Lifespan regulation in α/β posterior neurons of the fly mushroom bodies by Rab27

Abstract Brain function has been implicated to control the aging process and modulate lifespan. However, continuous efforts remain for the identification of the minimal sufficient brain region and the underlying mechanism for neuronal regulation of longevity. Here, we show that the Drosophila lifespan is modulated by rab27 functioning in a small subset of neurons of the mushroom bodies (MB), a brain structure that shares analogous functions with mammalian hippocampus and hypothalamus. Depleting rab27 in the α/βp neurons of the MB is sufficient to extend lifespan, enhance systemic stress responses, and alter energy homeostasis, all without trade‐offs in major life functions. Within the α/βp neurons, rab27KO causes the mislocalization of phosphorylated S6K thus attenuates TOR signaling, resulting in decreased protein synthesis and reduced neuronal activity. Consistently, expression of dominant‐negative S6K in the α/βp neurons increases lifespan. Furthermore, the expression of phospho‐mimetic S6 in α/βp neurons of rab27KO rescued local protein synthesis and reversed lifespan extension. These findings demonstrate that inhibiting TOR‐mediated protein synthesis in α/βp neurons is sufficient to promote longevity.

longevity and the size/weight of an individual as well as the reproductive success (Maklakov & Immler, 2016).
Animals need to sense, integrate, and adapt to changes from diverse physiological and environmental cues. In the mammalian hypothalamus, different clusters of neurons integrate internal and external inputs to regulate important life functions including appetite, body temperature, and sleep. For example, the murine arcuate nucleus integrates hormonal signals from the periphery including ghrelin and leptin to regulate food intake for energy homeostasis. Two other hypothalamic nuclei, the preoptic area, and the dorsomedial hypothalamus integrate peripheral inputs to maintain body temperature. This evidence illustrates the importance of the central nervous system (CNS) for the summation of peripheral signals to regulate systemic homeostasis. Whether the CNS also functions as a control hub for lifespan regulation; however, remains debatable.
Several recent studies suggest that the hypothalamus plays a pivotal role in longevity and it may be a regulator of systemic aging (Zhang et al., 2013;Zhao et al., 2017). In the fruit fly Drosophila, the functions of mammalian hypothalamus are divided among the mushroom bodies (MB) and the pars intercerebralis (Bang et al., 2011;Dus et al., 2015). Within the pars intercerebralis, the median neurosecretory cluster (mNSC) has been shown to regulate lifespan, as the ablation of mNSC results in altered systemic energy metabolism, enhanced stress tolerance, and lifespan extension (Broughton et al., 2005). The MB consists of ~ 2000 Kenyon cells that can be classified according to the axon innervation patterns into three major groups: the α/β, α'/β', and γ lobes. Although the MB is important for learning and memory similar to the hippocampus, the MB also shows functional analogy to the mammalian hypothalamus in regulating food-seeking behavior, courtship behavior, sleep, and temperature preference (Bang et al., 2011;Joiner, Crocker, White, & Sehgal, 2006;McBride et al., 2005). For example, the Drosophila MB integrates hunger and satiety signals to regulate feeding behavior to meet organismal needs (Tsao, Chen, Lin, Yang, & Lin, 2018). Also, reduction of insulin signaling in the MB decreases food consumption (Zhao & Campos, 2012). However, whether the MB contributes to the regulation of lifespan remains unknown.
Rab27 is a highly conserved small Rab GTPase widely expressed in various secretory cells, including endocrine, cancer, and immune cells, and is well recognized for its role in exosome secretion. We have reported that the Drosophila Rab27 is exclusively expressed in neuronal tissues predominantly in specific brain regions including the MB (Chan et al., 2011;Jin et al., 2012). At present, the neuronal functions of Rab27 remain poorly understood. Rab27 regulates the docking of dense-core vesicles in PC12 neuroendocrine cells (Tsuboi & Fukuda, 2006), whereas inhibition of Rab27 is found to impair synaptic transmission in C. elegans and giant squid (Mahoney et al., 2006;Yu et al., 2008). We have previously characterized rab27KO flies as viable and fertile, without apparent developmental defects (Chan et al., 2011). Here, we describe our investigation of how Rab27 functions to control lifespan in a small subset of neurons in the Drosophila MB and the underlying molecular mechanism.

| Loss of rab27 in adult neurons is sufficient to prolong lifespan
We have shown that rab27  homozygotes are viable (this strain was used for most experiments unless otherwise noted and is herein referred to as rab27KO) (Chan et al., 2011). Characterizing these flies in detail, we found that the homozygous rab27KO flies showed a pronounced lifespan extension compared with wild-type (WT) controls (average survival + 46.6%) and rab27 heterozygotes also had an intermediate yet significant lifespan extension in the populationbased longevity assay (Figure 1a,b, average survival + 20.3%; (Flatt et al., 2008); all of the lifespan statistics of the female were shown in the following figures and summarized in Table S1, whereas those of the male were summarized in Table S2). We generated another independent null allele of rab27 using CRISPR/cas9 (rab27 Crispr-KO , Figure S1a,b) which showed the same effect on lifespan (Figure 1c,d, average survival + 53.5%). These two rab27 knockout strains were further validated by the single-sex vial assays (Bjedov et al., 2010), and they exhibited similar trends of lifespan extension ( Figure S1c).
The effect of rab27KO on lifespan was reversed by the expression of rab27 cDNA, confirming that rab27 regulates longevity (Figure 1c,d).
Thus, the global loss of rab27 prolongs lifespan in Drosophila.
We have shown that Rab27 is expressed throughout life (Chan et al., 2011;Jin et al., 2012). We observed no difference in larval body length and pupation timing between WT and rab27KO  Figure S2f). Thus, reducing rab27 mRNA in the adult brain is sufficient to extend lifespan.

| rab27 controls systematic stress response and energy metabolism
Many long-lived mutants are reported to be highly resistant to oxidative stress and starvation. Indeed, we found that rab27KO flies lived longer than WT even under starvation or paraquat-induced oxidative stress (Figure 2a,b). Also, we observed a similar increase in the lifespan of rab27KO under starvation in an isogenized Canton-S background ( Figure S3a), suggesting that the extension effect is independent of genetic backgrounds. Moreover, the levels of Wolbachia DNA in our experimental strains were negligible ( Figure S3b) and the longevity of rab27KO flies remained after 3 generations of tetracycline treatment ( Figure S3c), demonstrating that the effect of rab27KO on longevity was not related to Wolbachia infection, which has been shown to reduce the lifespan in Drosophila (Min & Benzer, 1997).
The nuclear translocation of Forkhead box class O (dFOXO) in the peripheral tissues of Drosophila is associated with an extended lifespan, increased stress resistance, and altered lipid metabolism (Hwangbo, Gershman, Tu, Palmer, & Tatar, 2004). The activation of dFOXO in the fat body, as indicated by its nuclear localization, is a common feature linked to extended lifespan (Martins, Lithgow, & Link, 2016). We thus examined the subcellular dFOXO localization in adult fat bodies, a tissue that does not express rab27. In the rab27KO fat bodies, we found a reduced ratio of cytoplasmic-to-nuclear dFOXO compared with that of WT control (Figure 2c-d''), suggesting a cell non-autonomous effect of rab27KO. In rab27KO adults, the levels of circulating glucose, TAG, and food intake were significantly higher than those in WT controls, and the increases in metabolites could be reversed by rab27 expression (Figure 2eg). Notably, while many long-lived mutants are known to prolong lifespan at the expense of body size/weight and fecundity (Bai, Post, Kang, & Tatar, 2015;Bjedov et al., 2010;Gronke, Clarke, Broughton, Andrews, & Partridge, 2010), rab27KO flies exhibited comparable weight and fecundity with WT (Figure 2h,i). Also, the olfactory memory of rab27KO flies was equivalent to that of WT (Figure 2j). We concluded that removing rab27 increases resistance to starvation and ROS stress and affects specific aspects of energy metabolism.

| Reduction of rab27 in the α/β posterior (α/βp) neurons of the MB increases longevity
Rab27 protein is only detected in neurons (Jin et al., 2012). We Within the MB, knocking down rab27 in the α/β lobes leads to a modest but significant increase of lifespan (Figure 3c', average F I G U R E 1 Loss of rab27 in adult neurons is sufficient to prolong lifespan. (a) Survival of females from w 1118 control (black), rab27 heterozygote rab27KO/+ (blue), and homozygous rab27KO (red), p < .001. (b) Relative abundance of rab27 mRNA in strains in (a) determined by qRT-PCR in fly heads; mean ± SEM of three independent experiments. **p < .01, ***p < .001, one-way ANOVA. (c) Survival of females from two independent rab27 nulls, rab27 Crispr-KO (green), and rab27KO (red), comparing with w 1118 control (black) and rab27KO;UAS-rab27 (purple), p < .001. (d) Relative abundance of rab27 mRNA in strains in (c) in fly heads. *p < .05, ***p < .001, one-way ANOVA. (e) Body length measured at 24, 48, and 72 hr after egg collection, n = 3 independent experiments. (f) Developmental time from egg to pupa in w 1118 and rab27KO, n = 3 independent experiments. (g) Relative abundance of rab27 mRNA in the heads of elav-GS-Gal4 > UAS-rab27-RNAi flies fed with RU486 normalized to the solvent-fed control (EtOH). p < .01, significance determined by one-way ANOVA, also for Gal4 or UAS, controls is shown in Figure S2e. (h) Survival of females from RU486-induced elav-GS-Gal4 > UAS-rab27-RNAi (red) compared with the solvent-fed control (black), p < .001. All survival data were analyzed by log-rank tests. Please see Tables S1 and S2 for detailed information including mean lifespan and statistical comparisons survival + 17.5%), whereas reducing rab27 in the α'/β' lobes or γ lobe did not ( Figure S5c Underlined numbers at the upper right corner (c", d") indicate the ratio of cytoplasmic-to-nuclear dFOXO signal (n = 3; p < .05). Nuclei are labeled by DAPI (blue). Scale bars: 25 µm. (e) Levels of circulating glucose in females normalized to w 1118 , *p < .05, ***p < .001, n = 3 independent experiments. (f) Levels of TAG in whole female animals, normalized to both the body weight and the level in w 1118 . *p < .05, ***p < .001, n = 3 independent experiments. (g) Levels of food intake normalized to w 1118 . p < .01, n = 10 independent experiments. (h) Body weight of 21-day-old female w 1118 and rab27KO flies, n = 70 flies per condition. (i) Quantification of female fecundity shown by accumulated eggs laid from day 7 to day 42 post-eclosion. n = 20 flies per condition. (j) The performance index of an olfactory associative learning assay testing 7-day-old flies of the indicated genotypes. n = 3 independent experiments Importantly, expressing rab27 cDNA in the α/βp neurons significantly reversed the lifespan extension of rab27 Crispr-KO ( Figure 3f).
In sum, our data indicate that the α/βp neurons of MB regulate lifespan via a rab27-dependent mechanism.

| The α/βp neurons are required for lifespan maintenance
We next examined the requirement and mechanism of the α/βp Here, we showed that thor overexpression in the α/ βp neurons did not extend lifespan ( Figure S8). Altogether, rab27 deactivates TOR signaling for lifespan extension likely through an s6kspecific but thor-independent pathway.

| Rab27 binds with p-S6K and determines its subcellular localization
Because many Rab proteins are known regulators of vesicle trafficking, we performed sucrose gradient centrifugation of fly head extracts to determine whether Rab27 regulates the localization of TOR pathway components. In the head extracts of Rab27 EYFP flies, Rab27 co-fractionated with phosphorylated S6K (p-S6K) proteins ( Figure S9). Compared with the WT control, the distribution of p-S6K changed in rab27KO (Figure 5e), whereas two control proteins, Rab7 and ATP5A, showed no differences, suggesting that rab27KO did not abolish global protein distribution but specifically affected the subcellular localization of p-S6K. Also, we found that p-S6K co-im-

| Rab27 anchors activated S6K to the periphery of α/βp neurons for de novo protein synthesis
We then investigated the subcellular localization of Rab27 to further understand its function in the α/βp neurons. Since Rab27 has long been recognized to function in cargo docking during exocytosis (Fukuda, 2006;Kasai et al., 2005), we speculated that Rab27 may The in vivo association of Rab27 EYFP and p-S6K determined by the Proximity Ligation Assay (PLA). The brains were stained with anti-GFP antibodies to detect Rab27 EYFP (green) and anti-p-S6K antibodies (red), followed by the addition of PLA-specific probes to detect proximity ligation events (gray). White circles mark the post-synaptic, dendritic area of the α/βp neurons. and Figure 6m), and the reduction was rescued by s6-S5D expression was not affected by the expression of a constitutively activated S6K (s6k CA ) ( Figure S12a). In contrast, the lifespan was significantly reversed by the expression of phospho-mimetic s6-S5D specifically in the α/βp neurons of rab27KO ( Figure S12b). As a control, overexpression of rab27 or phospho-mimetic s6-S5D in the WT α/βp neurons did not reduce lifespan ( Figure S12c). Therefore, loss of rab27 reduced protein synthesis in the α/βp neurons, likely as the consequence of mislocalized p-S6K and reduced phosphorylation of S6.

| D ISCUSS I ON
Our data show that depleting rab27 in the α/βp neurons of the

Gems, & Withers, 2005). Growth is another life function that is
highly associated with longevity from flies to monkeys (Mattison et al., 2017). For example, mutants of chico, which encodes the insulin receptor substrate in Drosophila, are long-lived but have a metabolic imbalance and small body size (Bohni et al., 1999).
However, the link between lifespan and body weight seems to be more complicated, as Slack et al. (2015) have shown that a chico mutation that disrupts its interaction with Grb2/Drk extends lifespan without affecting body weight (Slack et al., 2015). Moreover, a mutation in the odorant receptor Or83b 2 increases TAG level, enhances starvation tolerance, and extends lifespan with no effect on body weight or female fecundity (Libert et al., 2007). These reports and our findings all point to the possibility to uncouple lifespan control from the performance in essential life functions measured in this study. We noticed that the aforementioned phenotypes, such as glucose level, TAG level, food intake, and stress tolerance, are detected in aged flies but not young ones. There are two possibilities. For one, Rab27 may be only required at old age so that rab27KO only causes physiological declines that are significant in old flies. Alternatively, the minute phenotypical differences that are below detectable levels at younger ages may accumulate over time and become obvious upon aging.
Most studies have attempted to modulate longevity at the level of individual animals. For example, systematic inhibition of the TOR signaling has been shown to extend lifespan from yeast to mammals. However, since the TOR pathway also regulates cell growth, ribosome biogenesis, and the process of translation, systemic inhibition of TOR signaling may come with a wide range of complications such as stomatitis, diabetes, and nephrotoxicity (Kaplan, Qazi, & Wellen, 2014). Instead of systematic manipulation, one plausible F I G U R E 6 Rab27 anchors p-S6K to the periphery of the α/βp neurons for de novo protein synthesis. (a-c''') Representative confocal fluorescence images of the dendrites of the α/βp neurons expressing UAS-mCD8-GFP (green) stained with anti-p-S6 (red) and DLG (postsynaptic marker, gray) in VT14429-Gal4 > UAS-mCD8-GFP (a-a'''), rab27 Crispr-KO ;VT14429-Gal4 > UAS-mCD8-GFP (b-b''') or rab27 Crispr-KO ;VT14429-Gal4 > UAS-rab27, UAS-mCD8-GFP (c-c'''). Scale bars: 10 µm. (d-i') Tracing de novo protein synthesis in the α/βp neurons with photoconvertible Kaede protein. Adult VT14429-Gal4 > UAS-Kaede or rab27 Crispr-KO ;VT14429-Gal4 > UAS-Kaede animals were exposed to ultraviolet light (UV) for 6 hr (d,f) to irreversibly convert all existing Kaede to red fluorescence (d',f'). Newly synthesized green Kaede protein was examined 1 hr after UV exposure in VT14429-Gal4 > UAS-Kaede (e-e') or rab27 Crispr-KO ;VT14429-Gal4 > UAS-Kaede (g-g'). The strain for rescue experiments: rab27 Crispr-KO ;VT14429-Gal4 > UAS-s6-S5D, UAS-mCD8-GFP (h-i'). Scale bars: 25 µm. (j) Quantification of p-S6 intensity by marking DLG positive areas and measured the red signal. The fluorescence intensity was normalized to the control levels. Data are represented as mean ± SEM measured in at least three brains, *p < .05, **p < .01. (k) Quantification of de novo protein synthesis expressed as the difference in green Kaede protein levels (ΔKaede) immediately after photoconversion and 1-hr post-UV. The ΔKaede level in rab27 Crispr-KO was normalized to the WT levels. Data are represented as mean ± SEM measured in at least three brains, **p < .01 strategy to avoid potential complications is to identify the minimal region required to promote longevity. The brain serves as a good candidate since it is the integration center of input and output signals and can regulate a wide range of physiological functions.
Previous reports have indicated that, in the Drosophila brain, the mNSC regulates systemic insulin levels by producing Dilps to mediate organism growth and lifespan (Broughton et al., 2005;Rulifson, Kim, & Nusse, 2002). Ablating mNSC leads to lifespan extension (Broughton et al., 2005), and limiting the secretion of Dilps from within the mNSC is also sufficient to increase longevity (Bai, Kang, & Tatar, 2012). In this study, we find that the α/ While the nervous system of C. elegans is not organized into a brain with higher-order structures and domains, our findings show that lifespan regulation by a small number of brain neurons may be an evolutionarily conserved phenomenon in the Drosophila brain, which shares functional analogous organization with higher organisms. In conjunction with these findings, we conclude that lifespan can be modulated by rab27 in TOR-mediated protein homeostasis in a small group of neurons. The consequence of reduced protein synthesis in the α/βp neurons of rab27KO remains to be investigated.

| Fly husbandry and stocks
Flies were maintained with standard cornmeal medium at 25℃,

| Generation of rab27 knockout flies using CRISPR/Cas9 system
The rab27 knockout flies (rab27 Crispr-KO ) were generated utilizing the CRISPR/Cas9 system as described in  with modifications. One pair of gRNAs was designed to target the start codon and 3' UTR of rab27 locus for the removal of the entire coding region. Further details are provided in Supporting Information.

| Lifespan analysis and antibiotic treatment of Wolbachia infection
Lifespan was measured with two independent methods. The "cage" method was modified from (Bai et al., 2015). Briefly, flies were raised at a density of approximately 200 larvae per bottle. Newly eclosed flies were allowed to mate within 48 hr and then transferred to experimental cages at a density of 100 males and 100 females in a 1-L cage with good ventilation. Fresh food was provided, and deaths were scored every 2-3 days. For the "vial" method, 10 newly hatched flies of the same sex were reared in a standard fly vial. Flies were transferred to fresh vials and dead flies were removed and scored every 2-3 days.
To create starvation-induced stress, 21-day-old adult flies were transferred to vials containing 1% agar. Flies were separated by sex into 10 flies per vial, and dead flies were counted every 6 hr. To assay oxidative stress resistance, flies were exposed to filter paper soaked with 5 mM paraquat (PQ) dissolved in 6.5% sucrose solution at the bottom of vials. The effects on survival were analyzed by the log-rank test.
The results of all lifespan experiments were summarized in Tables S1 and S2. For the antibiotic treatment, the flies were reared for three generations with 50 µg/ml tetracycline (Omics Bio) added to the food (Rottschaefer & Lazzaro, 2012). The infection status for Wolbachia was then verified via qRT-PCR with the wspB primers. Further details are provided in Supporting Information.

| Quantitative Real-Time PCR (qRT-PCR)
qRT-PCR was performed following the guideline of (Bustin et al., 2019). The procedures were as previously described in . Further details are provided in Supporting Information.

| Measurements of larval length and pupation timing
Body lengths were measured at 24, 48, and 72 hr after egg collection.
Larvae were fixed in 4% paraformaldehyde and then washed with phosphate-buffered saline (PBS). Images of larvae were taken with a Canon EOS-700D digital camera on a Leica S8APO microscope, and the body lengths were measured with ImageJ. To measure pupation time, eggs were collected for 24 hr and the number of pupae was recorded every 24 hr.

| Immunohistochemistry and confocal imaging
Adult fly brains were dissected, immunostained, and imaged following  with modifications. Further details are provided in Supporting Information.

| Quantification of dFOXO staining
A region of interest (ROI) was drawn around the DAPI positive nuclei, and an equally sized circle was drawn to mark the ROI in the cytoplasm in the same image. The green signal (dFOXO staining) was measured in these areas from individual confocal images using ImageJ. The ratio of cytoplasmic-to-+nuclear dFOXO was calculated by dividing the mean fluorescent intensity of cytoplasmic dFOXO to the mean fluorescent intensity of nuclear dFOXO from the same image. We then plotted all data in boxplots and used an unpaired Student's t test for significant differences.

| Triglycerides (TAG) and glucose measurement
TAG measurement was performed according to  with modifications. Further details are provided in Supporting Information.

| Feeding assay
21-day-old adult flies were starved for 6 hr. Subsequently, ten singlesex flies were transferred into a new vial containing standard cornmeal food and blue dye (0.0375 mg/ml, Sigma-Aldrich #3844-45-9) for 3 hr. Flies were homogenized with 400 µl PBS. After centrifugation, the amount of food ingested was determined by absorbance at wavelength 620 nm.

| Body weight measurement
Ten 21-day-old flies were weighed on a microbalance (Denver Instrument TB-124). Body weight measurements were performed in triplicates for each sex of each strain.

| Female fecundity
Eggs laid by mated female flies were counted daily from flies that were 7-day post-eclosion for 35 consecutive days. Fresh standard food was changed daily.

| Olfactory aversive memory
Conditioned odor avoidance was performed as previously described . Further details are provided in Supporting Information.

| Western blot and co-immunoprecipitation (co-IP)
Western blot was performed as previously described .

| Sucrose density gradient fractions
500 adult flies were flash-frozen in liquid nitrogen, vortexed, and passed through a small sieve to collect fly heads. Adult heads were homogenized using a pestle in lysis buffer (50 mM Tris pH8.0, 150 mM NaCl, 2 mM EDTA, 1% igepal, 0.5% sodium deoxycholate) with protease inhibitor cocktail (Roche). The lysate supernatant was layered on 20%-55% sucrose gradient. The gradient was centrifuged for 16 hr at 35,000 rpm at 4℃ in an SW-41 or SW-55 Ti rotor (Beckman). Serial fractions (1 ml each) were collected from the top of the tube and analyzed by Western Blotting. The band intensity was quantified with ImageJ software.
Further details are provided in Supporting Information.

| Kaede measurement
To measure the amount of newly synthesized Kaede proteins, preexisting Kaede was first photoconverted into red fluorescent proteins by UV irradiation. After 6-hr of UV irradiation, the flies were kept at 25℃, 60% humidity for 60 min. Next, the brains were dissected in PBS and fixed in 4% paraformaldehyde in PBS at room temperature for 45 min. The brains were then washed in 0.5% PBST three times for 20 min and mounted with Vectashield. The images were quantified and measured as described.

| Quantification and statistical analysis
Each experiment was performed at least three biological replicates in all graphs. For fluorescence images, the intensity was quantified double-blindly and measured using Adobe Photoshop CS6 and ImageJ. The band intensity of Western blotting or co-IP was quantified with ImageJ. All data were expressed as mean ± SEM and were compared using ANOVA followed by a Tukey test (for experimental groups ≥ 3) or an unpaired Student's t test (for experimental groups = 2). Survival data were analyzed by log-rank tests (Gronke et al., 2010). All statistical analysis was carried out using GraphPad Prism 5 software. A p < .05 was considered statistically significant: * indicates p < .05; ** indicates p < .01; *** indicates p < .001. All images were processed in Adobe Photoshop and assembled with Adobe Illustrator.

ACK N OWLED G EM ENTS
We would like to thank Drs. Jongkyeong Chung, Marc Tatar

CO N FLI C T O F I NTE R E S T S
The authors declare no competing interests.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available from the corresponding author upon reasonable request.