Rifampicin reduces advanced glycation end products and activates DAF-16 to increase lifespan in Caenorhabditis elegans

Advanced glycation end products (AGEs) are formed when glucose reacts nonenzymatically with proteins; these modifications are implicated in aging and pathogenesis of many age-related diseases including type II diabetes, atherosclerosis, and neurodegenerative disorders. Thus, pharmaceutical interventions that can reduce AGEs may delay age-onset diseases and extend lifespan. Using LC-MSE, we show that rifampicin (RIF) reduces glycation of important cellular proteins in vivo and consequently increases lifespan in Caenorhabditis elegans by up to 60%. RIF analog rifamycin SV (RSV) possesses similar properties, while rifaximin (RMN) lacks antiglycation activity and therefore fails to affect lifespan positively. The efficacy of RIF and RSV as potent antiglycating agents may be attributed to the presence of a p-dihydroxyl moiety that can potentially undergo spontaneous oxidation to yield highly reactive p-quinone structures, a feature absent in RMN. We also show that supplementing rifampicin late in adulthood is sufficient to increase lifespan. For its effect on longevity, rifampicin requires DAF-18 (nematode PTEN) as well as JNK-1 and activates DAF-16, the FOXO homolog. Interestingly, the drug treatment modulates transcription of a different subset of DAF-16 target genes, those not controlled by the conserved Insulin-IGF-1-like signaling pathway. RIF failed to increase the lifespan of daf-16 null mutant despite reducing glycation, showing thereby that DAF-16 may not directly affect AGE formation. Together, our data suggest that the dual ability to reduce glycation in vivo and activate prolongevity processes through DAF-16 makes RIF and RSV effective lifespan-extending interventions.


Introduction
Aging is an inescapable process in all living beings. Although most animals exhibit signs of age-related decline in body functions, some interestingly show negligible visible signs of senescence (e.g., some tortoises, sea anemones, lobsters). One of the hallmarks of aging is the accumulation of altered proteins (Hipkiss, 2006). With advancing age, cellular homeostatic processes that restrict damage caused by altered proteins decline in their efficiency, leading to various age-related pathologies (Thornalley, 2008). Importantly, modern lifestyle also contributes immensely to this process of disease development. One such posttranslational alteration of protein, which is at the center of many age-related diseases including type II diabetes, atherosclerosis, renal disorders, Alzheimer's disease, and rheumatoid arthritis, is the formation of advanced glycation end products (AGEs) (Thornalley, 2008;Luevano-Contreras & Chapman-Novakofski). AGEs are a complex and heterogeneous group of molecules formed via a series of parallel and sequential nonenzymatic reactions involving glucose or glucose-derived products and amino groups of proteins, lipids, or DNA (Singh et al., 2001;Ulrich & Cerami, 2001;Ahmed, 2005;Luevano-Contreras & Chapman-Novakofski, 2010). The carbonyl group of a sugar reacts with a protein to form an unstable aldimine intermediate called Schiff's base (Fig. S1A). The Schiff's base may undergo an Amadori rearrangement to form a stable 1-amino-1-deoxyfructose derivative with a stable ketoamine linkage, commonly known as Amadori product. The formation of Amadori product marks the reversible phase of glycation that is followed by irreversible oxidation, dehydration, condensation, fragmentation, or cyclization leading to the formation of AGEs such as pentosidine, N-ecarboxymethyllysine (CML), and pyraline (Singh et al., 2001;Ulrich & Cerami, 2001;Ahmed, 2005). Additionally, auto-oxidation of glucose by the Wolff pathway or of the Schiff's base by the Namiki pathway generates highly reactive dicarbonyl or oxoaldehyde molecules such as glyoxal, methylglyoxal, or 3-deoxyglucosone (3-DG). The dicarbonyl intermediates in turn give rise to AGEs such as pentosidine, CML as well as glyoxal-lysine dimer (GOLD), or methylglyoxal-lysine dimer (MOLD) through additional cross-linking (Khalifah et al., 1999;Singh et al., 2001;Ulrich & Cerami, 2001;Ahmed, 2005).
Methylglyoxal is mainly formed during the process of glycolysis; either enzymatically by methylglyoxal synthase, cytochrome 2E1, and semicarbazide-sensitive amine oxidase or nonenzymatically by spontaneous degradation of dihydroxyacetone phosphate and glyceraldehyde-3phosphate. Under euglycemic conditions, the glyoxalase system breaks down methylglyoxal and converts it to D-lactate Xue et al., 2011). However, this cellular detoxification machinery loses efficacy with increasing age. As a result, AGEs build up in different tissues as an organism gets older (Semba et al., 2010). AGE modifications often lead to abnormal protein function, oxidative stress, and inflammation; thus, they are closely associated with deteriorating cellular functions during aging and in pathologies of several age-related diseases (Luevano-Contreras & Chapman-Novakofski, 2010;Xue et al., 2011). In C. elegans, a powerful genetic model that has emerged as an effective system to study aging, glyoxalase I activity is markedly reduced with age that leads to increased oxidative stress. Interestingly, overexpression of the gene can decrease methylglyoxal-induced mitochondrial protein modifications and increase lifespan (Morcos et al., 2008). Thus, pharmaceutical interventions that can suppress AGE formation may be an effective way to increase lifespan and health span (Semba et al., 2010). In this study, we show that rifampicin (RIF), a potent glycation inhibitor (Golegaonkar et al., 2010), dramatically increases lifespan (up to 60%) as well as improve health of C. elegans by preventing AGE modifications of important cellular proteins. This is by far one of the largest increases in lifespan obtained using a pharmaceutical reagent. Apart from possessing antiglycating activity, RIF activates the FOXO transcription factor DAF-16, possibly through JNK pathway, to modulate transcription of a unique set of target genes, those that are not controlled by the Insulin-IGF-1 signaling pathway. Rifaximin (RMN), a rifampicin analog that lacks antiglycating activity in vitro and in vivo, but possesses the ability to activate DAF-16 fails to extend lifespan. Together, our study shows that the dual ability of RIF to reduce glycation as well as activate DAF-16 makes it a potent lifespan-extending intervention.

Rifampicin and rifamycin SV dramatically increase lifespan
Repositioning of known drugs for novel applications is an emerging strategy in the area of drug discovery. In our previous study using an in vitro mass spectrometry-based screen, the antibiotic rifampicin (RIF) was identified as a more potent glycation inhibitor than aminoguanidine (AMG), a well-known antiglycating agent (Golegaonkar et al., 2010). The higher efficacy of rifampicin as a glycation inhibitor was also evidenced in case of AGE-induced corneal stiffening (Brummer et al., 2011). Similar to RIF (apparent IC 50 0.114 mM), its analog rifamycin SV (RSV) also has similar glycation inhibition efficiency (apparent IC 50 0.216 mM) (Figs 1A,B, S1B). On the other hand, another analog, rifaximin (RMN), was found to possess low in vitro activity as a glycation inhibitor (Figs 1C, S1B). The differential activity of RIF, RSV, and RMN could be attributed to their chemical structures. All the three compounds possess macrocyclic lactam structure containing naphthyl moiety fused with a cyclopentanone ring as a common feature. However, only RIF and RSV possess the para-dihydroxynaphthyl moiety (highlighted in red in Fig. 1A,B), while RMN has a para-aminophenyl moiety fused in a ring system (highlighted in blue in Fig. 1C). As AGE accumulation with increasing age is linked to diseases and reduced longevity (Luevano-Contreras & Chapman-Novakofski, 2010), we asked whether glycation inhibitors RIF and RSV have any positive effect on lifespan. Using a modified C. elegans liquid lifespan assay with live E. coli OP50 (see Experimental procedures), we observed that RIF significantly increased lifespan of wild-type worms (Fig. S2A, details in Table S3). As E. coli OP50, which is used to feed C. elegans in the laboratory, can act as an opportunistic pathogen in older worms, antibiotics such as RIF may increase lifespan by simply killing the bacteria. To rule out this possibility, we repeated the above experiment using heat-killed bacteria. We found that RIF was still able to increase lifespan in a concentration-dependent manner ( Fig. 2A, Table S3) ruling out two scenarios: (i) Its effect on lifespan was due to bactericidal properties and (ii) bacteria metabolizing the drug to produce an unrelated product that indirectly affected lifespan. Additionally, antibiotic kanamycin did not have similar effect on lifespan (Fig. S2B). Therefore, we continued all our experiments using heat-killed bacteria and used 50 lM RIF. Next, we analyzed RSV and RMN for their effect on lifespan. While RSV increased lifespan significantly like RIF (Mean Lifespan AE Standard Error Mean on 0 lM RIF is 24.89 AE 0.28, while on 50 lM RIF is 39.57 AE 0.29 and 50 lM RSV is 30.09 AE 0.61, P ≤ 0.0001 in all cases; Fig. 2B), RMN treatment had no positive effect on lifespan (MLS AE SEM on 50 lM RMN is 21.48 AE 0.46 P ≤ 0.0001) (Fig. 2B, Table S3). In fact, RMN significantly reduced lifespan. Finally, treatment with RIF also enhanced two parameters that are representative of better health in worms, that is, decreased accumulation of autofluorescence in the gut (Fig. S2C) and delayed onset of proteotoxicity (Fig. S2D). These experiments suggested that the increased lifespan and health on RIF and RSV treatment may be due to their potent antiglycation activity in vivo.

Rifampicin reduces AGE accumulation in vivo
Increased AGE formation associated with progressing age is linked to many diseases including atherosclerosis and neurodegenerative disorders (Nedic et al., 2013). Using a monoclonal antibody that detects carboxymethyl lysine (CML), we performed Western blot analysis with proteins extracted from 3-, 6-, or 12-day-old adult wild-type worms. We found that accumulation of CML increased with progressing age in worms (Fig. 2C). As expected from in vitro glycation assays and lifespan analysis, treatment with RIF decreased CML formation considerably in 6-and 12-day-old worms ( Fig. 2D). However, RMN treatment had much lesser effect on CML formation on different days of adulthood, compared to RIF (Fig. 2D). These experiments show that AGE formation increases with age in C. elegans and RIF can act as a glycation inhibitor in vitro and in vivo to affect lifespan positively. Importantly, the property of RIF to increase lifespan can be linked to its antiglycating property. Methyl glyoxal (MG) is a highly reactive dicarbonyl that reacts with epsilon amino groups of lysine and sulfhydryl groups of proteins leading to the formation of AGEs. The levels of glyoxalase decline with age leading to increased AGE modification of proteins; AGE levels, including CML, also rise when glyoxalase is knocked down (Morcos et al., 2008). The glod-4 mutant lived significantly shorter than the WT (Fig. S3A), possibly due to the increased accumulation of AGEs. Interestingly, RIF was able to decrease CML modification of proteins in glod-4(gk189) ( Next, we used LC-MS E , a data-independent acquisition (DIA) strategy to identify in vivo AGE-modified proteins. We used the glod-4(gk189) strain as we thought that it may have relatively higher total AGE levels that will be easier to detect. LC-MS E is a unique approach wherein all eluted peptides are fragmented and the fragment ions are time-aligned with the retention time of the peptides (Silva et al., 2006). This method allows analysis of low abundance posttranslationally modified peptides (Blackburn & Goshe, 2009;Xie et al., 2009). A total of 49 AGE-modified proteins were detected in DMSO-and RIF-treated samples. In the DMSOtreated samples, 69 peptides carrying 89 AGE modifications were detected. However, in the RIF-treated samples, only 32 peptides were observed to carry 42 AGE modifications (Table 1)   changes in protein expression on RIF treatment (Table S4). We found that the beta subunit of the mitochondrial ATP synthase possessed the most number of modifications that were dramatically reduced on RIF treatment. VIT-2 as well as VIT-6, a vitellogenin that is reported to be heavily carbonylated in worms (Nakamura et al., 1999), also had many modifications that showed reduction on RIF treatment. Interestingly, modifications on several mitochondrial proteins including ATP synthase subunit beta, aconitate hydratase, isoform D of ATP synthase subunit alpha, dihydrolipoyl dehydrogenase, methylmalonate semialdehyde dehydrogenase, and chaperonin Hsp 60 as well as two fatty acid and retinol binding proteins also decreased on RIF treatment. Marked decrease in glycation was also reported in mitochondrial proteins when glyoxalase I was overexpressed in C. elegans (Morcos et al., 2008). Apart from metabolic proteins, RIF treatment resulted in a decrease in glycation on muscle-specific proteins such as paramyosin, tropomyosin, and myosin heavy and light chain as well as structural proteins such as actin and tubulin. Modifications on these important proteins may disrupt normal cellular functions leading to aging. RIF may help in retaining proper functioning of these proteins and thereby affect lifespan positively.

Rifampicin activates FOXO homolog, DAF-16
The Insulin-IGF-1 signaling (IIS) pathway in C. elegans and other organisms is one of the most important regulators of aging (Kenyon, 2010). Mutations in the components of IIS increase lifespan across the animal kingdom. Interestingly, genetic experiments in worms have revealed distinct temporal requirements for the longevity pathways to affect lifespan (Dillin et al., 2002). While IIS pathway knockdown starting on the first day of adulthood or even later is sufficient to extend lifespan, mitochondrial genes have to be knocked down during larval stages for increased longevity (Dillin et al., 2002). Similar to IIS pathway knockdown, RIF supplementation starting at day 1, 5, or even at day 9 of adulthood increased lifespan significantly (Figs 2B and 3A,B, Table  S3). This highlights the potency of RIF to act as a single-dose late-life longevity-extending intervention. As RIF has a temporal requirement similar to the IIS pathway mutants for regulating longevity, we studied its relation to this pathway. Mutations in the Insulin-IGF-1-like receptor, daf-2, reduces signaling through this pathway and increases lifespan (Kenyon, 2010). Mutation in the lipid phosphatase DAF-18 (homolog of mammalian PTEN) activates the IIS and suppresses the long lifespan of daf-2 mutants (Ogg & Ruvkun, 1998;Mihaylova et al., 1999). We observed that RIF failed to increase the lifespan in daf-18(e1375) mutant (MLS AE SEM on 0 lM RIF is 18.60 AE 0.26, while on 50 lM RIF is 19.84 AE 0.21, P = 0.0739; Fig. 3C, Table S3), suggesting that DAF-18 is required for RIF-induced lifespan extension. The increased lifespan of a daf-2 mutant is completely dependent on the FOXO transcription factor DAF-16 (Lin et al., 1997;Ogg et al., 1997). Further, the RIF-mediated lifespan extension also requires DAF-16 as the drug failed to increase lifespan in daf-16(mgdf50), a null mutant (MLS AE SEM on 0 lM RIF is 17.80 AE 0.25, while on 50 lM RIF is 17.11 AE 0.25, P = 0.122) (Fig. 3D, Table S3). These experiments suggested that RIF treatment may genetically interact with the IIS pathway components and activate DAF-16.
DAF-16 shuttles between the nucleus and cytoplasm; it translocates into the nucleus in response to low Insulin signaling or under stress. Activation of DAF-16 by RIF was supported by the enhancement of DAF-16::GFP protein (Henderson & Johnson, 2001) translocation into the nucleus in the presence of the drug (Fig. 4A) and consequently,  No change in glycation.
upregulation of sod-3, its direct target (Oh et al., 2006) (Fig. 4B). Similar to RIF, RSV was also able to translocate DAF-16 into the nucleus (Fig. 4A). Together, activation of DAF-16 is an additional mechanism of RIF-mediated lifespan extension, apart from its antiglycating property. Incidentally, RMN that lacked the glycation inhibition property also affected DAF-16 translocation (Fig. 4A). Thus, our data also indicate that a combination of potent antiglycating activity and positive regulation of DAF-16 is required for extending lifespan as observed on RIF treatment. Simply possessing DAF-16-activating property may not be sufficient to increase lifespan, as observed for RMN. There existed a possibility that activation of DAF-16 leads to the upregulation of an AGE detoxification system, resulting in increased AGE disposal and a consequential increase in lifespan. To evaluate this possibility, we grew daf-16(mgdf50) in the presence of RIF and measured the CML levels on days 6 and 9 of adulthood. We found that in daf-16(mgdf50), RIF was still able to decrease CML formation (Fig. 4C). Although we cannot rule out the possibility that other AGEs are unaffected or that small suppression of CML can have a dramatic effect on lifespan, our data suggest that DAF-16 may have other functions in lifespan regulation downstream of RIF, apart from detoxifying and disposing AGE-modified proteins.

Rifampicin may activate DAF-16 through JNK-1
Next, we asked whether RIF treatment reduces signals through the wellconserved Ser-Thr kinases, PDK-1 and AKT-1, that are downstream to the regulation by DAF-18, but upstream of DAF-16. Activation of these kinases leads to phosphorylation and exclusion of DAF-16 from the nucleus, while mutations in these kinase genes lead to prolonged lifespan in a DAF-16-dependent manner (Paradis & Ruvkun, 1998;Paradis et al., 1999). We found that in the activated alleles pdk-1 (mg142) (MLS AE SEM on 0 lM RIF on 29.18 AE 0.40, while on 50 lM RIF is 35.99 AE 0.47, P ≤ 0.0001, Fig. S4) and akt-1(mg144) (MLS AE SEM on 0 lM RIF is 18.55 AE 0.26, while on 50 lM RIF is 36.96 AE 0.70, P ≤ 0.0001, Fig. 5A), RIF could still increase lifespan, indicating that it may utilize a separate signaling pathway, bypassing PDK-1. As RIF treatment may activate an oxidative stress response pathway as in mammals (Sodhi et al., 1997), we speculated a role of JNK-1, a known DAF-16 and FOXO activator (Oh et al., 2005). In line with this supposition, RIF failed to considerably increase the lifespan of jnk-1 (gk7) (MLS AE SEM on 0 lM RIF is 39.52 AE 0.61, while on 50 lM RIF is 39.14 AE 0.51, P = 0.6307; Fig. 5B). In fact, the upstream kinase for JNK-1, JKK-1 (Oh et al., 2005), is also involved in this process as in jkk-1 (km2), RIF was not able to increase lifespan (MLS AE SEM on 0 lM RIF is 33.38 AE 0.43, while on 50 lM RIF is 33.96 AE 0.47, P = 0.3088; Fig. 5C). Together, these genetic data suggest that RIF requires some specific components of the IIS pathway and may utilize JNK pathway to activate DAF-16 and enhance lifespan. Further experimentation is mandated in dissecting the role of JNK pathway in RIF-mediated lifespan extension.

Rifampicin modulates expression of a different subset of DAF-16 target genes
To further understand the mechanism of RIF action, we performed a comparative microarray analysis which revealed that 102 genes are significantly upregulated, while 104 genes were downregulated by at least two fold (P < 0.05) upon RIF treatment as compared to DMSO ( Fig. 6A; Tables S1 and S2). While most of these genes are uncharacterized/unannotated, GO term analysis revealed that genes involved in cell cycle regulation, proteolysis, and organismal aging were enriched in the upregulated gene list (Fig. 6B). On the other hand, many genes involved in transcriptional regulation were downregulated (Fig. S5A). We also found genes that possibly function in countering AGEs to be

daf-18 (e1375)
Time ( downregulated (e.g., lysozymes, saposin-like proteins, peptidases); products of these genes may not be required due to the low intrinsic AGEs upon RIF treatment. Alternatively, the downregulation of lectins, saposins, and lysozymes may indicate that the worms may not be mounting innate immune response due to the presence of antibiotic in the media. However, two genes that have putative function as ubiquitin ligase and ubiquitin-conjugating enzyme were upregulated possibly to facilitate removal of nonfunctional proteins. Also, Vitellogenin genes (vit-3, vit-4 and vit-5) were found to be downregulated significantly. These proteins accumulate in the intestine following reproductive development and are subjected to oxidative damage leading to toxicity. So, lower levels of these proteins may promote longevity as shown earlier (Murphy et al., 2003;Van Nostrand et al., 2013). Additionally, genes involved in xenobiotic detoxification (e.g., cytochrome P450, UDP-glucuronosyltransferase, P-glycoprotein-related proteins) were upregulated (Table  S1). The latter group supports increased longevity as recently reported (Shore & Ruvkun, 2013) and may also be involved in detoxification of RIF. RIF treatment thus brings about a profound change in the C. elegans transcriptome that may also be responsible for the increased longevity.
A previous microarray study has revealed several hundred genes that are up-(Class 1) or downregulated (Class II) in a daf-2 mutant, dependent on DAF-16 (Murphy et al., 2003). We observed that although a few Class I DAF-16 target genes had increased expression on RIF treatment, some others of the same class were downregulated. Quantitative RT-PCR confirmed this as sod-3 was significantly upregulated in the presence of RIF, while scl-1 and hsp-12.6 were downregulated (Fig. 3B). Although both RIF and a mutation in daf-2 seem to activate DAF-16, there was only 2% overlap with Class I genes and 6% commonality with Class II genes between the two regimes, highlighting a separate activation profile of RIF (Fig. 6C). However, 92-95% of the promoters (5 kb upstream) of genes that are up-or downregulated following RIF treatment had consensus DAF-16 binding (GTAAAC/TAA) as well as DAF-16-associated (CTTATCA) elements (Fig. 6D, Tables S1 and S2) (Murphy et al., 2003), compared to 78% as would be expected by chance (P value is 8.28e-10 by hypergeometric test considering 20447 coding genes in WBcel235) (Kenyon & Murphy, 2006). Using quantitative real-time PCR, we checked the expression of 10 genes following RIF treatment and found seven of them to change in a daf-16-dependent manner (Fig. S5B). Thus, we conclude that RIF may activate DAF-16 differently and modulate a separate subset of target genes compared to the ones regulated by the IIS pathway.

Discussion
Drug repositioning is a relatively new area of pharmacology. It deals with discovering new molecular targets of approved drugs so that they may be used to treat morbidities unrelated to the ones for which the drug was developed. Some of the best-known examples of re-engineered drugs include viagra, colesevelam, requip, and gabapentin. Repositioning is profitable as it cuts cost of research and time of development, making drugs available to patients faster and cheaper. In this direction, our study shows that rifampicin and its analog rifamycin SV, developed initially as antibiotics to treat tuberculosis, are effective interventions in extending health and lifespan. Additionally, lifespan extension was also obtained when the drug treatment was started well into adulthood. It will be interesting now to evaluate their efficacy in higher organisms and determine whether RIF treatment late in adulthood can bestow benefits of health as well as longevity on these animals. Why does RIF and RSV have glycation inhibition properties while RMN does not? A close examination of the chemical structures of RIF, RSV, and RMN revealed that all the three compounds possess macrocyclic lactam and naphthyl moiety fused with a cyclopentanone ring as common features. However, only RIF and RSV possess the paradihydroxy naphthyl moiety (highlighted in red in Fig. 1), while RMN has a para-aminophenol moiety (highlighted in blue in Fig. 1). The p-dihydroxyl moiety present in RIF and RSV could undergo spontaneous oxidation to yield highly reactive p-quinone structures. On the other hand, the p-amino group in RMN cannot form p-quinone structures as it is locked up in the fused imidazole ring.
Formation of p-quinone structures from the p-dihydroxyl moiety could have multiple mechanistic consequences for the observed antiglycation properties of RIF and RSV in vitro. First, the quinones, due to their higher reactivity could compete directly with glucose for the primary amino groups on Insulin in vitro and other proteins in vivo. Benzoquinones and napthaquinones have been shown to inhibit glycation and lipid peroxidation (Jung et al., 2005). Secondly, the p-quinone moiety could also react with other side chain residues on proteins such as cysteine, tryptophan, and tyrosine via a Michael addition reaction, which might decrease the formation of glucosylated Insulin (Schiff's base). Earlier studies using tritium labeling have shown irreversible binding of 3 H-rifampicin-quinone to albumin which was inhibited by co-incubation with ascorbic acid (Bolt & Remmer, 1976), implicating quinole↔quinone redox system. Additionally, it has been reported that rifampicin quinone, but not reduced rifampicin, showed immunosuppressant activity and prolonged graft acceptance (Konrad & Stenberg, 1988). Therefore, the structure-specific inhibitory activity observed with RIF and RSV could be attributed to the presence of highly reactive quinone moieties. We posit that similar mechanisms may partly explain the robust antiglycating effects of RIF and RSV in vivo in C. elegans.
We show that the lifespan increase on RIF treatment requires DAF-18/ PTEN but not AKT-1 and PDK-1. On the other hand, RIF requires a wildtype jnk-1 and jkk-1 allele for its lifespan effects. While the IIS pathway negatively regulates DAF-16, JNK activates it (Oh et al., 2005;Wang et al., 2005). It is possible that the strong mutant allele of daf-18 leads to a highly activated IIS pathway and strong DAF-16 sequestration in the cytoplasm. Under this condition, activating JNK (by RIF) is not sufficient to translocate DAF-16 into the nucleus. On the other hand, activating mutations in PDK-1 or AKT-1 may not be sufficiently strong to sequester DAF-16; JNK activation can override this inhibition. It is also possible that DAF-18 and JNK-1 interact independent of PDK-1 or its downstream kinases, by some mechanism as indicated in other systems (Vivanco et al., 2007;Hubner et al., 2012). As JNK-1 is a stress-associated kinase (Matsukawa et al., 2004), RIF treatment may activate a stress response through this cascade, involving its upstream kinase JKK-1, leading to the activation of DAF-16 that results in transcription of genes required for increased longevity. However, this action of RIF alone is not sufficient to increase lifespan as RMN having similar DAF-16 activating property failed to positively affect longevity in wild-type worms. In fact, previous studies have shown that simply activating DAF-16 is not sufficient to increase lifespan (Lin et al., 2001). Together, these observations underscores the importance of our finding that a combination of two complementary properties, viz. potent glycation inhibition and DAF-16 activation, are required for affecting longevity in a positive manner, as shown by RIF and RSV (Fig. 6E). This dual property of RIF is reflected in its mechanism of action that is both at the transcriptional as well as posttranscriptional level. At the posttranslational level, RIF treatment reduces AGE formation on major cellular proteins that control metabolism and support musculature. At the transcriptional level, it activates genes involved in aging, proteolysis, cell cycle, and cellular detoxification. Interestingly, in both the cases, RIF targets the vitellogenins that are required for growth and development of progeny. However, the vitellogenin proteins appear to be toxic at old age in worms as they undergo AGE modifications, and knockdown of vit-2 as well as vit-5 by RNAi increases lifespan (Murphy et al., 2003). Vitellogenins are required during the self-fertile reproductive phase of hermaphrodites and later for reproduction by cross-fertilization. In the absence of cross-fertilization in older worms as well as in aging Drosophila, vitellogenins accumulate in the body cavity and are subjected to oxidative damage. In later life, they become one of the most prevalent proteins in aggregates (Nakamura et al., 1999;David et al., 2010;McGee et al., 2011). AGE modifications on these proteins may accelerate aggregation and toxicity. Thus, decreased vitellogenin protein accumulation and reduced AGE modification may contribute to extended longevity imparted by RIF. Additionally, we found that several genes involved in xenobiotic detoxification are upregulated on RIF treatment. These genes have recently been implicated in lifespan regulation (Shore & Ruvkun, 2013) and may constitute one of the arms through which RIF has a positive effect on longevity. From the proteomics data, we found that proteins important in muscle function have several AGE modifications that are reduced on RIF treatment. These proteins include myosin, tropomyosin, paramyosin as well as actin and tubulin. By reducing glycation on major muscle-specific proteins, RIF may positively affect muscle function as evidenced in our paralysis assays. Apart from these proteins, enzymes involved in important cellular processes including glycolysis (Glyceraldehyde 3-phosphate dehydrogenase), Kreb's cycle (citrate synthase, aconitase), and amino acid metabolism (arginine kinase, methylmalonate semialdehyde dehydrogenase) develop lower number of AGE modifications in the presence of RIF. This may lead to efficient metabolism required for extended lifespan as observed on RIF treatment. Thus, the multidimensional effects of RIF result in one of the most dramatic effects of a drug on longevity in C. elegans.
In conclusion, our study using the most suitable model organism for aging research shows how structure-dependent differential activity of three FDA-approved drugs may help in repositioning them to potentially treat age-associated pathologies and increase longevity.

Experimental procedures
Detailed experimental procedures are provided as Data S1. Worms were bleached, and eggs were placed on OP50-Amp seeded NGM-Amp plates. They were grown at 20°C till young adult stage, rinsed off, and washed 4 times with M9 buffer. Thereafter, worms were resuspended in S-complete buffer and diluted further to contain 10-15 worms per 10 lL. For each experiment, six-eight wells were used for the same treatment and controls were set up in each plate. Thus, the total number of worms was~60 at the beginning of the experiments. Bacteria for liquid culture were prepared using 14 h/37°C-grown OP50-Amp that was resuspended in S-complete to 100 mg mL À1 and heat-killed at 100°C for 45 min. A master mix was then prepared such that each well of a noncoated sterile 96-well culture plate contained 80 lL S-complete, 10 lL resuspended worms, 9.9 lL heat-killed OP50-Amp, 1.6 lL carbenicillin (50 mg mL À1 stock), and 0.6 lL amphotericin B (250 lg mL À1 stock). Later, to each well, 30 lL of 0.6 mM 5-Fluoro-2 0 -deoxyuridine (FuDr; Sigma-Aldrich, St. Louis, MO, USA) was added and, the next day, 10 lL of a solution in S-complete containing the drugs (dissolved in DMSO, stored at 4°C) or only DMSO was added. The worms were fed again on the 5th day with 9 lL heat-killed OP50-Amp/well. Every other day, the worms in the wells were exposed to light from the diascopic base of a stereozoom microscope and the plates gently agitated by tapping; based on their movement, the worms were scored as dead or alive. Dead worms were not extracted from the wells; a count of total number of live worms in each well was kept to plot the lifespan. Lifespans were repeated multiple times, and consolidated data are shown (details in Table S3). In vivo AGE modification of proteins was determined by LC-MS E , a data-independent acquisition. For proteotoxicity assay, CL2006 (Link, 1995) was grown as above and worms from a single well were transferred to an OP50-seeded NGM plate for scoring. For protein extraction, in a 250-mL flask, 17.78 mL S-complete, 2.20 mL worm suspension (~5000 worms), 1.9 mL heat-killed OP50-Amp, 33 lL Carbenicillin, 13.2 lL amphotericin B, and 6.6 mL of 0.6 mM FuDr was added. The next day, 2.2 mL of S-complete containing 200 lM RIF (dissolved in DMSO) or only 0.2% DMSO with or without glucose was added and grown for the indicated number of days at 20°C before harvesting. For qRT-PCR analysis, cDNA was synthesized using 5 lg of total RNA and SuperScript III cDNA synthesis kit (Life Technologies, Carlsbad, CA, USA). Microarray analysis was carried using an Affymetrix, USA platform. Data were analyzed using Genespring GX 11.1 with Affymetrix default analysis settings. The microarray data are available at http://www.ncbi.nlm.nih.gov/geo/ query/acc.cgi?token=xjcdjagswaygerq&acc=GSE45292. Statistical analyses for survival were conducted using Mantel-Cox log-rank test with OASIS software available at http://sbi.postech.ac.kr/oasis. P value calculations for other experiments were performed by unpaired t-test using SIGMAPLOT 10.0 (Systat Software, Inc., San Jose, CA, USA).

Supporting Information
Additional Supporting Information may be found in the online version of this article at the publisher's web-site.

Table S1
List of genes upregulated on Rifampicin treatment determined by Microarray analysis along with the number of DAF-16 binding sites in their 5 kb upstream region.

Table S2
List of genes downregulated on Rifampicin treatment determined by Microarray analysis along with the number of DAF-16 binding sites in their 5 kb upstream region.