Accumulation of F-actin drives brain aging and limits healthspan in Drosophila

Summary The actin cytoskeleton is a key determinant of cell and tissue homeostasis. However, tissue-specific roles for actin dynamics in aging, notably brain aging, are not understood. Here, we show that there is an age-related increase in filamentous actin (F-actin) in Drosophila brains, which is counteracted by prolongevity interventions. Critically, modulating F-actin levels in aging neurons prevents age-onset cognitive decline and extends organismal healthspan. Mechanistically, we show that autophagy, a recycling process required for neuronal homeostasis, is disabled upon actin dysregulation in the aged brain. Remarkably, disrupting actin polymerization in aged animals with cytoskeletal drugs restores brain autophagy to youthful levels and reverses cellular hallmarks of brain aging. Finally, reducing F-actin levels in aging neurons slows brain aging and promotes healthspan in an autophagy-dependent manner. Our data identify excess actin polymerization as a hallmark of brain aging, which can be targeted to reverse brain aging phenotypes and prolong healthspan.


Summary 25
The actin cytoskeleton is a key determinant of cell and tissue homeostasis. However, 26 tissue-specific roles for actin dynamics in aging, notably brain aging, are not understood. Here,27 we show that there is an age-related increase in filamentous actin (F-actin) in Drosophila brains, 28 which is counteracted by prolongevity interventions. Critically, modulating F-actin levels in 29 aging neurons prevents age-onset cognitive decline and extends organismal healthspan. 30 Mechanistically, we show that autophagy, a recycling process required for neuronal homeostasis, 31 is disabled upon actin dysregulation in the aged brain. Remarkably, disrupting actin 32 polymerization in aged animals with cytoskeletal drugs restores brain autophagy to youthful 33 levels and reverses cellular hallmarks of brain aging. Finally, reducing F-actin levels in aging 34 neurons slows brain aging and promotes healthspan in an autophagy-dependent manner. Our 35 data identify excess actin polymerization as a hallmark of brain aging, which can be targeted to 36 reverse brain aging phenotypes and prolong healthspan. 37 Introduction dietary restriction or treated with rapamycin, two evolutionarily conserved approaches to lifespan 94 extension 38,39 , both show a reduction in F-actin in aged brains. To establish causal relationships, 95 we have identified multiple interventions targeting neuronal actin dynamics that can slow brain 96 aging and prolong healthspan. More specifically, we show that adult-onset, neuron-specific 97 inhibition of Fhos (Formin homology 2 domain containing ortholog), a FHOD class formin that 98 nucleates actin filaments 40 , improves cognitive function in aged flies and dramatically improves 99 multiple markers of organismal healthspan. Using both genetic and pharmacological 100 approaches, we show that excess F-actin polymerization leads to impaired autophagic activity 101 and the accumulation of dysfunctional mitochondria in the aged brain. Remarkably, we show 102 that treating aged animals with cytoskeletal drugs, to disrupt actin polymerization, can reverse 103 age-onset impairments in brain autophagy and improve cognitive performance. Finally, we show 104 that improvements in autophagy in the aged brain are necessary for the beneficial effects of 105 neuronal F-actin modulation. Together, our findings reveal neuronal dysregulation of actin 106 dynamics as a novel hallmark of aging, which can be targeted to restore autophagic activity, 107 improve brain function and prolong healthspan. 108 model. Using immunofluorescent microscopy (IF), we began by comparing wild type fly brains 117 collected from young flies (10 days) to those isolated from middle-age (30 days) and late-age (45 118 days) flies using phalloidin to stain for F-actin 15,17,37,41 . Remarkably, we detected a significant 119 increase in total F-actin levels when comparing young brains to middle-age and late-age brains 120 (Fig 1a,b). At increased magnification, we observed F-actin-rich rods accumulating in the aging 121 brain that were absent in young brains (Fig 1c,d). These observations were corroborated with an 122 additional control laboratory strain (Extended Data Fig1a,b). To further validate these findings, 123 we collected protein homogenates from the heads of young and aged flies for enzyme-linked 124 immunosorbent assays (ELISAs). Both middle-age and later-aged fly heads had significantly 125 more F-actin compared to young controls (Fig 1e). Interestingly, both cytoplasmic actin isoforms 126 expressed in Drosophila neurons, Act5c and Act42a 42 , were observed to increase 127 transcriptionally by quantitative polymerase chain reaction (qPCR) in aged head samples when 128 compared to two other genes considered to be expressed at a steady state: GAPDH and RPL32 129 (Extended Data Fig 1c, d). Although actin genes are widely considered to be 'housekeeping 130 genes', these findings argue that actin expression is dynamic with age in Drosophila brains. 131 To further extend our observations of increased F-actin polymerization in aged fly brains, 132 we used reporter lines expressing Act5c and Act42a tagged with green fluorescent protein (GFP). 133 Flies expressing Act5c-GFP in neurons showed an increase in Act5c with age that colocalized 134 with phalloidin staining (Fig 1f, Extended Data Fig 1e). Neuronal expression of Act42a-GFP 135 showed actin-rich structures in aging brains that were absent in young brains (Extended Data Fig 136 f,g). Interestingly, the distribution of Act42a-GFP in aging Drosophila brains showed a pattern 137 distinct from that of Act5c-GFP and phalloidin, suggesting a different distribution of this actin overall actin intensity in aged brains compared to young brains (Extended Data Fig h,i). The 140 pharmacological reagents cytochalasin D and latrunculin A are commonly used to depolymerize 141 actin filaments 43,44 . Feeding aged flies cytochalasin D (Extended Data Fig j,k) or latrunculin A 142 (Extended Data Fig l,m) for one week was sufficient to ablate the age-associated accumulation of 143 actin-rich rods in brains. Cumulatively, these observations support a model in which F-actin 144 polymerization increases in Drosophila brains with age. 145 To assess if F-actin polymerization in aged brains was reflective of aging health or if it 146 occurred universally with chronological age, we assessed flies from two widely studied lifespan 147 extension strategies. Dietary restriction (DR) and/or protein restriction is an evolutionarily 148 conserved approach to slow aging and promote lifespan 45 . Flies fed a low protein diet had a 149 significantly longer lifespan compared to flies provided a high-protein diet (Fig 1g). Using IF, 150 we observed actin-rich rods in the brains of flies on a rich diet at young middle-age (21 days 151 post-eclosion) that were absent in the brains of flies undergoing DR (Fig 1h,i). We next tested 152 the effect of treating flies with rapamycin, a small molecule that has also been shown to prolong 153 lifespan in evolutionarily diverse species via inhibition mTORC1 39 . Consistent with previous 154 observations 46,47 , feeding flies rapamycin significantly extended their lifespan compared to 155 vehicle-fed controls (Fig 1j). Remarkably, aged flies fed rapamycin had significantly fewer actin-156 rich rods in the brain compared to age-matched controls (Fig 1k,l). Together, these findings 157 suggest that age-associated F-actin polymerization in Drosophila brains reflects aging health and 158 can be delayed using prolongevity strategies. 159 160

Genetic targeting of neuronal F-actin extends organismal healthspan and lifespan 163
Since our observations found a correlation between aging and F-actin accumulation in the 164 brain, we next decided to test if targeting neuronal F-actin polymerization genetically could 165 affect organismal healthspan. We screened several genes related to actin and actin stabilization, 166 including actin isoforms, actin-binding proteins (ABPs), and actin assembly factors, and assessed 167 changes to age-related F-actin polymerization in the brain and to organismal lifespan. We found 168 that neuronal knockdown of Formin homology 2 domain containing ortholog (Fhos), 169 the Drosophila homolog of the FHOD sub-family of formins, had the most robust effect on these 170 parameters. Fhos promotes actin nucleation for filament assembly 40 . Using the pan-171 neuronal Elav-Gene-Switch (elavGS) driver line 48 , we expressed UAS-Fhos-RNAi in adult fly 172 neurons upon administration of the inducing agent RU486 in food. This system allows for cell-173 type specific induction of genetic constructs in a time-and dose-dependent manner. Flies in each 174 experiment come from the same parental crosses and undergo identical developmental 175 conditions. Phalloidin staining in aged brains revealed that neuronal Fhos-RNAi induction 176 abrogated the actin-rich rods observed in aged control brains (Fig 2a, b,c). We found that midlife 177 knockdown of the dominant actin isoform expressed in adult Drosophila neurons, Act5c, 178 similarly reduced the number of actin-rich rods in the aged brain (Extended Data Fig 2a, b,c). 179 Actin plays an essential role in neuron polarity and, consequently, function 49 . To assess 180 physiological brain function, we tested associative learning and memory using olfaction aversion 181 training 50 . Briefly, flies were conditioned to associate a neutral odor (3-octanol, OCT) with a 182 series of electric shocks. After one hour of rest, they were placed in a T-maze and allowed to 183 choose between OCT and a second neutral odor (4-methylcyclohexanol). Young flies avoided 184 the shock-associated OCT significantly more than aged flies (Fig 2c). Furthermore, aged flies expressing Fhos-RNAi in neurons showed a significantly better memory recall response than 186 uninduced age-matched controls (Fig 2c). Remarkably, treating aged flies for one week with the 187 actin destabilization drug cytochalasin D also significantly improved associative learning and 188 memory (Fig 2d). Hence, reducing F-actin polymerization in aged brains, both genetically and 189 pharmacologically, improved learning and memory in aged flies. 190 Changes in food intake can extend lifespan and delay age-related brain F-actin 191 polymerization (Fig 1g, h,i). To test if neuronal expression of Fhos-RNAi in adult flies affected 192 feeding behavior, we performed the consumption-excretion ("Con-Ex") feeding assay 51 . 193 However, we observed no differences in food consumption and excretion (Fig 2e). With 194 neuronal Fhos-RNAi induction correlating with a reduction in age-associated actin-rich rods in 195 the brain and an improvement in memory, we next decided to test if organismal lifespan and 196 additional parameters of health also improved. Remarkably, flies expressing Fhos-RNAi in 197 neurons showed a dramatically increased lifespan compared to controls (Fig 2f). Similarly, RNAi induction conferred an increase in spontaneous daytime activity with no detectable 205 nighttime restlessness in aged flies (Fig 2i,j). 206 Intestinal barrier dysfunction is an evolutionarily-conserved characteristic of aging 207 associated with systemic inflammation, frailty, and mortality 53 . To assess if neuronal knockdown of Fhos could prolong intestinal integrity, we performed the 'Smurf assay' 54,55 . In 209 agreement with prolonged lifespan and improved parameters of aging health, we observed a 210 delay in gut leakiness in aged flies expressing Fhos-RNAi in neurons (Fig 2k). 211 To confirm that the effects of neuronal Fhos and Act5c knockdown were not an artifact of 212 RNAi expression or RU486 administration, we generated flies expressing a construct of double-213 stranded RNA of GFP in neurons (elavGS>UAS-dsGFP). Importantly, induction of dsGFP did 214 not affect age-associated brain F-actin polymerization (Extended Data Fig 2e,f), memory and 215 learning (Extended Data Fig 2g), or feeding behavior (Extended Data Fig 2h). Furthermore, 216 providing RU486 to elavGS>UAS-dsGFP flies did not extend organismal lifespan (Extended 217 Data Fig 2i). Together, these findings indicate that neuronal knockdown of the actin nucleation 218 gene Fhos, as well as tsr and Act5c, can significantly delay parameters of aging and extend 219 organismal lifespan. 220 221 Age-associated neuronal F-actin polymerization impairs brain autophagy 222 Actin dynamics are essential in the biogenesis and transportation of most cellular 223 vesicles, including autophagosomes 33 . Defects in autophagy is considered a primary hallmark of 224 aging, resulting in impaired proteostasis and decreased organelle turnover 20,23,30,56 . 225 Furthermore, autophagy and vesicular trafficking defects have been identified in 226 neurodegenerative diseases, including Alzheimer's disease, Parkinson's disease, and 227 Huntington's disease 37,57 . With the changes observed to actin dynamics in the aging brain (Fig  228   1), we next sought to evaluate if interventions targeting age-associated F-actin polymerization 229 would affect autophagy. Using endogenous LC3/ATG8 as a marker of autophagy, we found an 230 age-related increase in autophagosomes reflective of reduced autophagic activity 53,58,59 (Fig 3a,b). Neuronal-specific knockdown of actin nucleation factor Fhos resulted in a dramatic 232 reduction in ATG8a accumulation, closely reflecting what was observed in young brains (Fig 233 3a,b). To further evaluate autophagic flux in the aging brain, we used a reporter line expressing 234 GFP-mCherry-ATG8a ("ATG8a-tandem") ubiquitously under the control of the endogenous 235 ATG8 promoter 60 . As autophagosomes fuse with lysosomes, GFP signal on the ATG8a tandem 236 protein is quenched due to its sensitivity to low pH. Remaining mCherry-only foci indicate 237 autolysosomal activity. When investigating the brains of young flies expressing ATG8a-tandem, 238 we observed a striking density of red-only puncta and a near absence green signal. Conversely, 239 aged brains showed a mixture of yellow and red-only puncta, with significantly fewer mCherry 240 foci indicative of fewer autolysosomes (Fig 3c,d). These findings suggest that young brains 241 display extensive autophagic flux that becomes stalled with age. When Fhos-RNAi was 242 expressed in the neurons of adult ATG8a-tandem flies, we observed fewer yellow puncta and 243 more red-only puncta in brains compared to aged controls, indicating recovery of autophagy (Fig 244 3c,d). 245 To complement our findings with ATG8, we tested additional readouts of protein 246 homeostasis (proteostasis) and autophagy. Decline in proteostasis is another major cellular 247 hallmark of aging 20,56 , and it has been well characterized that aged tissues in Drosophila 248 accumulate aggregates of ubiquinated proteins 29,31,32,46,58,61,62 . We observed that neuronal Fhos-249 RNAi induction significantly reduced age-associated protein aggregates in the brain (Fig 3e,f). 250 Ubiquinated proteins can be targeted for autophagic degradation by the adaptor protein 251 p62/SQSTM1, with an accumulation of p62 indicating reduced turnover and breakdown similar 252 to ATG8a accumulation 53,58 . Aged brains showed significantly more p62 puncta compared to young brains, and neuronal knockdown of Fhos also reduced the accumulation of p62 in aged 254 brains (Extended Data Fig 3a,b). 255 Next, we sought to test if a pharmacological intervention targeting actin polymerization 256 could improve readouts of brain autophagy. We had earlier found that one-week treatment at 257 midlife with cytochalasin D was sufficient to abrogate age-associated brain F-actin 258 polymerization (Extended Data Fig 1j,k). Here, we tested the effect of cytochalasin D on flies 259 expressing the ATG8a-tandem reporter. In a given cohort, brains were collected from young flies 260 (10 days), aged flies before treatment (37 days), and aged flies after 1 week of treatment with 261 cytochalasin D or vehicle from day 37 to day 44. Aged brains collected at day 37 and day 44 fed 262 vehicle showed significantly fewer red-only ATG8a-tandem puncta compared to young brains, 263 indicating a decline in autophagic flux. Remarkably, 1 week of cytochalasin D treatment in aged 264 animals significantly increased red-only ATG8a-tandem foci compared to both day 44 and, 265 critically, day 37 brains (Fig 3g,h). These findings indicate a reversal in age-related decline in 266 autophagic flux when treating animals with a pharmacological inhibitor of actin polymerization. 267 We next followed the same drug treatment paradigm while investigating proteostasis. Consistent 268 with what was observed using the ATG8a-tandem reporter, treatment of wild-type flies for 1 269 week with cytochalasin D was sufficient to reverse age-related accumulation of ubiquinated 270 proteins in the brain (Fig i,j). These findings imply that therapeutic targeting of age-associated 271 actin polymerization may reverse both cellular hallmarks of brain aging and improve brain 272 function. 273 One major role of autophagy is to mediate the turnover and clearance of damaged or 276 superfluous mitochondria 63 . Accordingly, we next sought to understand if age-associated F-actin 277 polymerization in the brain interfered more specifically with mitophagy and mitochondrial 278 function. To visualize mitophagic flux, we used the mito-QC reporter line that encodes a tandem 279 GFP-mCherry fusion protein targeted to the outer mitochondria membrane 64 . Mitochondria 280 degraded in acidic lysosomes (mitolysosomes) appear as mCherry-only puncta as GFP is 281 quenched. Consistent with our observations in autophagy, we found significantly more red-only 282 foci in the aged brains of flies with neuronal knockdown of the F-actin nucleation factor Fhos 283 compared to age-matched controls (Fig 4a, b). In agreement, midlife treatment of flies with 284 cytochalasin D resulted in significantly more mitolysosomes in the brain compared to vehicle-fed 285 controls (Extended Data Fig 4a, b). Consistent with reduced clearance of mitochondria, brains 286 from aged Drosophila showed an accumulation of mitochondrial content compared to young 287 adult controls (Fig 4c,d) as described 29 . Neuronal knockdown of Fhos resulted in a significant 288 reduction in mitochondrial content to an amount similar to that detected in young brains (Fig 289 4c,d). As an additional readout of brain mitochondrial content, we examined mitochondrial DNA assess function, we examined mitochondrial membrane potential using the potentiometric dye intensity compared to young controls, while targeting neuronal F-actin polymerization via Fhos 299 knockdown resulted in recovery of mitochondrial membrane potential (Fig 4e,f). These data 300 suggest that genetic targeting of F-actin polymerization in adult Drosophila brains results in 301 increased mitophagy, reduced age-associated accumulation of mitochondria, and improved 302 mitochondrial function. 303 Consistent with improved autophagy and mitophagy (Extended Data Fig 3g,h;4a,b), 304 treating middle-aged flies with cytochalasin D significantly reduced the accumulation of 305 mitochondrial content in aged brains (Fig 4g,h). Notably, less brain mitochondrial content was 306 detected at day 44, after 1 week of cytochalasin D treatment, compared to brains collected from 307 flies of the same cohort at day 37 before treatment (Fig 4g,h), consistent with a reversal of this 308 hallmark of brain aging. Mitochondrial content was reduced in aged brains at multiple 309 concentrations of the drug (Extended Data Fig 4i,j). Treating aged flies for 1 week with an 310 independent actin depolymerization agent, latrunculin A, also resulted in reduced mitochondrial 311 content in aged brains in a dose-dependent manner (Extended Data Fig 4k,l). Importantly, TMRE 312 staining revealed that pharmacologically depolymerizing age-associated F-actin by cytochalasin 313 D significantly improved mitochondrial homeostasis in the aged brain (Fig 4i,j). Together, these 314 findings indicate that genetic and pharmacological targeting of age-associated F-actin 315 polymerization improves mitophagy and mitochondrial homeostasis in Drosophila brains. 316 317

Neuronal reduction of F-actin polymerization slows aging via autophagy 318
Our findings indicate that age-associated F-actin polymerization in the brain disrupts 319 autophagy, mitochondrial homeostasis, and proteostasis. Next, we set out to determine whether 320 the beneficial effects of decreasing actin polymerization on brain and organismal aging are due to improved neuronal autophagy. First, we observed that Fhos-mediated modulation of F-actin 322 levels during brain aging proceeds in an autophagy-independent manner. To block the autophagy 323 pathway, we targeted the expression of Atg1 (Autophagy-related 1, the Drosophila homolog of 324 mammalian ULK1). This Ser/Thr protein kinase regulates the initiation of the formation of the 325 autophagosome 65 . Concomitant knockdown of neuronal Atg1 and Fhos in elavGS>UAS-Atg1-326 RNAi,UAS-Fhos-RNAi flies resulted in reduced age-associated actin-rich rods in the brain (Fig  327   5a ,b). In contrast to extended lifespan with neuronal knockdown of Fhos alone (Fig 2f), induced 328 elavGS>UAS-Atg1-RNAi,UAS-Fhos-RNAi flies showed no difference in lifespan compared to 329 vehicle-fed controls (Fig 5c). Consistent with this finding, we detected no difference in intestinal 330 barrier integrity in flies with neuronal knockdown of both Fhos and Atg1 (Fig 5d). Hence, 331 although age-associated F-actin polymerization in the brain was reduced with neuronal Fhos 332 knockdown, resulting health and lifespan benefits were dependent on autophagy. 333

Furthermore, concomitant knockdown of neuronal Atg1 and Fhos in elavGS>UAS-Atg1-334
RNAi,UAS-Fhos-RNAi flies showed an age-associated accumulation of mitochondrial content in 335 the brain (Fig 5e,f). Improvements to brain mitochondrial function that were found with 336 disrupting age-associated F-actin polymerization were also lost with neuronal Atg1 knockdown 337 (Fig 5g,h). Together, these results indicate that improved mitochondrial homeostasis associated 338 with reduced brain F-actin polymerization, like improvements to organismal health and lifespan, 339 are dependent on autophagy. Cumulatively, these data are consistent with a model in which age-340 associated F-actin polymerization in Drosophila brains disrupts autophagy and, thereby, drives 341 paradigms of brain and organismal aging. 342

Discussion 345
Actin filaments show a loss of stability and deterioration in aged yeast cells 66 and in 346 multiple non-neuronal tissues of aged C. elegans 5 . In contrast, we find a striking age-related 347 increase in F-actin and actin-containing rods in the fly brain, which contribute to brain aging and 348 drive organismal health decline. Our findings are consistent with studies showing excess actin 349 stabilization drives neurotoxicity in models of Alzheimer's disease (AD) and related tauopathies 350 15,16 and Parkinson's disease (PD) 17,67 . As advanced age is a major risk factor for sporadic forms 351 of both AD and PD, actin hyperstabilization may be a shared pathogenic mechanism of age-onset 352 neurodegeneration. Future work will be required to determine the precise cellular mechanisms 353 that lead to excess F-actin and the formation of actin-rich rods in the aged brain. The formation 354 of rodlike inclusions (rods) composed of actin and actin assembly-regulatory proteins has been 355 shown in cultured neurons in response to various stressors, including oxidative stress and 356 mitochondrial energetic impairments 68-70 . As mitochondrial dysfunction is a major hallmark of 357 brain aging, it is reasonable to suggest that there could be a 'vicious cycle' whereby 358 mitochondrial dysfunction and actin hyperstabilization drive brain aging. In addition, it has been 359 shown that de novo actin polymerization is required to form model Hirano bodies 71 . In the 360 context of our study, it is interesting to speculate that actin nucleation is driving excess actin 361 stabilization, and actin-rich rod formation, in the aged brain. The most pronounced phenotype 362 that we observe with respect to lifespan extension is mediated by neuron-specific RNAi of Fhos, 363 which shares the capacity of other formins to nucleate and bundle actin filaments 40 . We also 364 observe increased expression of actin transcripts in the aged brain. Hence, it is possible that the 365 overall increase in F-actin in the aged brain is due to a combination of increased actin expression accumulation to be an important hallmark of brain aging, which should be considered in the 368 context of existing hallmarks of aging 21 . 369 One of the key hallmarks of brain aging which drives age-onset pathology is disabled 370 autophagy 21,72 given its demonstrated capacity to remove aggregated proteins and damaged 371 organelles linked to neurodegenerative disease. Hence, identifying interventions that restore 372 autophagy in the aged brain is a promising therapeutic avenue, but that depends on the step in the 373 process that is impaired and the nature of the impairment. Unfortunately, our understanding of 374 the nature of autophagy impairments in the aged brain and in most cases of neurodegeneration is 375 limited, making it uncertain whether the same type of intervention is likely to work in all 376 diseases and at all stages of those diseases. Indeed, it has been proposed that dysfunctional 377 autophagy in aged animals, linked to blockage of autophagy at a late stage, may contribute to 378 age-onset health decline 30,73 . Hence, it is possible that interventions that induce early stages of 379 autophagy may not prove effective when targeted to aged animals. In this study, we show that 380 inhibiting F-actin polymerization in aged neurons prevents the age-related loss of autophagic 381 activity in the brain. Moreover, we show that treating middle-aged flies with an actin 382 polymerization inhibitor, cytochalasin D, restores brain autophagy to pre-treatment levels and 383 leads to lower amounts of protein aggregates and mitochondrial content in the brain than before 384 the treatment began. Interestingly, our data indicate that the accumulation of F-actin in the aged 385 brain is not due to impaired autophagy. We show that the ability of neuronal Fhos inhibition to 386 prevent F-actin accumulation in the aged brain is autophagy-independent. However, reducing F-387 actin levels in the aged brain while disrupting autophagy fails to improve mitochondrial 388 homeostasis in the aged brain or prolong healthspan. Hence, our current working hypothesis is that disrupting actin polymerization in aging neurons prolongs healthspan via improvements in 390 brain autophagy. 391 The functional significance of abnormal F-actin accumulation in cellular health and 392 disease is highlighted by clinical data studying neurodegenerative diseases. There is an 393 emerging consensus that F-actin containing intracellular inclusions disrupt neuronal function and 394 are a likely cause of synaptic loss without neuronal loss, as occurs early in dementias 74,75 . Here, 395 we show that disrupting actin polymerization in the brains of middle-aged animals robustly 396 improves a well-established paradigm of olfactory learning 50,76 ; the ability of flies to associate 397 an odor with an aversive stimulus. Our findings reveal that inhibiting actin polymerization in 398 aged animals can slow or even reverse aspects of brain aging. It is important to consider, 399 however, that it is unlikely that disrupting actin polymerization in every cell and tissue-type 400 would promote organismal health. It is likely that actin polymerization is essential to cell and 401 tissue homeostasis in numerous contexts. Hence, in order to translate these findings to benefit 402 human health, future work could focus upon identifying cell-type and tissue-specific approaches 403 to inhibit actin polymerization in aged organisms.   185-199 (1987).

Olfactory training 675
Aversion training was performed as described in 50 using a system from MazeEngineers 676 (Conduct Science). Briefly, flies were exposed to a neutral odor (3-octanol) by air pump in a 677 training chamber for one minute under low red light conditions. They were then exposed to the 678 odor in a series of twelve 60-V shocks for 1.25 seconds followed by rest for 3.75 seconds for a 679 total of one minute. Flies recovered for one hour before being placed in a T-maze with the trained scent on one side and a second neutral scent (4-methylcyclohexanol) on the other side of 681 the maze. After two minutes of exploration under dim red light conditions, flies in either 682 chamber of the maze were counted. 683 684

Intestinal barrier dysfunction (Smurf) assay 685
Intestinal integrity assays were performed as previously described 55 . Flies were aged to the 686 indicated time points with standard RU-or RU+ food as indicated. To conduct the "Smurf" 687 assay, flies were then transferred to new vials containing standard medium with 2.5% wt/vol 688

Consumption-Excretion (Con-Ex) assay 707
Con-EX assays were performed as previously described 51 . Adult flies were transferred to new 708 empty vials (10 flies per vial with a total of 6 vials) and fed from feeder caps containing standard 709 medium with 2.5% wt/vol F&D blue dye # 1 for 20h at 25 ℃. Feeder caps were discarded at the The sample was diluted in sample diluent at a ratio of 1:20 before being loaded to a microplate 736 that was pre-coated with an antibody specific for F-actin. A biotinylated secondary antibody was 737 added to the microplate and subsequently incubated with HRP-avidin followed by peroxidase 738 substrate. Concentrations of F-actin were determined using a Epoch BioTek microplate reader 739 and compared to a serially diluted standard provided by the manufacturer. 740 741

Statistics 742
GraphPad Prism 9 (GraphPad Software, La Jolla, CA, USA) was used to perform statistical 743 analysis and graphically display data. Significance is expressed as p values as determined by 744 two-tailed, unpaired, parametric, or non-parametric tests as indicated in figure legends. When 745 comparing two groups, unpaired t-tests were used when data met criteria for parametric analysis 746 and Mann-Whitney tests were used for non-parametric analysis. To compare more than two 747 groups when parametric tests were appropriate, one-way ANOVAs with Tukey's multiple 748 comparisons tests were performed. To compare more than two groups sampled from a Gaussian To analyze more than two groups when data did not meet requirements for parametric tests, 751 Kruskal-Wallis tests with Dunn's multiple comparisons post hoc tests were used. When 752 performing grouped analyses with multiple comparisons, two-way ANOVAs with Šídák's 753 multiple comparisons test were performed. Bar graphs depict mean ± standard error of the mean 754 (SEM). The number (n) of biological samples used in each experiment can be found in figure  755 legends. Log-rank (Mantel-Cox) tests were used to compare survival curves. No statistical 756 methods were used to pre-determine sample sizes but our sample sizes are similar to those 757 (c) Immunostaining of brains at 63x magnification from young (10-day-old) and aged (30-dayold and 45-day-old, as indicated) Canton S flies, showing actin-rich rods (red channel, Phalloidin). Scale bar is 5 µm. Accompanying diagram indicates brain region where imaging was conducted.
(g) Survival curves of Canton S flies given a rich diet (5.0% yeast extract) versus those undergoing dietary restriction (DR, 0.5% yeast extract) from day 4 post eclosion onwards.
(b) Quantification of actin-rich rods by phalloidin staining per 1mm 2 area of brain optic lobes as shown in (a). n = 3 flies per condition. *p = 0.0249; unpaired t-test.
(f) Quantification of actin-rich rods in brains as shown in (d). n = 4-8 flies per condition, as indicated. *p = 0.0259, **p = 0.0075, unpaired t-tests. (g) Performance index in olfactory aversion training in 37-day-old elavGS>UAS-dsGFP flies with or without RU486-mediated transgene expression from day 5 onward, assessed by the number of flies avoiding a shock-associated scent versus the total number of flies participating in the assay. ns = non-significant, *p (young vs aged RU-) = 0.0294, *p (young vs. aged RU+) = 0.0162, one-way ANOVA, Tukey's multiple comparisons test.