Intracellular protein glycosylation modulates insulin mediated lifespan in C.elegans.

O-linked-β-N-acetylglucosamine (O-GlcNAc) modification is a regulatory, nuclear and cytoplasmic post-translational glycosylation of proteins associated with age-related diseases such as Alzheimer's, Parkinson's, and type II diabetes. Global elevation of O-GlcNAc levels on intracellular proteins can induce insulin resistance, the hallmark of type II diabetes, in mammalian systems. In C. elegans, attenuation of the insulin-like signal transduction pathway increases adult lifespan of the nematode. We demonstrate that the O-GlcNAc cycling enzymes OGT and OGA, which add and remove O-GlcNAc respectively, modulate lifespan in C. elegans. Median adult lifespan is increased in an oga-1 deletion strain while median adult life span is decreased upon ogt-1 deletion. The O-GlcNAc-mediated effect on nematode lifespan is dependent on the FoxO transcription factor DAF-16. DAF-16 is a key factor in the insulin-like signal transduction pathway to regulate reproductive development, lifespan, stress tolerance, and dauer formation in C. elegans. Our data indicates that O-GlcNAc cycling selectively influences only a subset of DAF-16 mediated phenotypes, including lifespan and oxidative stress resistance. We performed an affinity purification of O-GlcNAc-modified proteins and observed that a high percentage of these proteins are regulated by insulin signaling and/or impact insulin pathway functional outcomes, suggesting that the O-GlcNAc modification may control downstream effectors to modulate insulin pathway mediated cellular processes.


INTRODUCTION
pharmacologically or genetically, has been demonstrated in mammalian systems to induce insulin resistance. Aberrant O-GlcNAc modification is implicated in human diabetes, and age related neurodegenerative diseases [5][6][7].
In C. elegans, the conserved insulin-like signaling pathway regulates lifespan as well as reproductive development, stress response, and dauer formation [8,9]. Upon insulin binding, the insulin receptor (DAF-2) catalyzes receptor tyrosine phosphorylation events to activate phosphatidylinositol-3 kinase (PI3K/AGE-1) [10]. AGE-1 activates the effector kinases, AKT-1 and SGK-1, which antagonize the forkhead box O (FoxO) transcription factor DAF-16 by preventing its nuclear localization (see Figure 6). A reduction in DAF-2-me-diated signaling allows DAF-16 to localize to the nucleus where it regulates a large number of genes essential for metabolism, energy storage, reproductive development, immunity, stress resistance, and other novel processes which collectively modulate lifespan in C. elegans [11][12][13]. Adult lifespan extension upon reduction in insulin signaling has also been reported in Drosophila, mice, and recently in a centenarian human population [14,15]. DAF-16 function is controlled by additional mechanisms beyond its nuclear localization, as neither constitutive nuclear localization nor over-expression is sufficient to induce maximum lifespan extension [16].
The mechanism by which insulin signaling mediated through FoxO/DAF-16 produces distinct biological outcomes in response to varying levels of insulin signaling remains ambiguous [12,17]. extended adult lifespan while the ogt-1(ok1474) mutant has a reduced lifespan relative to wild-type animals, assayed at 20°C. Lifespan curves are based on pooled data from multiple replicates described in Table 1 The C. elegans genome contains a single OGT ortholog, ogt-1, and a single OGA ortholog, oga-1. In the current study we demonstrate that the O-GlcNAc cycling enzymes modulate median adult lifespan in C. elegans. Intriguingly, we delineate downstream functions of the insulin signaling pathway that are O-GlcNAc dependent, and others that appear to be O-GlcNAc independent. Finally we used a biochemical approach to identify putative O-GlcNAc target proteins, and found that the majority of these proteins are regulated by insulin signaling, and a significant percentage functionally impact lifespan. Our work therefore suggests a potential mechanism by which the post-translational O-GlcNAc modification of proteins downstream of the insulin pathway could differentially modulate insulin-mediated signaling outcomes.
The ogt-1 mutant animals appear to age prematurely, as exemplified by markedly slower movement at day 12 of adulthood relative to still-active wild-type and oga-1 mutant adults (Suppl. Movie 1). ogt-1 null mutant animals have normal developmental timing (43 ± 3.9 hrs vs. 41 ± 0.7 hrs for ogt-1 and wild type, respectively, p = not significant, >0.5), and generate fertile progeny numbers similar to wild-type animals (263 ± 80 vs. 289 ± 32 for ogt-1 and wild type, respectively, p = ns), suggesting that the reduced movement and earlier death of ogt-1 mutant adults is not due to a general 'sickness'. The oga-1(ok1207) mutant demonstrates a substantial increase in O-GlcNAcmodified protein levels as determined by immunoblot of whole animal lysate, while the ogt-1(ok1474) mutant has only residual staining ( Figure 1B). To determine whether adult lifespan extension in oga-1 mutants arises from excessive O-GlcNAc-modifications, we analyzed the lifespan of the oga-1; ogt-1 double mutant which has a markedly reduced level of O-GlcNAc-modified proteins similar to the ogt-1 mutant (data not shown). The adult lifespan of the oga-1; ogt-1 double mutant is similar to that of wild type (Table 1), confirming that lifespan extension observed in oga-1 mutants is dependent on excessive O-GlcNAc-modified protein levels.

O-GlcNAc cycling enzymes modulate insulinmediated signaling in C. elegans
Inactivation of insulin-mediated signaling increases C. elegans adult lifespan in a DAF-16-dependent manner [21]. We observed that the lifespan extension in the oga-1(ok1207) mutant is also dependent on intact DAF-16 ( Figure 1C). Inactivation of OGT-1 function in daf-2(e1370) mutants dramatically reduces its adult lifespan arguing that O-GlcNAc modification of cellular proteins is a requirement for the daf-2 mutant long lifespan ( Figure 1D). Lifespan extension regulated by insulinmediated signaling in C. elegans is mainly achieved through its downstream effector kinases, such as AGE-1 and SGK-1 ( Figure 6). Inactivation of the effector kinases AGE-1 and SGK-1 also results in significant adult lifespan extension [22,23]. To further test whether O-GlcNAc modification is essential for insulin signaling mediated adult lifespan regulation, we inactivated O-GlcNAc transferase function in both the long-lived age-1(hx546) and sgk-1(ok538) mutants. We observed that adult lifespan extension in age-1 and sgk-1 mutants are completely suppressed to wild-type levels upon combination with the short-lived ogt-1(ok1474) mutant allele (Figure 2A and 2B).
In a conventional genetic analysis, genes that function in parallel pathways with the same functional output generally demonstrate additive interactions. However, combining the long-lived oga-1(ok1207) mutant allele with long-lived insulin signaling pathway mutants, daf-2, age-1, or sgk-1, did not extend lifespan further, suggesting that O-GlcNAc-modified proteins function in the same genetic pathway ( Figure 2C and 2D, Table 1). To further investigate where O-GlcNAc cycling impinges on the insulin signaling pathway, we analyzed both akt-1 and pdk-1 gain-of-function (gf) mutant alleles that constitutively activate the insulin signaling pathway [24,25]. Both akt-1(mg144)gf and pdk-1(mg142)gf mutant lifespans are shorter than that of wild type ( Table 1). The combination of akt-1(mg144)gf or pdk-1(mg142)gf mutant alleles with the long-lived oga-1(ok1207) allele failed to suppress the lifespan extension associated with the oga-1 mutant (Table 1). This data suggests that O-GlcNAc-modified proteins act downstream of the insulin pathway effector kinases to regulate insulin signaling outcomes. However both mg144 and mg142 alleles also do not suppress the extended lifespan of the age-1(hx546) mutant [24]; and thus this conclusion is not definitive.
In C. elegans, the insulin signaling pathway regulates numerous biological functions through DAF-16 by restricting its nuclear localization including, dauer entry, longevity, stress resistance, fertility, and reproductive development [12]. We sought to determine whether perwww.impactaging.com turbing the O-GlcNAc cycling enzymes OGT-1 or OGA-1 would affect DAF-16 subcellular localization. Inactivation of the insulin signaling pathway effector kinase AGE-1 (by RNAi) produced virtually complete nuclear localization of DAF-16 ( Figure 3). Interestingly, inactivation of either OGA-1 or OGT-1 produced a similar but modest increase in DAF-16 nuclear localization ( Figure 3). The observation that inactivating either OGA-1 or OGT-1 produces similar effects on DAF-16 nuclear localization but opposite effects on insulin pathway outcomes, suggests that O-GlcNAc cycling does not functionally modulate the insulin pathway at the level of DAF-16 nuclear localization.

Perturbation in O-GlcNAc cycling affects only a subset of insulin-mediated functions
Lifespan extension in insulin pathway mutants is correlated with oxidative and thermal stress resistance as well as changes in animal development [26,27]. We observed that the long-lived oga-1 mutant is resistant to oxidative stress in a DAF-16-dependent manner similar to that observed with the daf-2 mutant ( Figure 4A). Coinactivation of OGT-1 and DAF-2 in daf-2(e1370); ogt-1(ok1474) double mutant animals significantly reduced the oxidative stress resistance associated with the daf-2 mutant ( Figure 4A). This data implies that protein O-GlcNAc modification contributes to the oxidative stress The adult lifespan extension associated with the oga-1(ok1207) mutant is not synergistic or additive with mutations of the long-lived insulin signaling pathway kinase age-1 (C) or sgk-1 (D). Lifespan curves (A-D) are based on pooled data from multiple replicates as described in Table 1. Control curves are re-plotted to facilitate comparisons in panels A-D.
www.impactaging.com tolerance that is prominent in insulin pathway mutants. In contrast to oxidative stress resistance, the long-lived oga-1 mutant differs from long-lived daf-2 mutants in that the oga-1 mutant does not exhibit increased thermal stress tolerance, delayed developmental timing defects, or reduced fecundity, which are prominent in daf-2 mutants ( Figure 4B, 4C, and 4D, respectively) [16,26,28,29]. Oxidative stress resistance in the insulin pathway mutants arises from the DAF-16mediated expression of several genes including super-oxide dismutase SOD-3, while the increased thermotolerance is correlated with the induction of several heat-shock proteins, such as HSP-16 [27,30,31]. We observed that the inactivation of OGA-1 increased the expression of a sod-3::GFP reporter but not a hsp-16.2::GFP reporter ( Figure 5A and 5B, respectively). Our data indicates that O-GlcNAc modification of proteins is critical for only a subset of the insulin pathway functional outcomes that are downstream of DAF-16. www.impactaging.com

Identification of putative O-GlcNAc target proteins through affinity purification
In an effort to identify the proteins that are modified by O-GlcNAc in C. elegans, we affinity purified O-GlcNAcmodified proteins from oga-1(ok1207) and ogt-1(ok1474) mutant adults, and then identified the affinity purified proteins by liquid chromatography coupled to tandem mass spectrometry (LC-MS/MS). Non-specific proteins would be expected to be present in both purifications, while O-GlcNAc-modified proteins (or proteins tightly-associated with O-GlcNAc-modified proteins) would be specific for the oga-1 mutant lysate. Proteins that were identified from affinity purifications of both oga-1 and ogt-1 mutant lysates were censored as non-specific. Using this approach, we identified 13 distinct proteins that were specific for the O-GlcNAc affinity purification from oga-1 mutant lysate ( Table 2). The 13 proteins encompass a total of 21 proteins when including paralogs with identical or nearly identical protein sequences that could not be distinguished by mass spectrometry. The majority of the putative O-GlcNAc target proteins are regulated by the insulin signaling pathway and a substantial percentage functionally modulate lifespan or oxidative stress (Table 2 and discussion below), suggesting that OGT-1 and OGA-1 may control a subset of insulin signaling functional outcomes by directly targeting the O-GlcNAc modification of proteins that act downstream of the insulin signal transduction pathway.

DISCUSSION
In C. elegans, O-GlcNAc cycling enzyme mutants are viable and therefore provide a useful model to elucidate the role of O-GlcNAc protein modifications, which have been implicated in numerous cellular processes. In our current study we demonstrated that a loss of O-GlcNAc transferase (OGT-1) activity reduces adult lifespan in the nematode C. elegans. Conversely, inactivation of O-GlcNAcase (OGA-1) leads to a moderate increase in adult lifespan. While the accumulation of excessive levels of O-GlcNAc-modified proteins in oga-1 mutant animals is associated with lifespan extension, it is not associated with obvious abnormalities in reproductive development, as oga-1 mutants have an overtly wild-type appearance, develop normally into adults, and generate normal numbers of fertile offspring. Similarly, lifespan reduction in ogt-1 null mutants does not appear to be due to a general 'sickness' of the mutant animals, as ogt-1 mutants also develop normally into fertile adults, and generate normal numbers of offspring ( Figure 4).
Love et al. recently reported similar results that OGT-1 inactivation shortens both wild-type and daf-2 mutant lifespans [32]. However, their findings differ from ours in that they reported that OGA-1 inactivation did not significantly affect lifespan in a wild-type background but slightly extended lifespan in a daf-2 mutant background [32]. We observed that oga-1 mutant lifespan was significantly longer than that of wild type (three independent experiments showed significant differences; p<0.0001, n=66-89 for each experiment); and double mutant analysis of oga-1 with each of three separate components of the insulin signal transduction pathway, daf-2, age-1, and sgk-1, did not show additional lifespan increases in our experiments. www.impactaging.com The failure to observe additive lifespan extension when combining oga-1 and insulin pathway mutations is consistent with O-GlcNAc-modified proteins working in the same genetic pathway as insulin signaling. In C. elegans, insulin signaling regulates gene expression by restricting DAF-16, whose activity is essential for the extended lifespan that is observed in insulin pathway mutants [11,12]. We found that DAF-16 is required for the lifespan extension in oga-1 mutant animals, which is also consistent with O-GlcNAc cycling functioning in the same genetic pathway as insulin signaling.
The observation that elevated O-GlcNAc levels (in the oga-1 mutant) mimics only a subset of insulin pathway mutant phenotypes (lifespan extension and resistance to oxidative stress) but not others (delayed developmental timing, reduced fecundity, and increased thermotolerance) suggests that O-GlcNAc cycling only alters a subset of functional outcomes under the control of insulin-mediated signaling.
The insulin signaling pathway regulates DAF-16 nuclear localization. If OGA-1 and OGT-1 impacted the insulin pathway upstream of DAF-16, it would be expected that oga-1 and ogt-1  -1 and ogt-1 mutants generate normal numbers of offspring while the daf-2 mutant has reduced fecundity at 20°C. The reduced fecundity in the daf-2 mutant is not rescued by loss of O-GlcNAc modification in the daf-2; ogt-1 double mutant. Asterisks above bars denote statistically significant differences from wild type (* p < 0.05 and ** p < 0.01); while asterisks above solid lines connecting two genotypes denote statistically significant differences between those genotypes. www.impactaging.com mutants would have opposite effects on DAF-16 nuclear localization. However, we observed that both mutants have the same modest enhancement of DAF-16 nuclear localization, suggesting that O-GlcNAc cycling modulates insulin signaling outcomes downstream of DAF-16.
We performed an affinity purification of O-GlcNAcmodified proteins that identified 13 proteins ( Table 2). The mammalian counterparts of 6 of the 13 proteins are known to be O-GlcNAc modified, suggesting that these are conserved targets of O-GlcNAc transferase (Table 2).
Strikingly, the expression of 10 of the 13 putative O-GlcNAc target proteins is controlled by the insulin signaling pathway (Table 2). More significantly, 4 of the 13 putative O-GlcNAc target proteins have been reported to modulate lifespan in C. elegans. Inactivation of either of the F0F1-type ATP synthases ATP-2 or H28O16.1 is observed to increase adult lifespan and increase the expression of a sod-3::GFP reporter [33]. Inactivation of the glyceraldehyde-3-phosphate dehydrogenase GPD-2 further extends the lifespan of daf-2 mutants [34]; while inactivation of the malate dehydrogenase MDH-1  www.impactaging.com shortens daf-2 mutant lifespan [35]. Additionally, inactivation of GPD-2 and GPD-3 allows the daf-2 (e1370) mutant to escape dauer entry, thereby demonstrating a critical role for GPDs in regulating one of the functional outcomes under the control of the insulin signaling pathway [36]. Our work therefore demonstrates that downstream targets of the insulin pathway are O-GlcNAc modified. This provides the possibility that O-GlcNAc cycling enzymes act downstream of DAF-16 to directly functionally modulate specific insulin-regulated processes.
Increased O-GlcNAc levels inhibit insulin stimulated activation of AKT, and disrupt insulinmediated glucose transport in adipocytes demonstrating a direct involvement of O-GlcNAc cycling in the attenua-tion of insulin-mediated signaling functional outcomes. A reduced ability of insulin to activate glucose transport into cells, i.e., insulin resistance, is marked by high glucose and high insulin levels in circulating blood, and aberrant insulin-regulated gene functions [10]. Abnormalities in carbohydrate and fat metabolism are characteristic of insulin-resistant tissues, and are implicated in elevated plasma sugar and fatty acid levels in patients with type II diabetes [38]. In C. elegans, similar disarray in carbohydrate and fat storage is observed in insulin pathway mutant animals [9]. Previous studies also demonstrated abnormal carbohydrate and fat storage upon disruption of O-GlcNAc cycling [18].

METHODS
Lifespan analysis. All lifespan assays were performed using 85-125 animals for each genotype per assay at 20°C on NGM plates seeded with OP50 bacteria following established procedures [8]. Adults were transferred to fresh plates every alternate day during the fertile period, and later when necessary. Adults were counted as dead when they failed to respond to repeated head and tail prodding. Multiple assays were pooled for final analysis.
Statistical analyses. Survival curves were analyzed by the Log-rank (Mantel-Cox) and Wilcoxon tests using GraphPad Prism (version 5.0) to determine the significance between mutant and wild-type controls as indicated in Table 1. In other experiments, statistical significance was determined with the unpaired two-tailed Student's t-test.
Stress tolerance assay. Oxidative stress resistance assays were performed in multi-well plates by immersing adults in M9 media containing 100 mM paraquat (N,N'dimethyl-4,4'-bipyridinium dichloride) using 10-20 animals for each genotype per assay at 20°C [39,40]. A complete absence of swimming movement was scored as death, and also reconfirmed by transferred animals not moving upon prodding on agar plates. Death was scored every 3, 6, 9, and 12 hrs. Experimental data from multiple assays were pooled together for final analysis. Thermal resistance assays were performed on NGM plates pre-heated at 35°C using at least 50 animals for each genotype per assay [29,41]. Adults were incubated at 35°C, and were scored every hour by response to touch until all animals were dead. Experimental data from multiple assays were pooled together for final analysis.
Developmental timing and fecundity assays. Postembryonic developmental timing was assayed by collecting comma-stage embryos from animals maintained at 20°C. The exact time of hatching was recorded, and then animals were transferred onto separate OP50-seeded NGM plates. After the third molt, animals were inspected more frequently (every 15 mins) to record the exact time of the fourth molt. The assay was repeated several times using at least 10 animals for each genotype per assay, and multiple experimental data were pooled together for final analysis. For fecundity analysis, individual larval stage animals were placed on OP50seeded NGM plates at 20°C. Offspring were counted by moving them onto a fresh plate one-by-one. The assay was repeated multiple times with at least 10 animals per genotype in each assay, and data were pooled together for final analysis.
Western immunoblotting. Adult animals were flash frozen in liquid nitrogen in lysis buffer (50mM HEPES pH 7.8 and 300mM NaCl). Frozen worms were crushed in a pestle pre-chilled with liquid nitrogen. The frozen ground worms were re-suspended in cold lysis buffer (with DNaseI and protease inhibitor cocktail from Roche) and subjected to sonication for 6-8 seconds followed by centrifugation at 16,000 rpm for 1 hour at 4°C. The middle portion (not including the pellet or floating lipids) www.impactaging.com of each tube was collected for analysis. An equal amount of whole animal lysate was separated on a 10% NuPAGE Bis-Tris gel (Invitrogen) followed by transfer onto a PVDF membrane (Millipore). The membrane was probed with anti-O-GlcNAc antibody from Affinity Bioreagents (1:2000 in 3% BSA dissolved in 1XTBS + 0.05% Tween20). The immunoblot was visualized using a chemiluminescent substrate (Super Signal West Pico from Pierce). The same membrane was re-probed with anti-α-tubulin antibody (Sigma) to normalize protein loading.
All RNAi experiments were performed at 20°C unless otherwise mentioned using E. coli strain HT115 transformed with pPD129.36 clones from an RNAi library provided by Julie Ahringer for gene inactivation experiments, as previously described [42,43].
GFP reporter assays. Experimental animals for analysis were grown on feeding RNAi for more than one generation. All animals were imaged at single shutter setting, and equal nuclear and cytoplasmic areas were chosen for light intensity measurement using OpenLab 5.5 software. The average of three mean light intensity measurement values per cell (in arbitrary units) for both nuclear and cytoplasmic regions were recorded for multiple cells in each treated and untreated animals. The assay was repeated several times with at least 10 animals per genotype in each assay at 20°C. Data from multiple experiments were pooled for final analysis. Both sod-3::gfp and hsp-16-2::gfp reporters were assayed in 10day-old adults.
Affinity purification and identification of O-GlcNAcmodified proteins. Adult animals were collected and flash frozen in liquid nitrogen in 50mM HEPES pH 7.8 and 50mM NaCl buffer. Frozen worms were crushed with a mortar and pestle pre-chilled with liquid nitrogen. The frozen ground worms were re-suspended in the above buffer (with DNaseI and protease inhibitor cocktail from Roche), and disrupted by sonication followed by centrifugation at 16,000 rpm for 1 hour at 4°C. Clear supernatant was collected and O-GlcNAcmodified proteins were identified as previously described [44].