their well-maintained muscle mass (Depuydt et al., 2013) and mitochondrial function (Brys et al., 2010), locomotive ability (Gems et al., 1998), and metabolic capacity upon mechanical stimulation (Braeckman et al., 2002). Therefore, we believe that reduced insulin/IGF-1 signaling results in an energy-saving state along with the improved protection of proteins. The preserved somatic integrity in daf-2 possibly relies on chemical chaperones (Depuydt et al., 2016; Honda et al., 2010) or small heat shock proteins (Walther et al., 2015). Our findings and suggestions are also compatible with the hyperfunction theory, which proposes that the attenuation of protein synthesis might reduce hypertrophy, thereby retarding the accumulation of irrelevant macromolecules that otherwise cause age-related cellular malfunction (Blagosklonny, 2012; Gems and de la Guardia, 2013).

Experimental Procedures

Strains

GA154 glp-4(bn2ts)I;daf-2(e1370)III (long-lived) and GA153 glp-4(bn2ts)I daf-16(mgDf50)I; daf-2(e1370)III (reference strain) were kindly provided by David Gems at the University College of London. The strains were maintained as described previously (Depuydt et al., 2013). The transgenic lines generated by biolistic transformation and crossing were daf-2(e1370)III;jrIs5[unc-119(+) rps-0p∷mls∷Dendra2] and daf-16(m26)I;daf-2(e1370)III; jrIs5[unc-119(+) rps-0p∷mls∷Dendra2]. The strains glp-4(bn2)I;daf-2(e1370)III;jrEx17[rla-1p∷rla-1∷Dendra2] and glp-4(bn2)I;daf-2(e1370)III;jrEx18[asp-4p∷asp-4∷Dendra2] were created by microinjection.

SILeNCe

Metabolic ¹⁵N and ¹⁴N Labeling of the Escherichia coli Bacteria

E. coli K12 was freshly grown overnight at 37°C while shaking at 120 rpm in minimal medium containing 42 mM Na₂HPO₄, 22 mM KHPO₄, 9 mM NaCl, trace elements (17 mM EDTA, 3.1 mM FeCl₃·6H₂O, 0.62 mM ZnCl₂, 0.076 mM CuCl₂·2H₂O, 0.042 mM CoCl₂·6H₂O, 0.16 mM H₃BO₃, and 0.0068 mM MnCl₂·6H₂O), glucose 20% (w/v), 1 mM MgSO₄, 0.3 mM CaCl₂, 0.0041 mM biotin, 0.0033 mM thiamin, and 93 mM ¹⁵N- or ¹⁴N-containing NH₄Cl (0.22-μm filter, sterilized). Fully labeled bacterial cultures were obtained by inoculating a single colony with an inoculation loop into 250 ml of minimal medium. Overnight cultures reached an optical density of approximately 1.5 (A₅₅₀) and were concentrated 50-fold.

Culturing and Sampling of C. elegans

Synchronized L1 nematodes were plated on nitrogen-free agarose (1.2%) plates containing NaCl (0.3%), cholesterol (0.0005%), 1 mM CaCl₂, 1 mM MgSO₄, and 25 mM K₂HPO₄/KH₂PO₄ (pH 6.0) and a lawn of freshly grown ¹⁴N-labeled E. coli K12 cells. To prevent dauer formation, glp-4(bn2ts)I; daf-2(e1370)III were grown at 17°C until the third larval stage (L3) and then switched to 24°C for the remainder of the experiment. Worms that had freshly molted to adults (adult day 0) were washed three times and transferred into culture flasks (not exceeding 1,500 worms/ml) containing S-basal (100 mM NaCl, 50 mM potassium phosphate [pH 6.0]), 12.93 μM cholesterol, 75 μM 5-fluoro-2′-deoxyuridine (Acros Organics), and ¹⁴N-labeled E. coli K12 cells (A₅₅₀ = 1.0). On day 2 of adulthood, worms were collected and washed thoroughly (2 × S-basal, 1 × S-basal containing 2.5 mM EDTA, 2 × S-basal) to remove bacteria. Next, worms were pulsed in new culture flasks containing S-basal, 12.93 μM cholesterol, 75 μM 5-fluoro-2′-deoxyuridine (Acros Organics), and ¹⁵N-labeled E. coli K12 cells (A₅₅₀ = 1.0). From this moment, worms were sampled after 0, 15, 20, 27, 40, and 72 hr of ¹⁵N pulsing. Animals were washed thoroughly (2 × S-basal, 1 × S-basal containing 2.5 mM EDTA, 2 × S-basal) to remove bacteria, after which dead worms were removed via Percoll (Sigma-Aldrich) washing. The worm pellet was resuspended in 200 μl denaturing solution (8 M urea, 1 mM EDTA, 10 mM DTT, and 50 mM TrisHCl [pH 8.0]) and immediately dripped into liquid nitrogen and stored at −80°C. The sample size was chosen based on our previous study (Depuydt et al., 2013). We sampled all the different post-pulse time points in five biological replicates for both glp-4(bn2ts)I daf-16(mgDf50)I; daf-2(e1370)III and glp-4(bn2ts)I; daf-2(e1370)III.

Quantitative Proteomics

Randomized Study Design

Samples were processed in batches comprising all the biological replicates of both reference and daf-2 worms sampled at one time point post-labeling. Samples within a batch were analyzed blindly and in random order (Table S3).

Preparation of the Tryptic Peptide Samples

The frozen worm beads (∼1,500 worms in denaturing solution) were homogenized using a BioPulverizer (BioSpec) pre-conditioned in liquid nitrogen. The fine powder was recovered into a 1.5-ml tube, thawed, centrifuged (2 min, 5,000 rpm), and sonicated for 30 s in a 5510 Branson ultrasonic water bath (Branson Ultrasonics). Protein concentrations were determined using a Coomassie assay. Further sample processing steps, LC/MS peptide identification, and data processing strategies are provided in the Supplemental Experimental Procedures.

Functional Analysis and Data Visualization

The UniProtKB Protein Knowledgebase (UniProt 2015), the Database for Annotation, Visualization, and Integrated Discovery (DAVID 2014) (Dennis et al., 2003), and Wormbase (version WS246) were used for protein annotation and evaluation of functionally and spatially related proteins (Table S4). Heatmap visualization of different protein half-lives was done with the MultiExperiment Viewer, part of the TM4 microarray software suite (Saeed et al., 2003). Euclidean distances between proteins were calculated and used as a distance metric in a hierarchical clustering analysis (HCA) together with an average pairwise distance as linkage method. Graphs and statistical tests (including tests for normality) were performed with GraphPad Prism version 6.05 for Windows (GraphPad).

Correlation Analysis of Protein Abundance and Half-Lives

In total, 167 common proteins were detected comparing the SILeNCe dataset with the previously obtained proteomic dataset of Depuydt et al. (2013) (Table S2). The “dietary restriction” condition was removed from the latter, and the set was re-normalized. Pearson correlation analysis was performed between log₂-transformed ratios of protein abundance and protein half-lives (log₂ fold change) (GraphPad Prism version 6.05 for Windows).

Dendra2

jrIs5[unc-119(+)rps-0p∷mls∷Dendra2] was created using biolistic transformation (PDS-1000/He System, Bio-Rad) to study mitochondrial turnover. This reporter is a concatenate of the ubiquitous ribosomal protein subunit promoter rps-0p, the gas-1 mitochondrial localization sequence, and the Dendra2 gene (Evrogen). This transgenic strain was crossed with the daf-2(e1370)III mutant. The reference strain was obtained by crossing the latter strain with the daf-16(m26)I genetic background. An device emitting intense violet light was hand-built (4 × 20 W, 415-to420-nm light emitting diodes (LEDs), Zhuhai Tianhui Electronics) for mass photoconversion of worms without affecting lifespan and fecundity (data not shown). Residual UV and heat produced by the LEDs were eliminated using UV filters and heat mirrors, respectively. Day 1 adult worms were photoconverted during 30 min and, subsequently, red Dendra2 fluorescence intensity was measured at regular time points using a Victor² 1420 multilabel counter (PerkinElmer). Fluorescence was normalized to protein content as determined using a BCA protein assay kit (Thermo Scientific). The slopes, representing mitochondrial Dendra2 turnover, were evaluated in GraphPad Prism using linear regression analysis.

Ribosomal and lysosomal protein degradation was assayed by microinjecting an rla-1p∷rla-1∷Dendra2 or asp-4p∷asp-4∷Dendra2 construct containing a 2-kb upstream and 1-kb downstream sequence of the respective gene in the glp-4(bn2)I;daf-2(e1370)III mutant. Synchronized L1 worms were grown on nematode growth medium (NGM) plates seeded with freshly induced RNAi bacteria (L4440 empty vector and daf-16), incubated at 17°C until the third larval stage (L3), and then switched to 24°C for the remainder of the experiment. daf-16 RNAi resulted in the expected abrogation of the longevity phenotype of glp-4;daf-2 worms (Figure S5). On the second day of adulthood, worms (n = 6/condition) were photoconverted and immediately imaged using confocal microscopy. Subsequently, worms were rescued and replaced on fresh RNAi plates. After 48 or 96 hr, the same individuals were re-imaged using the same camera settings specific for each animal. Confocal images were acquired using a Nikon TiE-C2 confocal microscope with objectives Plan 4× 0.10 chrome-free infinity-corrected (CFI) (numerical aperture [NA] 0.1, dry), 20× CFI Plan Apochromat VC (NA 0.75, dry), and Plan Apo VC 60×A WI differential interference contrast (DIC) N2 (NA 1.20, water immersion [WI]) and the NIS Elements imaging software (version 4.13.03). The Dendra2 fluorochrome was excited with a 561-nm solid-state laser, and the emission was collected with a 605/55-nm band-pass filter. The mean intensity per pixel of red Dendra2 fluorescence was determined with ImageJ (Schneider et al., 2012). Protein half-lives were calculated as the fractional decline of red fluorescence per hour based on two measured time points.

Data Availability

The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium (Vizcaínoetal., 2014) via the PRIDE partner repository with the dataset identifiers PXD002317 and 10.6019/PXD002317.

Highlights.

• Approximately 56% of the daf-2 proteome showed decreased turnover

• This decrease was most prominent for components of the translation machinery

• Functional and spatial groups of proteins show distinct turnover patterns

• High protein turnover is not essential to support lifespan extension in daf-2