Increasing breakdown of S-adenosyl-methionine may extend fruit fly lifespan

has a single SAM synthase () (). We first knocked downto test whether it increases the lifespan in, as reported in.-RNA interference (RNAi) crossed with an ubiquitous driver,, showed developmental lethality, indicating thatis an essential gene for proper development. Next, we checked the lifespan of male flies with adult-specific knockdown ofby, which is the RU486-inducible binary expression system in the whole body. We were able to bypass the developmental lethality of-RNAi, and thus, we analysed the adult lifespan in the presence or absence of RU486. In contrast to-RNAi in, we observed a significantly shortened lifespan in(). Drosophila melanogaster sams sams Drosophila C. elegans UAS-sams da-Gal4 sams sams Tub -Gal4 sams sams C. elegans Drosophila Fig. 1a Fig. 1b GS

It was possible that-RNAi had tissue-specific negative effects. Because methionine metabolism is active in the liver of mammals, we knocked downspecifically in the fat body, a counterpart of the liver and white adipose tissue in. We compared the lifespan of-RNAi flies to UAS-only control flies, as, but not, was backcrossed in this experiment. Both the female and male lifespans were significantly decreased by knocking downin fat body (). Although this manipulation is not adult specific, asis also expressed during the developmental stage, inhibiting SAM synthesis indid not demonstrate positive effects on lifespan, raising a question about what causes the different outcomes between flies and worms. sams sams Drosophila FB>sams FB-Gal4 UAS-sams-RNAi sams FB-Gal4 Drosophila Supplementary Fig. 1a,b

We suspected that there were differences in the metabolic regulation of SAM. To find out how SAM metabolism was affected by-RNAi, we analysed the metabolites of methionine metabolism by ultra-performance liquid chromatography tandem mass spectrometry (UPLC–MS/MS) (). Surprisingly,-RNAi in the fat body using two different fat-bodydrivers (and) did not result in a systemic SAM reduction (), despite the marked decrease in Sams proteins (). Methionine was drastically elevated, indicating that the synthesis of SAM from methionine was indeed inhibited (). Fruit flies, mice and humans, but not worms, have Gnmt () that catalyses the conversion of glycine to sarcosine (-methyl-glycine) by methyl group transfer using SAM (). Gnmt works as a regulator of SAM levels in metabolic organs such as liver in mammals or the fat body in. Knockdown ofin fat body led to reduced Gnmt proteins (), elevated SAM levels () and decreased Sar levels (), although there is an unexpected slight increase in all three metabolites in negative-control flies (). When methionine was added to food, the increase in SAM levels was limited, whereas the amount of sarcosine was increased in a dose-dependent manner (). Conversely, sarcosine levels significantly decreased inknockdown male flies (), in contrast to negative-control flies (where they were increased for unknown reasons) (), indicating that Gnmt maintains SAM levels by reducing SAM consumption. Interestingly,-RNAi in the fat body led to a reduction in Gnmt protein (), which corresponded to a reduced amount of sarcosine, suggesting the existence of transcriptional or post-transcriptional regulation of Gnmt expression for maintaining SAM levels. Considering that-RNAi resulted in a decrease in the amount of SAM in, the buffering of SAM levels by Gnmt may explain the different-RNAi phenotype. However, there are other possibilities, including differences in the knockdown efficiency ofor the number ofgenes (four in, but only one in). sams sams Gal4 FB-Gal4 r -Gal4 N Drosophila gnmt sams sams sams1 C. elegans sams sams sams C. elegans D. melanogaster Fig. 1c–l Fig. 1d,g Fig. 1c Fig. 1e,h Fig. 1a ^{6 12} Fig. 1c Fig. 1d,g Fig. 1f,i Fig. 1j–l Supplementary Fig. 1c–e Fig. 1f,i Fig. 1l Fig. 1c ¹³ 4 http://www.genome.jp/kegg/↗

We further analysed how these metabolites are affected by overexpressing eitherorubiquitously or in a fat-body-specific manner (;).overexpression decreased Met and increased SAM. Introducing the R275H mutation, which was observed in several human hypermethioninemia patients who had a dominantmutation, led to increased Met without affecting SAM levels. This result resembled the-RNAi phenotype. MAT1A in mammals forms dimers (called MAT III) or tetramers (called MAT I). R264 is essential for dimerization, and the R264/R264H dimer is enzymatically inactive. Thus, we considered that Samsmay behave as a dominant-negative form inas well. On the other hand,overexpression decreased SAM levels, although subtly, as well as methionine levels. The reason methionine was also reduced in-overexpressing male flies was unknown; however, we thought that the reduction in SAM accelerated SAM synthesis from methionine as a feedback mechanism for maintaining SAM levels. We also established a mutant version of aoverexpression line by introducing a serine 145 to alanine substitution. Since S145A is located at the SAM-binding site of Gnmt, we speculated that this mutation might result in reduced enzymatic Gnmt function. Indeed, the effect ofoverexpression on SAM and methionine was attenuated, although it was not complete (;). sams gnmt sams MAT1A sams Drosophila gnmt gnmt gnmt gnmt Fig. 2a–i supplementary Fig. 2a ^{14 14 15} Fig. 2a–i supplementary Fig. 2b–d R264H R275H S145A

We then assessed whether SAM reduction in-overexpressing flies had a positive impact on lifespan. Whenwas overexpressed using, we observed a slight but statistically significant increase in male lifespan (). Several enzymes, including Gnmt, have functions other than enzymatic activity, making us question whether Gnmt-induced longevity is dependent on its enzymatic activity. Overexpression ofbydid not show any significant effect on lifespan, confirming that enhancing enzymatic activity of Gnmt is necessary for lifespan extension (). Unexpectedly, we did not observe the lifespan increase in female flies when we compared lifespans ofwith those of+RU (). However, there was a significant increase inwhen compared with, even in the absence of RU. It is possible that leaky expression of thedriver was enough to extend lifespan in females, or that the position () effect ofcould have different effects on males and females. Therefore, we also checked the effect of Gnmt overexpression on lifespan using a differentoverexpression line, which was inserted at thesite (-2). In line with this, we observed a significant increase in male and female lifespans (), suggesting that Gnmt overexpression can be beneficial for both sexes. gnmt gnmt Tub -Gal4 gnmt Tub -Gal4 gnmt gnmt gnmt gnmt Tub attP40 UAS-gnmt gnmt attP2 gnmt GS S145A GS S145A GS Fig. 3a,b ¹⁶ Fig. 3a,b Supplementary Fig. 3a,b Fig. 3c,d

Genetic and pharmacological interventions targeting key metabolic pathways such as insulin/IGF-1 signalling (IIS) or the target of rapamycin (TOR) pathway also increase organismal lifespan. Despite accumulating genetic studies using model organisms, our knowledge of the mechanisms underlying lifespan extension by DR or IIS/TOR inhibition is not complete. For example, inhibiting the IIS pathway exerts its effect on longevity through the transcription factor FoxO. However, the precise molecular targets and mechanism of FoxO-dependent lifespan increase have not been fully revealed. Interestingly, studies inindicatedexpression is negatively regulated by, an orthologue of FoxO, implying that activation of FoxO under reduced IIS/TOR signalling or DR can extend longevity through repression ofmesenger RNA. In, dFoxO transcriptionally induces, but it did not repressat least at the whole-body level (), implying that Gnmt may contribute to IIS/FoxO-dependent lifespan extension. We observed in our experimental conditions that adult-specific overexpression ofin male flies significantly increased lifespan (). As expected, we showed that knockdown ofpartially attenuated the lifespan extension of(), suggesting that Gnmt mediates lifespan extension under reduced IIS activity. This was not due to the dilution of Gal4 by introducing two UAS lines simultaneously, as we observed the same degree ofinduction by RU treatment in both lines (). Interestingly,-RNAi alone did not affect lifespan, indicating that increases in SAM do not have a negative impact on the lifespan (). This was also supported by the data thatoverexpression in male flies did not affect the lifespan (). ^{1 9 17 18 19 12} Supplementary Fig. 3c Fig. 3e,f Fig. 3e,f Fig. 3g Fig. 3e,f Fig. 3h C. elegans sams-1 daf-16 sams-1 Drosophila gnmt sams InR gnmt Tub >InR InR gnmt sams DN GS DN DN

We also tested whether DR-induced longevity requires Gnmt. We used 20% SY diet as nutrient-rich food and 5% SY diet as DR diet. When compared with nutrient-rich condition, DR significantly extended the lifespans of both male and female flies in our lab conditions (). Then, we subjected theloss-of-function mutant, which is the protein null mutant that we characterized previously and in which neither Gnmt protein nor sarcosine could be detected, to DR lifespan analysis. Althoughhas a different genetic background since it showed lethality when backcrossed to, we found that DR-induced longevity was not observed inmutant (), suggesting the possibility that Gnmt-dependent SAM catabolism mediates DR longevity. In addition,expression is positively regulated by the oxidative and xenobiotic stress-responsive factor CncC/Nrf2 (), which is also suggested as a mediator of DR longevity, indicating that Gnmt is a common downstream target for longevity pathways. Fig. 3i–l ¹² Fig. 3i–l Supplementary Fig. 3d–h ^{20 21} gnmt gnmt w gnmt gnmt Mi Mi 1118 Mi

Although Gnmt overexpression extended longevity, and Gnmt is required for DR-induced longevity, it is still unknown whether Gnmt activity and SAM metabolism changes during physiological ageing. We quantified Met and SAM in two different wild-type strains analysing both young and aged male flies. Met levels decreased during ageing, whereas SAM levels increased (). We also found that the amount of sarcosine decreased during ageing (), suggesting Gnmt activity to buffer SAM declines in an age-dependent manner. However, the expression of Gnmt was induced in aged male flies, as determined by quantitative real-time PCR (qRT–PCR) and western blotting using whole-body homogenates (;). Sarcosine levels are negatively regulated by sarcosine dehydrogenase (sardh), and Sardh was also induced transcriptionally during ageing (). Interestingly, whenexpression was knocked down, sarcosine levels were high in young male flies and further induced by ageing (), suggesting that total Gnmt activity is indeed increased during ageing. Gnmt was increased in the fat body, because fat-body-specific knockdown ofabrogated elevation of Gnmt in whole-body homogenates (). This increase was dependent on dFoxO (;). Induction of Gnmt expression during ageing might be an adaptive response against an increase in SAM levels. Indeed, whenwas knocked down in the fat body, SAM levels in aged male flies were further elevated (). Therefore, we concluded that the dFoxO–Gnmt pathway is activated, but not sufficiently for complete suppression of the SAM increase in aged flies. Importantly, overexpression of, but not, inhibited the age-dependent SAM increase (;), indicating that Gnmt-induced lifespan extension is caused by suppression of the age-dependent increase in SAM. Fig. 4a–d Fig. 4e Fig. 4f,g Supplementary Fig. 4a Fig. 4f,h Fig. 4i Supplementary Fig. 4b Fig. 4j Supplementary Fig. 4c Fig. 4k Fig. 4k Supplementary Fig. 4d,e 12 S145A sardh gnmt dFoxO gnmt gnmt

Since Gnmt is required for DR-dependent lifespan extension, we analysed how DR altered systemic SAM levels. We analysed young (1 weeks (w)) and aged (5 w) male flies from two wild-type strains as well as amutant strain maintained on a DR- (5% SY) or nutrient-rich (20% SY) diet. Since the 20% SY diet contains much Met, young flies maintained on 20% SY diet showed a slight increase in Met levels; however, changes in SAM levels were not statistically significant (;), suggesting that Gnmt buffered SAM increase. Indeed, the lack of regulation byresulted in increased SAM in male flies on the 20% SY diet compared with that of files on the 5% SY diet (). As previously mentioned, Met levels tended to decrease with ageing in all three genotypes, but this phenotype was not affected by the diet (;). In contrast, an age-dependent increase in SAM levels was suppressed by DR (;). Themutants showed high SAM levels in young flies, but this was not further elevated by ageing (), suggesting that a threshold for SAM increases exists in aged flies. gnmt Gnmt gnmt Fig. 5a,b Supplementary Fig. 5a,b Fig. 5a Fig. 5b Supplementary Fig. 5b Fig. 5a Supplementary Fig. 5a Fig. 5a

In addition to DR, we checked whether reduced IIS activity affected systemic SAM levels by analysing. Compared with control flies (), in which SAM was increased during ageing regardless of RU486 treatment (), SAM levels were rather significantly decreased inold male flies upon RU486 treatment than that in the control flies (old flies without RU486 or young flies with RU486), although the reason behind the RU486-induced SAM elevation inyoung flies is unknown (). Interestingly, an age-dependent SAM increase was also rescued by overexpressing the dominant-negative form of the TOR () (). These data suggested that the suppression of SAM increases in aged flies was a common mechanism underlying lifespan extension by DR or reduced IIS/TOR pathway. Tub >InR Tub >lacZ Tub >InR Tub >InR Tub >TOR GS DN GS GS DN GS DN GS TED Fig. 5c Fig. 5d Fig. 5e

Flies were reared on a standard diet containing 4% cornmeal, 4% baker's yeast (Oriental Yeast), 10% glucose and 0.8% agar with propionic acid and nipagin at constant 25 °C, 60% humidity under 12–12-h light–dark conditions. For most biochemical analysis other than lifespan study, all flies were collected within 2 days after adult eclosion and maintained for 5 days with free access to food and mating for adult maturation, unless otherwise stated.

,and Canton S were utilized as wild-type strains.-RNAiwere generated in our previous study.was kindly provided by S. Pletcher.was kindly provided by D. Bohmann.,(K1409A),-RNAi (HMS00422) andwere obtained from the Bloomingtonstock center.-RNAi (v25983),-RNAi (v103143) and-RNAi (v107052) were obtained from the Viennaresource center.,,,,,,-RNAi andRNAi were all backcrossed at least six generations into.was maintained on its original genetic background, because it shows lethality when backcrossed onto. Thus, we used theline as a control.,,andwere generated as described below. For overexpression of metabolic enzymes in fat body,was used because it is stronger thanFor most biochemical analyses of adult male flies, we crossed the same batch ofvirgin females (once mixed after virgin collection, and then separated into each group) with different males including wild-type orlines as control. UAS-only controls were also analysed for some experiments, which we believe is essential. w yw UAS-gnmt Tub -Gal4 UAS-CncC r -Gal4 UAS-InR UAS-dFoxO UAS-dFoxO Drosophila UAS-gnmt UAS-sams UAS-keap1 Drosophila da-Gal4 FB-Gal4 Mef2-Gal4 pxn-Gal4 NP1-Gal4 elav-Gal4 UAS-gnmt UAS-lacZ- w yw;gnmt w yw UAS-gnmt UAS-gnmt UAS-sams UAS-sams r -Gal4 FB-Gal4. Gal4 lacZ 1118 sh GS 4 DN 1118 Mi 1118 S145A R275H 4 ¹²

We establishedoverexpression lines by phiC31-integrase-mediated transgenesis as follows. For making, a PCR-amplified 870-bp fragment of the full-length complementary DNA (cDNA)was reinserted into thevector byRI/I, and it was then inserted into the attP40 site using phiC31-mediated transgenesis (Best Gene). For, we subcloned full-lengthcDNA once into thevector byRI/I and introduced the TCC→GCA mutation by PCR-based site-directed mutagenesis using the following primers: 5′-GGCAACGCATTTGCCCACTTGATGGAC-3′, and 5′-GGCAAATGCGTTGCCCAAGCAAATGAC-3′, followed by reinsertion into thevector. For, a 1.2-kbp cDNAfragment was amplified by PCR from total RNA of adultusing the following primers: 5′-GAATTCATGCCGCAAAAGAC-3′, and 5′-CTCGAGTCAGTTGTCAATCTCC-3′ and cloned into thevector.cDNA was amplified by PCR usingas template and the primers: 5′-GAAGATCTATGCCGCAAAAGAC-3′, and 5′-GGGCTCGAGTCAGTTGTCAATCTCC-3′ and subcloned once into the pUAST-3xFlag vector. The vector was then sequenced. Full-length, 1.2-kbpcDNA without a Flag tag was digested byII/I and subcloned into thevector. For UAS-,cDNA was subcloned once into thevector and a AGA→CAC mutation was introduced by PCR-based site-directed mutagenesis using the following primers: 5′-CTTACAGGACACAAAATCATTGTAGATACT-3′, and 5′-AATGATTTTGTGTCCTGTAAGCCCAGCGTC-3′. Transgenes were inserted into thesite on the second chromosome orsite on the third chromosome using phiC31-mediated transgenesis (Best Gene) and maintained as homozygous lines without crossing with balancer lines. gnmt UAS-gnmt pUAS-attB Eco Xho UAS-gnmt gnmt pBSSK Eco Xho pUAS-attB UAS-sams sams w pT7 sams pT7-sams sams Bgl Xho pUAST-attB sams sams pBSSK attP40 attP2 ¹² S145A 1118 R275H

For DR experiments, we prepared a 20% SY diet and 5% SY diet. The 20% SY diet contains 20% baker's yeast (Oriental Yeast), 20% sucrose (Wako), and 1.5% agar (Kishida Chemical) with propionic acid and nipagin. The 5% SY diet contains 5% baker's yeast (Oriental Yeast), 5% sucrose (Wako) and 1.5% agar (Kishida Chemical) with propionic acid and nipagin. For making these diet, first yeast and sucrose were dissolved in water and boiled for 15 min to kill the living yeast completely. This mixture was added into agar, which was dissolved in water with 10-min heating in the different pot. After that, propionic acid and nipagin were added directly into the diet. For the gene switch (GS) system, 10% SY diet was prepared and RU486 (Sigma) was added into the 10% SY diet at a final concentration of 200 μM to drive gene expression. For the high-Met diet, we added 1, 2 or 5 ml of 100 mM stock methionine solution in a 100-ml standard diet. ²⁵

To control environmental effects during development, we crossed the same number of female and male flies in the same period on a standard diet. Newly hatched flies were collected within, at most, 2 days, and then they were kept for 2 days under the mixed sex condition. Then, males and females were separated and grouped into 20 flies per vial. Flies were transferred to fresh food every 3 or 4 days, and at that time, we counted the number of dead flies.

For western blot analysis, five adult male flies were homogenized in PBS supplemented with 0.1% SDS, 10 mM dithiothreitol and 1 × protease inhibitor cocktail (Roche). Samples were boiled for 5 min at 98 °C after mixing with 6 × Laemmli sample buffer. Whole-body proteins corresponding approximately to 0.15 fly were subjected to SDS–polyacrylamide gel electrophoresis and transferred onto polyvinylidene difluoride membrane. After 30 min blocking with 5% skimmed milk, membranes were incubated with the 1st antibody at 4 °C overnight. Anti-Gnmt antibody (1:3,000) and anti-Sardh antibody (1:1,500) were generated previously. Anti-Sams antibody (1:1,000) was produced in guinea pigs using a synthetic peptide corresponding to the C terminus 382–400 ofSams as an antigen. For loading control, anti-α-tubulin monoclonal antibody (1:3,000, Sigma DM1A) was used. Representative images were shown from at least two independent experiments with reproducible results. Full blot images for main figures are shown in. For qRT–PCR, total RNA was purified from five adult males using the Qiazol and RNeasy micro kit (Qiagen). cDNA was made from 100 or 200 ng DNase-treated total RNA by the Takara PrimeScript RT Reagent Kit with gDNA Eraser. Quantitative PCR was performed using Takara Premix Ex Taq II (Tli RNaseH Plus) and the Light Cycler 480 system (Roche). For internal controls,was predominantly used, butwas also checked to rule out the possibility that changes in target gene expression resulted from altered expression of internal controls. Primer sequences are available in. ¹² Supplementary Fig. 6 Supplementary Table 1 Drosophila Rpl32 RNA pol II

Methionine, SAM and sarcosine levels were measured by UPLC equipped with tandem MS, TQD (UPLC–MS/MS, Waters). Briefly, five adult males were homogenized in 50% methanol, deproteinized by acetonitrile and evaporated completely. Pellets were solubilized in 10 mM HCl followed by filtration using 0.22-μm polyvinylidene difluoride filters (Millipore). Samples were mixed in equal volume of 50 mM Tris-HCl pH 8.8 with 100 μM dithiothreitol for Met and SAM measurements. For the sarcosine measurement, samples were derivatized after filtration by the AccQ-Tag Ultra Derivatization Kit (Waters). Samples were subjected to UPLC system with Acquity UPLC BEH C18 column. Separated solutions were ionized by electrospray ionization in positive-ion mode (ESI+) and detected by TQD with the following/transitions: Methionine, 149.98>132.90, SAM, 399.20>250.00 and derivatized sarcosine, 260.00>171.50. ¹² m z

Because absolute metabolite amounts can easily vary between experiments due to technical (for example, machine conditions) as well as biological (for example, season, food, genotype and so on) conditions, we always prepared a control sample in the same analytical round to compare the quantities as ratios relative to the control. To analyse the metabolites precisely, the quality of parents, the developmental environment and sampling time (to avoid circadian changes) were carefully controlled, and each graph was constructed from the same round of analysis.