Polygenic adaptation to overnutrition reveals a role for cholinergic signaling in longevity

Overnutrition by high-sugar (HS) feeding reduces both the lifespan and healthspan across taxa. Pressuring organisms to adapt to overnutrition can highlight genes and pathways important for the healthspan in stressful environments. We used an experimental evolution approach to adapt four replicate, outbred population pairs of Drosophila melanogaster to a HS or control diet. Sexes were separated and aged on either diet until mid-life, then mated to produce the next generation, allowing enrichment for protective alleles over time. All HS-selected populations increased their lifespan and were therefore used as a platform to compare allele frequencies and gene expression. Pathways functioning in the nervous system were overrepresented in the genomic data and showed evidence for parallel evolution, although very few genes were the same across replicates. Acetylcholine-related genes, including the muscarinic receptor mAChR-A, showed significant changes in allele frequency in multiple selected populations and differential expression on a HS diet. Using genetic and pharmacological approaches, we show that cholinergic signaling affects Drosophila feeding in a sugar-specific fashion. Together, these results suggest that adaptation produces changes in allele frequencies that benefit animals under conditions of overnutrition and that it is repeatable at the pathway level.

The adaptation of complex phenotypes arises from a composite of physiological and 28 morphological traits under simultaneous selection (1,2). Aging is one complex trait of interest 29 that is accompanied by pathophysiology across a wide variety of outcomes (3). In many cases, 30 long-lived individuals also display increases in the healthspan, and age is a risk factor for a 31 range of pathophysiologies (4-7). Although it is generally true that the longest-lived in a 32 population display healthier quantitative traits than short-lived, the correlation between traits can 33 be weak and the mechanisms that underlie each correlation are not well understood (8,9). In 34 addition, such traits as aging are highly polygenic, with many loci of small effect sizes 35 contributing to the phenotype. The polygenic nature of complex traits creates challenges for 36 researchers attempting to map the genetic basis of said traits (1,2,10-13). One approach to 37 address this question that has met with success has been evolve-and-resequence (E&R) 38 studies using model organisms such as Drosophila melanogaster (abbreviated as Drosophila 39 here). These studies have used a variety of selection pressures on life-history traits such as 40 increased longevity (11,14) and late-life reproduction (15)(16)(17)(18)(19)(20)(21) to identify loci that increase 41 lifespan. An advantage of E&R studies in uncovering physiologically relevant adaptation is the 42 ability to leverage replicate populations from a common genetic background. Parallelism across 43 replicates thus increases confidence in potentially causal mechanisms at the SNP, gene, and/or 44 pathway levels.
4 131 DNA isolation and sequencing 132 At generations 0, 5, 10, and 15, pools of 120 flies per replicate were collected and frozen at -133 80 o C until processing (48,49). Frozen flies were homogenized in PBS (pH 7.4) and DNA was 134 extracted per the manufacturer's instructions using Qiagen DNeasy Blood and Tissue kit 135 (Qiagen, Germantown, MD, USA Cat. #69504). Samples were then subjected to ethanol 136 precipitation as previously described (50). Total DNA was quantified using Qubit 2.0 fluorometer 137 HS Assay kit (Life Technologies/ThermoFisher Scientific, Waltham, MA, USA Cat. #32850). 138 Libraries were constructed and sequencing was performed by the Genome Technology Access 139 Center at the McDonnell Genome Institute (Washington University in St. Louis, Missouri). DNA was 140 sequenced on the Illumina NovaSeq 6000 platform and mapped to BDGP release 6.27. Generation 141 0 had an average number of mapped reads of over 332,000,000 at an average 88% mapped reads. 142 Generation 0 had an average read depth of >320x. Generation 10 had an average number of 143 mapped reads over 320,000,000 and an average 92% mapped reads to give a sequencing depth of 144 >240x across all populations. 145 146 RNA isolation and sequencing 147 At generation 10, flies were aged on 1M sucrose food media for three weeks and frozen in 148 groups of 20 per replicate at -80°C until processing. Flies were then homogenized by electric 149 mortar and pestle in Ribozol (VWR Cat. N580-100ML). RNA was treated with DNase (VWR Cat. 150 PIER89836) and then extracted per the manufacturer's instructions using the Qiagen RNeasy 151 Mini Kit (Qiagen, Germantown, MD, USA Cat #74104). Total RNA was quantified using Qubit 152 2.0 fluorometer HS Assay kit (Life Technologies/ThermoFisher Scientific, Waltham, MA, USA Cat. # 153 10210). Libraries were constructed after FastSelect (Qiagen, Germantown, MD, USA Cat. 154 #334385) rRNA subtraction and sequencing was performed by the Genome Technology Access 155 Center at the McDonnell Genome Institute (St. Louis, Missouri) using an Illumina HiSeq platform. All 156 samples produced over 11,000,000 mapped reads with an average of 12,360,000 mapped reads 157 and an average 93% mapped reads. RNA reads were aligned to the Ensembl release 76 genome 158 using STAR (v. 2.5.1a) (51). 159 160 Gene identification 161 To identify SNPs that were likely to have changed in allele frequency as a result of selection to 162 high-sugar, we used Popoolation2 (48). Popoolation2's Fisher's exact test was used to identified 163 SNPs significantly differentiated in allele frequency between selected populations at generation 164 10 and the starter population at generation 0 with a Bonferroni corrected P value of 0.05 (raw P 165 vulaues of 0.05 adjusted to ~10 -8 . To eliminate the effects of selection to other sources (e.g. lab 166 rearing conditions) we further identified SNPs that differentiated between the selected 167 populations at generation 10 and the control populations at generation 10 with a Bonferroni 168 corrected P value of 0.05. Genes of interest were called from SNPs if at least one significant 169 SNP was within an annotated gene or within 2 kb upstream or downstream of a gene. In 170 combination with the DNA analysis after selection, we conducted a differential expression 171 analysis on generation 10 RNA-seq data. Gene counts were normalized using the TMM method 172 in the R package EdgeR (52). Differential expression analysis was then conducted using the R 173 package Limma and Benjamini-Hochberg procedure was used for multiple testing correction . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made To understand more about the target gene mAChR-A and its function, we performed a network 189 analysis to learn its relationship with other genes that have been previously studied. First, we 190 conducted a gene ontology (GO) analysis using gProfiler to select the genes that were 191 significant and associated with GO terms of interest, such as DNA binding, and ribosome 192 biogenesis (55). 360 genes were selected from the male group. We then used GeneMANIA (56) 193 to identify 8 genes that were directly linked to mAChR-A and utilized Cytoscape (57) to produce 194 a network plot which contains an overall network and the types of interactions among these 9 195 genes. 196 197 Feeding assay 198 Feeding assays were performed similarly to those described in (58). Flies were aged for 1 or 3 199 weeks on control and experimental food media then transferred to corresponding media 200 supplemented with 2% FD&C Blue #1 for 2 hrs (11am-1pm) and frozen immediately at -80°C 201 until processing. Flies were homogenized in groups of 4 with mortar and pestle and 202 supernatants were read at 630 nm on a VersaMax microplate spectrophotometer (Molecular 203 Devices). Statistical analyses were conducted using a two-tailed Student's t-test or one-way 204 ANOVA on Prism software (v. 9.4.0). 205 206 Results 207 208 Experimental selection to a high-sugar-diet increased longevity on control and selective diets 209 To identify genes that increase lifespan on a high-sugar (HS) diet, we undertook an 210 experimental evolution approach. Large census populations (~6000 flies) were reared on control 211 (LS) diets, then aged on LS or HS in single-sex cages until approximately half the flies in the 212 selected population had died (Fig 1; see methods for more detail). The starting population was a 213 diverse outbred population created by round-robin mating of wild-caught isofemale lines 214 (citation; see methods for more details). Replicate populations were paired as shown (Fig 1) and 215 matched in husbandry and developmental timing. 216 . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted June 14, 2023. ;https://doi.org/10.1101https://doi.org/10. /2023 217 218

226
In response to 10 generations of selection to adult feeding on a high-sugar diet, all Selected 227 populations exhibited a significant increase in longevity relative to both generation 0 (G0) and to 228 the paired control population when challenged with the selective HS ( Fig. 2A-B,E-F). 229 Interestingly, selection to HS adult feeding also extended lifespan in most populations on the 230 non-selective, LS diet. It is also notable that all control populations showed an increase in 231 lifespan on LS compared to generation 0, suggesting that even at a modest 10-15% mortality 232 rate selection for increased lifespan occurred ( death on respective diets. The significance corresponding to number of symbols is as follows: *P < 0.5, **P < 0.01, 242 ***P < 0.001, ****P < 0.0001 by log-rank Mantel-Cox. # represents the significance for generation 10 vs generation 0. 243 $ represents the significance for Selected population vs Control population at generation 10.

245
Interestingly, the magnitude of changes in lifespan were sex-specific, with Selected males 246 exhibiting greater increases in survival on HS (1.74-fold increase in median day of death 247 compared to the control 1.1-fold increase). By contrast, selected females exhibited an average . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made between the S1-C1 pairing, 5,574 between S2-C2, 10,111 between S3-C3, and 20,001 between 279 the S4-C4 pairing. SNPs were then further filtered to those within a gene or within 2 kb 280 upstream or downstream of a gene. While there was a sizable proportion of genes with overlap 281 between at least two selected populations (21%), there still existed a sizable proportion of 282 genes unique to each evolved population (79%) (Fig 3A). Interestingly, there was a much 283 stronger overlap in significant gene ontological (GO) categories as compared to the individual 284 genes (Fig 3B,C). This suggests strong evidence for parallel evolution in the pathways that are 285 involved but that unique genes within those pathways responded to selection in the different 286 replicate populations. One such class of neuronal categories that were consistent across 287 populations included learning/memory, generation of neurons, neuronal development, GPCR 288 signaling, and behavior (-log 10 (P) > 7.6 for all populations, Fig  (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted June 14, 2023. ; https://doi.org/10.1101/2023.06.14.544888 doi: bioRxiv preprint 292 293 294 overlap of significant identical genes (purple lines) and genes with the same ontology tag (blue lines). Terms with P < 301 0.01 and enrichment score >1.5 were counted as significant categories (C) Heatmap with hierarchical clustering of 302 significant ontological terms identified in each respective population pair. Heatmap coloring depicts significant P 303 values. Terms with P < 0.01 and enrichment score >1.5 were counted as significant categories.

305
Dynamic gene expression profiles observed across selected population pairs 306 We next measured changes in gene expression between selected and control populations after 307 10 generations of selection. Because expression profiles differed so much between males and 308 females, we analyzed these datasets separately ( genes were identified in a pair-wise fashion, comparing each Selected population to its Control 310 population partner. Genes with false-discovery rate adjusted P values less than or equal to 0.05 311 were considered differentially expressed. Transcriptomes differed between Selected and 312 Control populations when aged on a HS diet for 3 weeks (Fig 4A-B). Clusters of co-regulated 313 genes in each sex had similar expression patterns, consistent with what would be expected 314 from coordinated regulation of signaling pathways. With the exception of the S4-C4 pairing, . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted June 14, 2023. ; https://doi.org/10.1101/2023.06.14.544888 doi: bioRxiv preprint

315
females exhibited a greater number of significant alterations in their transcriptome from their 316 paired population. (S2 Table). As was the case for allele frequencies, the fewest differences in 317 expression were seen between populations S2 and C2. Using principal component analysis 318 (59), we clustered expression data from HS-fed flies and noted that there is modest clustering in 319 expression profiles among the selected populations (S8 Fig). Among females, S3 and C3 had 320 the most pronounced differences for the genes in PC1 (S8 Fig) and among males, S4 and C4 321 had the best separation over PC1, consistent with the large improvements in lifespan seen in 322 the two Selected populations. These scores plots highlight the diversity among population pairs 323 and suggest that the Selected populations exhibit similar expression trends, but limited direct 324 overlap in specific DE transcripts, as was observed for our gene-level analyses ( Fig 3A). Having 325 identified an overabundance of SNPs in neuronal genes, we next assessed these pathways in 326 greater detail. 327 328 329 (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made Given the prevalence of reduced cognition upon aging and the allele frequency differences 346 associated with cognition and neuronal function between the Selected and Control populations, 347 we assessed the expression of genes in this category ( Fig 4C). Several cholinergic signaling 348 genes exhibited statistically significant allele frequency changes and differential expression 349 between paired Selected and Control populations. Both nicotinic and muscarinic receptors were 350 affected, and changes in these and acetylcholine regulators were observed across populations 351 (S1 Table, S9 Fig). We thus assessed the expression of genes associated with neuronal 352 function in the fly that were differentially expressed in at least one population. Here we observed 353 patterns of differential expression in neuronal genes, with the muscarinic acetylcholine receptor 354 mAChR-A standing out in particular ( Fig 4C). This gene, which exhibited allele frequency 355 differences in four comparisons, was down regulated in females in two of the Selected 356 populations and in males in three of the Selected populations (S1 Table, Fig 4C-D). The 357 mAChR-A gene encodes a G protein-coupled receptor (GPCR) orthologous to the human 358 stimulatory muscarinic acetylcholine receptors M1, M3, and M5, which trigger intracellular 359 calcium release upon ligand binding (60-62). mAChR-A is expressed strongly and specifically in 360 the male and female adult brain (63). This gene was also found in PC1 for differential 361 expression in both males and females (S8 Fig). Interestingly, several genes that exhibited 362 similar expression patterns in our study have been previously described as having co-363 localization, co-expression, or shared protein domains with mAChR-A ( Fig 4D). Several of these 364 genes have been shown to be expressed in the brain and control neural physiology and 365 behavior (37,64-69). These genes showed similar trends of expression across Selected and 366 Control populations as mAChR-A, typically exhibiting lowered expression in the Selected 367 populations ( Fig 4D). Taken together, these data suggest that variation in mAChr-A and related 368 genes contributes to adaptation to the HS diet and led us to measure behaviors that may have 369 been beneficial to lifespan on a HS diet. 370 371 372 mAChR-A alters lifespan and feeding behavior in a sugar-specific manner 373 To explore potential links between cholinergic signaling and longevity in relation to HS feeding, 374 we selected mAChR-A because of its interesting differential expression pattern (Fig 4C-D) and 375 changes in allele frequency in multiple populations (S1 Table). This receptor was targeted with 376 RNA interference in the brain by crossing syb-GAL4 with UAS-mAChR-A RNAi , hereafter referred 377 to as "mAChR-Ai". The control was a cross between the driver and the insertion site control 378 genotype. Both the control and mAChR-Ai larvae were reared on a 5% dextrose-cornmeal-379 yeast-agar media and aged as adults on 0.15M or 1M sucrose. To address the role of the 380 muscarinic acetylcholine receptor in longevity, we first measured lifespan. There was a highly 381 significant 0.16 fold decrease in lifespan in mAChR-Ai males, but no significant difference was 382 observed in transgenic females aged on the HS diet (Fig 5A,B) or for either mAChR-Ai males or 383 females aged on 0.15M sucrose (Fig 5C,D), compared with the control genotype. This suggests . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted June 14, 2023. ; https://doi.org/10.1101/2023.06.14.544888 doi: bioRxiv preprint 384 that mAChR-A signaling is important for survival during HS feeding and operates in a diet-385 specific and sex-specific fashion.

393
Because the CNS controls feeding behavior and previous studies have shown HS affects 394 feeding (64,65), which in turn can have profound effects on longevity, we measured the feeding 395 rate of control and mAChR-Ai flies using a 'Con-Ex' assay (58). Control and mAChR-Ai animals 396 were aged as adults on 0.15M or 1M sucrose for 3 weeks before being transferred to media 397 supplemented with 2% FD&C Blue #1 dye for 2 h. We again observed diet-and sex-dependent 398 effects, as mAChR-Ai significantly reduced feeding only in male flies aged on HS while there 399 were no significant effects on consumption in control-reared males or in females on either diet 400 (Fig 5E,F). This result might be expected given the higher expression of mAChR-A in males, . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted June 14, 2023. ; https://doi.org/10.1101/2023.06.14.544888 doi: bioRxiv preprint 401 compared with females, in both our whole-animal expression data and public adult brain 402 expression data from Dr. Julian Dow and colleagues (S2 Table, (63)). 403 404 405 atropine produced sex-specific and diet-specific effects, and atropine also produced different 420 results in the Control and Selected populations. The largest effect on lifespan observed in 421 Control population C3 males, where it exhibited a 0.125-fold reduction in median day of death ( 422 P= 0.0009) on the HS diet (Fig 6A, atro) while the corresponding population S3 males were not 423 significantly affected by atropine supplementation (P = 0.53, Fig 6A). Again, this tracked with 424 expression, as the largest effect of atropine was observed in the flies with the highest 425 expression of mAChR-A, population C3 males. The opposite effect was observed in the females 426 where population S3 females exhibited a modest 0.06-fold increase in median day of death (P = 427 0.039, Fig 6B) while the Control C3 females were not significant affected ( P= 0.09, no fold 428 increase, Fig 6B). Because mAChR-Ai flies exhibited a feeding phenotype, we examined the 429 effects of atropine on feeding rates (58). To measure the effects of atropine on feeding, mated 430 adult flies were aged for 3 weeks on +/-atropine HS and LS food. Atropine did not have a 431 significant effect on feeding for many of the comparisons on HS diets (Fig 6C-F), although a 432 significant reduction in feeding was observed in Control population C3 females ( Fig 6D). 433 Interestingly, atropine eliminated the dramatic difference in feeding between population S3 and 434 C3 females on HS (Fig 6D). These sex-dependent results of atropine on feeding of the HS diet 435 were consistent with differential expression of its target receptor, mAChR-A, in males from 436 populations S3 and C3 (fed on the HS diet, compared to control-fed males, as shown on the 437 heatmap in Fig 4C). 438 439 To further explore potential roles for muscarinic signaling in feeding, we tested atropine on the 440 control (LS) diet. A more compelling trend toward reduced feeding was apparent in atropine-fed 441 flies on the control diet (Fig 6E,F). Females from both populations and males from population 442 C3 all reduced intake after atropine supplementation (Fig 6E,F). As for longevity, there was no 443 effect of atropine on feeding in S3 males (Fig 6E). Females from population S3, who typically 444 reduce feeding on the HS diet, no longer did so when atropine was present ( Fig 6D,F P =  445 0.0003 via one-way ANOVA). Taken together, our data are consistent with a diet-and sex-. CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made

465 466
. CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted June 14, 2023. ; https://doi.org/10.1101/2023.06.14.544888 doi: bioRxiv preprint

467
Discussion 468 469 In this study, we used an evolve and resequence (E & R) approach to identify key genes and 470 pathways in age-and overnutrition-associated pathophysiology phenotypes. Laboratory 471 selection prompted a highly polygenic and sex-biased response across replicate populations. 472 There was evidence for parallel evolution at the levels of biological processes, pathways, and 473 gene families, although there was limited overlap in the SNPs and individual genes identified. 474 Genes involved in cholinergic signaling, as well as a number of potentially interacting genes, 475 exhibited significant differences in allele frequency and/or expression across selected 476 populations. We verified one such gene with both genetic and pharmacological manipulations. A 477 knockdown of mAChR-A reduced male lifespan and consumption behavior in a sugar-specific 478 fashion. Furthermore, flies treated with atropine, an antagonist of muscarinic acetylcholine 479 receptors, also showed sex-specific alterations in lifespan and feeding. These results suggest 480 that our E & R approach was successful at identifying causal variants contributing to lifespan 481 under the conditions used. 482 483 Increases in lifespan 484 Evolve and resequence studies in Drosophila have been successful at producing rapid 485 responses to selection and at mapping the genetic basis for a variety of complex traits 486 (11,14,(17)(18)(19)(71)(72)(73)(74)). In the current study, we observed a robust phenotypic response to 10 487 generations of adaptation on two food media that correlated with polygenic changes in the 488 genome and differential expression (Fig 2-4). By generation 10, all Selected populations 489 exhibited significantly increased lifespan on both HS and LS diets, while Control populations 490 had increased lifespan primarily on the LS (Fig 2, S1- survival on HS food, despite being high-sugar naïve, after 10 generations (Fig 2). These data 497 suggest that loci contributing generally to longevity may have undergone genomic changes in 498 C1 and C4 that extend male lifespan on both diets, although the effect is attenuated on HS. 499 Given these overlaps, a potential future avenue may be to explore loci linked with longevity 500 across diverse environmental conditions. 501 502 Evidence for parallel evolution 503 Understanding whether adaptation is predictable and can proceed via parallel evolution from 504 standing variation is a major question in evolutionary genetics. Our results support the 505 hypothesis that the predictability of evolution depends upon the hierarchical level being tested 506 (79). All Selected populations showed significant increases in lifespan on HS as compared to 507 Controls, suggesting parallelism at the level of overall fitness. Selected populations also showed 508 a marked overlap in ontological categories based on significant changes in allele frequencies, 509 providing support for parallel evolution at the levels of biological processes, organ systems, 510 signaling cascades, and gene families. Perhaps most striking was the prevalence of CNS genes . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted June 14, 2023. ;https://doi.org/10.1101https://doi.org/10. /2023 511 associated with neuronal development, behavior, and synaptic signaling. We observed a 512 diverse array of acetylcholine receptors and related cholinergic genes with significant changes 513 in allele frequencies in all of our Selected populations (S1 Table). All Selected populations had 514 similar changes in the expression of genes functionally related to mAChR-A (Fig 4D). 515 Considering these parallels, it may be surprising to note that there was little evidence of parallel 516 evolution at the polymorphism level. Only five genes overlapped across Selected populations 517 ( Fig 3A); although each gene did share multiple polymorphisms with significant changes in allele 518 frequencies across all four population pairs (P < 1.0 -10 for all Selected populations). Further, 519 some, but not all, Selected populations exhibit an advantage over the Control on the LS diet (Fig  520  2, S1-4 Fig). This, coupled with the modest genetic overlap between Selected populations, 521 suggests that the populations evolved unique solutions to our selective pressure at the 522 individual gene and locus levels but many of the adaptive changes influenced similar biological 523 processes. Given the rapid, high mortality rate in Selected populations at early generations, it is 524 perhaps no surprise that the populations diverged at the genomic level. Beneficial alleles, 525 especially for a highly polygenic trait like lifespan, might easily be lost due to chance under 526 strong selection. These data are consistent with other studies showing that stochasticity plays 527 an important role (79-81), especially in populations with extensive standing variation (79,82,83). 528 Interestingly, among the neuronal GO terms, genes associated with development were the most 529 enriched in our SNP gene lists despite our selective pressure being applied exclusively to 530 adults. Taken together, these data suggested to us that the brain might have been particularly 531 responsive to our selective pressure, whereas shifts in metabolic flux may have complex or 532 antagonistic effects that make it difficult to achieve overall benefits to development, lifespan, 533 and healthspan. 534 535 mAChR 536 One of the genes associated with learning, memory, and synaptic transmission, mAChR-A, 537 provided strong evidence for parallel evolution as it showed significant differences in allele 538 frequencies in multiple Selected populations as well as reduced expression in Selected males 539 from population S3, compared with Control population C3 ( Fig 3A, Fig 4C, S1 Table). A network 540 analysis of one cluster of differentially expressed genes in our male dataset revealed genes 541 connected to mAChR-A through previously established co-expression or co-localization. Like 542 mAChR-A, expression of these genes decreased in males across all of our Selected 543 populations. Both polycomb-like (Pcl) and single-minded (sim) have been shown to be involved 544 in behavior likely to be relevant to lifespan and healthspan during HS overnutrition. The gene 545 sim has been associated with the development of the midline cells of the embryonic nerve cord 546 and the brain central complex (67-69) and sim mutants exhibit defective locomotion behavior 547 (66). Moreover, it has been demonstrated that Pcl, in its role as part of the Polycomb 548 Repressive Complex 2.1, is an integral epigenetic regulator of sugar-dependent taste and 549 feeding in D. melanogaster (37,64,65). This complex binds the chromatin of sweet-sensing 550 gustatory neurons, repressing transcription involved in sweet taste; this thus modulates 551 appetitive behavior and promotes obesity (37,64,65). Reduced Pcl expression might therefore 552 benefit Selected flies by attenuating feeding. 553 . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted June 14, 2023. ; https://doi.org/10.1101/2023.06.14.544888 doi: bioRxiv preprint

554
The Drosophila muscarinic acetylcholine receptors or mAChRs comprise a family of three G-555 protein coupled receptors (type A, B, and C) -two of which, A, and B, are strongly expressed in 556 the central nervous system (63). Mammals possess two classes of mAChRs, M1-type and M2-557 type, with mAChR-A sharing the closest homology with the excitatory M1-types (60-62). 558 Recent studies have shown both stimulatory and inhibitory roles for muscarinic signaling in 559 different types of Drosophila neurons (84,85). In addition, mAChR-A has shown age-dependent, 560 spatially restricted expression patterns in neurons of the adult Drosophila brain (86) (86). These 561 data, coupled with a general decrease in expression of mammalian muscarinic acetylcholine 562 receptors with age suggest that mAChR-A may play a conserved role in senescence (87-94). In 563 the present study, mAChR-Ai produced a reduction in both lifespan and consumption in HS-fed 564 males (Fig 5). Because our populations exhibited reduced mAChR-A expression in response to 565 selection correlating to their increased lifespan, we anticipated that knockdown of mAChR-A 566 would improve fitness on HS. The reduction in lifespan we observed in the knockdown flies may 567 result from a critical effect on feeding or other cholinergic-regulated behaviors that negatively 568 affect lifespan. It may be too simplistic too expect basic whole-gene manipulation to recapitulate 569 a complex role. Selected flies may have co-adapted protective genes that help to maintain 570 fitness in the face of reduced feeding, a genetic context that is impossible to replicate in a 571 transgenic animal. To complement the knockdown, we administered the S3-C3 population pair 572 atropine, an inhibitor of mAChR (Fig 6). Here, we observed a trend toward marginal 573 improvements in female lifespan, although the C3 population males (with higher expression of 574 mAChR-A) exhibited a significant reduction in lifespan, similar to the knockdown flies. Given that 575 there is a complex interaction between feeding and nutritional geometry that may cause variable 576 intake of the drug, water, and nutrients, it is difficult to interpret these paradoxical results to 577 determine exactly how atropine impacts lifespan. In addition, mAChRs are broadly expressed at 578 different types of synapses and may regulate developmental and behavioral processes that 579 these data are unable to capture (60,85,88,(90)(91)(92)(95)(96)(97). Furthermore, atropine targets all 580 mAChRs and thus may have affected the type-B and type-C receptors in a way the knockdown 581 did not. 582 583 Curiously, the mAChR-Ai knockdown exhibited a sex-specific phenotype in that only males 584 showed a reduction in lifespan and feeding. This phenotype is consistent with mAChR-A having 585 higher expression in Drosophila males in our study (Fig 4C). The creators of FlyAtlas2, Leader 586 et al., observed higher expression of mAChR-A and mAChR-B in adult male, compared to 587 female, brains (63). Xia et al. further demonstrated that the type-C mAChR is more highly 588 expressed in male heads than in female heads (90). Drosophila males may therefore be more 589 sensitive to the interaction between mAChR signaling and high sugar due to their higher 590 endogenous expression of these receptors. It has also been reported that mammalian mAChRs 591 display sex-specific expression patterns, and drugs targeting the receptors differ in their effects 592 by sex (91-93). Taken together, these data are consistent with other models where the effects 593 of genotype on aging are sex-and brain-dependent. These results were also consistent with a 594 broader trend of sex-specific quantitative traits present in the Selected and Control populations. 595 . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted June 14, 2023. ;https://doi.org/10.1101https://doi.org/10. /2023 596 Sex-specific effects of adaptation 597 The Selected populations exhibited sex-specific alterations at both the expression (Fig 4) and 598 phenotypic (Fig 2) hierarchical levels. Males showed a greater increase in longevity than 599 females after adaptation, compared to both Gen 0 and paired Controls. This ran counter to our 600 expectations, given the dietary nature of the selective pressure, as mated female Drosophila 601 have previously been shown to consume more than males (58,98). Females died more quickly 602 on the HS diet, so that stronger selective pressure was applied to females, compared with 603 males (Fig 2). Therefore, one might assume that the HS diet would have the most detrimental 604 effects on females and elicit a stronger adaptive response. By contrast, our findings, which were 605 consistent across genotypes, suggest males exhibit a greater response to HS than do females 606 (Figs 2,5,6). This is consistent with previous studies showing that male flies fare worse during 607 overnutrition (39,40,99,100). 608 609 Sex differences in the response to selection may also be a result of our unisex aging approach; 610 cohousing flies reduces their lifespan (101). Both males and females suffer fitness costs during 611 the courtship process (101)(102)(103)(104)(105). Courtship behavior and the production of ejaculate are 612 energetically costly for males. Females, likewise, suffer significant energy costs during courtship 613 and during oocyte and yolk production (102-105). Thus, unisex housing post-mating may have 614 imposed a unique type of sex-specific selection pressure. In the absence of females, males will 615 not engage in the cost of courtship. Females, by contrast, will continue to produce eggs even in 616 the absence of males (though this declines with time). Therefore, we expected the influence of 617 mating on lifespan to be relatively low in our study. In addition to the demands of courtship, 618 mating, and egg/sperm production, males and females experience interlocus sexual conflict 619 (101)(102)(103)105). Here, male-or female-biased selection may result in loci that confer fitness 620 advantages for one sex at a cost to the other sex (102,103,(105)(106)(107)(108). For example, male 621 Drosophila courtship behavior involves persistent "chasing" and aggressive attempts at 622 mounting that are detrimental to female survival (101,104,105). Such behaviors in males are 623 energetically costly to maintain, and the alleviation of such costs, as well as other types of 624 conflict, may have contributed to the sex biases we observed in survival. The underlying 625 mechanisms behind these sex differences will require future studies to elucidate them. 626 627 Conclusions 628 In summary, our data indicate a robust phenotypic and genotypic response to selection for long-629 lived flies in two dietary environments using Drosophila. All replicate Selected populations 630 exhibited increased lifespan on HS accompanied by genomic changes at many loci. 631 Interestingly, we observed parallel adaptation at the pathway level across replicates with limited 632 overlap at the gene or nucleotide levels. We identified a number of neuronal genes with putative 633 roles in the physiology of aging on HS. Once fully characterized, mAChR-A and associated 634 genes may provide important insight into the physiological mechanisms by which organisms 635 maintain fitness over the lifespan in the face of overnutrition. 636 637 638 Acknowledgements 639 . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted June 14, 2023. ;https://doi.org/10.1101https://doi.org/10. /2023

655
. CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made  Selected populations in warm colors. The significance corresponding to number of symbols is as follows: *P < 0.05, 671 **P < 0.01, ***P < 0.001, ****P < 0.0001 by log-rank Mantel-Cox. # represents the significance for generation 15 vs 672 generation 0. $ represents the significance for the Selected population vs Control population at generation 15.

674
. CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made  . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made

729
The top six changes ranked by lowest P value using a Fisher's exact test all differed with P < 10 -5 between S3 and 730 C3. g0 denotes differences that reached the same degree of significance from generation 0. association meta-analysis of human longevity identifies a novel locus conferring survival . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted June 14, 2023. ; https://doi.org/10.1101/2023.06.14.544888 doi: bioRxiv preprint . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted June 14, 2023. ; https://doi.org/10.1101/2023.06.14.544888 doi: bioRxiv preprint . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted June 14, 2023. ;https://doi.org/10.1101https://doi.org/10. /2023